Professional Documents
Culture Documents
ah of Southern
Sinai ................................................................................................. 565
Rudolf de Jong
Classical and Colloquial Arabic Archaisms ..................................... 595
Alan S. Kaye
Do Tey Speak the Same Language? Language Use in Juba
Local Courts .................................................................................... 607
Catherine Miller
Paradigmatic Stability and Final Laryngeals in Nigerian Arabic:
Why History Repeats itself, without Actually Doing so ............. 639
Jonathan Owens
Some Aspects of Diglossia as Refected in the Vocabulary of
Literary and Colloquial Arabic ..................................................... 653
Judith Rosenhouse
contents ix
Everything you Always Wanted to Know about l, yil to say
in Egyptian Arabic ......................................................................... 675
Manfred Woidich
Index .................................................................................................... 701
x contents
PREFACE
From when we began to compile a Festschrif for Kees Versteegh on the
occasion of his sixtieth birthday, the frst problem we had to tackle was:
Who should we invite to contribute? Te broad range of his scholarly
interests, his expertise in diferent felds, his academic contribution
to them, as well as his network of global contacts made it clear that
the circle of his colleagues and friends is too large to invite all to con-
tribute. So we decided to confne this volume to Kees Versteeghs core
contribution: Arabic linguistics. Tis decision will surely be regretted
by his colleagues working in related felds of scholarship in which Kees
Versteegh is engaged as well, but we had to make a choice. Moreover,
our time schedule forced us to only select the frst set of early contribu-
tions brought in. In the end, what does it matter? Someone will have
to prepare another for his sixty-ffh!
Even with this limitation, the present Festschrif has become volumi-
nous, since the eagerness of scholars in the feld of Arabic linguistics
to contribute has been overwhelming. It clearly demonstrates: Kees
Versteegh has been widely considered to be one of the most eminent
scholars in this discipline. He also is appreciated as a tirelessly working
editor and co-editor of renowned book series, collective volumes and
encyclopae dias. Moreover, he has won many friends by his kindness
and dependability.
With the list of Kees Versteeghs publications at hand, one realises the
number, diversity, and depth of his scholarly interests, from Hellenistic
elements in Arabic linguistics and other felds of Islamic culture, to
the history of Arabic grammar as a scholarly discipline from classical
to modern times, grammatical and linguistic phenomena described
by scholars of Arabic grammar, changes in written and spoken Arabic
through the ages, as well as early Qurnic exegesis as source for the
beginnings of Arabic grammar. In addition, he ranslated and commented
on Arabic texts, grammatical treatises and other genres including a novel.
His books Greek Elements in Arabic Linguistic Tinking; Pidginization
and Creolization: the Case of Arabic; Arabic Grammar and Qurnic
Exegesis in Early Islam; and Te Arabic Language are handbooks. Every
single scholar and student of Arabic linguistics can not bypass them.
Several of his publications have been translated into Arabic, and others
will follow.
At Nijmegen University, where Kees Versteegh is working since
1972, frst as lecturer and later as full professor, he organized several
workshops about the History of Arabic Grammar (1984 and 1987)
and the Model of Arabic Grammar in other Languages (1997). With
intense pleasure the participants of these conferences remember these
fruitful academic gatherings occurring in a very informal and convivial
atmosphere. Tese gatherings resulted in two collective volumes he
edited in cooperation with one of his many colleagues. Tanks to his
co-editorship, international publication projects such as the History of
Language Sciences and the Encyclopaedia of Arabic Language and Lin-
guistics, Arabic linguistics acquired a frm place in the feld of linguistics
in general. Generations of Dutch students, translators and interpreters
of Arabic will know his name from the two volumes of the dictionary
Woordenboek Arabisch-Nederlands/Nederlands-Arabisch of which he is
one of the editors.
Tis Festschrif will be a monument for Kees Versteegh in Arabic
studies, as a homage to his scholarly oeuvre. Several contributions
camd from former students who wrote their Ph.D. theses under his
supervision. As a matter of fact, all contributors studied with him. We
divided the collected articles into three chapters refecting the foci of
his scholarly oeuvre: history of Arabic grammar, Arabic linguistics, and
Arabic dialectology. We have put the history of Arabic grammar frst,
since it is Kees Versteeghs true domain. He wrote his Ph.D. in this feld,
dedicated most of his publications to it, and returns to it. He translated
and commented upon al-Zajjjs Kitb al-dh , a theoretical treatise on
Arabic grammar and one of the classical works on this issue. Since the
1980s, his research and publications also have been directed towards
other historical and thematical topics in linguistics, and he certainly
will recognize many of his own ideas in the contributions in all three
sections. His studies on pidginization and creolization in Arabic and his
interest in Arabic dialects express his feeling for the historical develop-
ment of the Arabic language.
Kees Versteegh is a polyglot. When invited to lecture outside the
Netherlands he enjoys lecturing in the language of is audience, in Eng-
lish, French, German and Arabic, as well as Spanish or Czech. From
the start of his academic careerhe frst studied Greek and Latinhe
cherished a fondness for these languages. In his research this became
xii preface
apparent in his search for Hellenistic elements in Arabic linguistics.
Together with a colleague of Classical Studies, he also teaches regular
classes on linguistics. His interest inwell, let us sayexotic languages
is exceptional. His eyes sparkle when he tells how he learned Inuktitut,
the language of the Inuit, from an indigenous scholar in Alaska or how
he mastered Hottentot clicks in South Africa.
Many colleagues enjoyed Kees Versteeghs hospitality in his sextons
house in picturesque Batenburg. Tey will never forget the tasty meals
he prepares himself as the evenings conversation quickly passes from
scholarly issues to personal and private ones, as for example his experi-
ences as director of the Netherlands-Flemish Institute in Cairo or the
birds he suddenly noticed by a Zamalek window. One may be engaged
with Kees in a really basic discussion of Arabic linguistics, only to be
interrupted by his quick sprint outdoors to observe an ornithological
event, aferwards resuming the discussion exactly at the comma or full
stop of his last intervention. Bird watching remains his hobby and on
a walk along the Maas (or any other river), he never forgets his bin-
oculars. Besides linguistics, a guest learns a lot about the birds of the
region, their names in Dutch and other languages, including Arabic.
Guests get free access to his rich private library with its rare books
on Arabic linguistics and even the privilege of borrowing without any
check for solvability.
Te study of Arabic has been Arabic-speaking scholars domain for
many centuries. Kees Versteegh has unearthed their understanding of
the language and translated it for Western scholarship. Western scholars
linguistic study of Arabic, according to current standards, started only
at the beginning of the 19th century. So while the western linguistic
approach is much younger than that of those native scholars, Western
Arabic linguistics are on par with even native intellectual eforts, and
Kees Versteegh has contributed to this state of afairs. Had classic schol-
ars of the Arabic language like Sbawayhi or al-Zajjj the change to
witness the linguistic achievements of their modern Western colleagues
like Kees Versteegh, they would have been impressed. Tey might have
cited the verses of Ibn H amds (Sicily, 447527/10561133):
al adawti t uyrin fs h in al anna afs ah ah ajamu
lahunna ardu inda l-h alli muhammalatu l-wazni l tulamu
turajjiu fh durba l-luh ni fa-tut ribun wa-hya l tufamu
On the tunes of birds that speak true Arabic,
but the most eloquent of them are foreigners
preface xiii
who use metres that are neglected by al-Khall and unknown
in which they quaver notes of various melodies,
delighting us although they are unintelligible.
Tis Festschrif would not have seen the light without several helping
hands. Greetje Heemskerk compiled the bibliography of his publications.
Marjolein van der Heul and Ine Smeets produced the raw version of
the index. Elizabeth Bishop copy-edited the English contributions. Joed
Elich of Brill has been willing to publish the book. Ingrid Heijckers
lead us safely through the whole of the production process and beyond.
Renee Otto supervised the publication process. We are very grateful for
their commitment.
Everhard Ditters and Harald Motzki
xiv preface
BIBLIOGRAPHY KEES VERSTEEGH
1977
Greek Elements in Arabic Linguistic Tinking. (= Studies in Semitic
Languages and Linguistics, 7). Leiden: E. J. Brill. Also published as
PhD thesis University of Nijmegen. [Arabic translation by Mah mud
Kankr, 2000: Ans ir ynniyya f l-fkr al-luaw al-arab. Amman:
Jamiyyat umml al-mat bi at-tawuniyya].
1978
Te Arabic Terminology of Syntactic Position. Arabica 25, 261281.
1979
Die Mission des Kyrillos im Lichte der arabo-byzantinischen Bezie-
hungen. Zeitschrif der deutschen Morgenlndischen Gesellschaf 129,
233262.
Reviews
Ephrem Hunayn Festival Baghdd 47/2/1974. 1974. Baghdad: Mat bt
Majma al-Lua al-Suryniyya, Bibliotheca Orientalis 36, 9697.
Rundgren, Frithiof. 1976. ber den griechischen Einfu auf die arabische
Nationalgrammatik. Uppsala: Acta Universitatis Upsaliensis. N.S. 2(5),
119144. Bibliotheca Orientalis 36, 235236.
1980
Hellenistic Education and the Origin of Arabic Grammar. Progress in
Linguistic Historiography: Papers from the International Conference
on the History of the Language Sciences, Ottawa, 2831 August 1978.
Edited by Konrad Koerner. Amsterdam: J. Benjamins. 333344.
Logique et grammaire au dixime sicle. Histoire, pistmologie, Lan-
gage 2. 3952.
Notice bibliographique. Histoire, pistmologie, Langage 2. 6775.
Te International Project Onomasticon Arabicum. Bibliotheca Orientalis
37, 291294.
Te Origin of the Term qiys in Arabic Grammar. Zeitschrif fr ara-
bische Linguistik 4, 730.
Te Stoic Verbal System. Hermes 108, 338357.
xvi bibliography kees versteegh
Reviews
Ambros, Edith. 1979. Sieben Kapitel des arh Kitb Sbawaih von ar-
Rummn in Edition und bersetzung. Wien: Verlag des Verbandes der
wissenschaflichen Gesellschafen sterreichs. Bibliotheca Orientalis
37, 361362.
Malti Douglas, Fedwa and Genevive Fourcade. 1976. Te Treatment
by Computer of Medieval Arabic Biographical Data: An introduction
and guide to the Onomasticum Arabicum. Paris: Centre national de
la recherche scientifque. Bibliotheca Orientalis 37, 362.
Traini, Renato. 1977. Sources biographiques des Zadites (annes 1221200
h.): Letters alif-ha. Paris: Centre national de la recherche scientifque.
Bibliotheca Orientalis 37, 362363.
1981
A Dissenting Grammarian: Qut rub on declension. Historiographia Lin-
guistica 8, 403429. [See also Te History of Linguistics in the Near
East. Edited by Kees Versteegh, Konrad Koerner and Hans-Josef
Niederehe. 1983. Amsterdam: J. Benjamins. 167193].
De taalsituatie in de Arabische wereld. De taal van de Islam: Opstellen
over Arabische, Turkse en Afghaanse cultuur. Nijmegen: Nederlandse
Vereniging voor de Studie van het Midden-Oosten en de Islam.
1938.
La conception des temps du verbe chez les grammairiens arabes. Ana-
lyses, thorie 3, 4768.
Reviews
Klein-Franke, Felix. 1980. Die klassische Antike in der Tradition des
Islam. Darmstadt: Wissenschafliche Buchgesellschaf. Bibliotheca
Orientalis 38, 734737.
1982
Progress and Change in the History of Arabic Grammar. Linguistics
in the Netherlands 1982. Edited by Saskia Daalder and Marinel Ger-
ritsen. Amsterdam: North-Holland. 3950.
Structural Change and Pidginization in the History of the Arabic
Language. Papers from the 5th International Conference on Histori-
cal Linguistics, Galway, April 610 1981. Edited by Anders Ahlqvist.
Amsterdam: J. Benjamins. 362373.
Vertalingen voor 1
ste
jaarsstudenten Arabisch (with Gert Borg). Nijmegen:
Katholieke Universiteit Nijmegen.
bibliography kees versteegh xvii
Reviews
Dagorn, Ren. 1981. La geste dIsmal daprs lonomastique et la tradi-
tion arabes, Genve: Droz; Paris: Champion. Bibliotheca Orientalis
39, 720726.
Rowson, Everett and Seeger Bonebakker. 1980. A Computerized Listing
of Biographical Data from the Yatmat al-Dahr by al-Talib. Malibu:
Undena Publications. Bibliotheca Orientalis 39, 727729.
1983
A Dissenting Grammarian: Qut rub on declension. Te History of Lin-
guistics in the Near East. Konrad Koerner, Hans-Josef Niederehe, and
Kees Versteegh (Eds.). Amsterdam: J. Benjamins. 167193. [See also
Historiographia Linguistica 1981. 8, 403429].
Arabic Grammar and the Corruption of Speech. Arab Language and
Culture. Edited by Ramzi Baalbaki. Beirut: American University of
Beirut. [= al-Abh th 31]. 139160.
Current Bibliography on the History of Arabic Grammar. Zeitschrif
fr arabische Linguistik 10, 8689.
Current Bibliography on the History of Arabic Grammar. Zeitschrif
fr arabische Linguistik 11, 8486.
History of Eastern Linguistics in the Soviet Union. Historiographia
Linguistica 10, 289307.
Te History of Linguistics in the Near East. (= Studies in the History of
Linguistics, 28). Editor (with Konrad Koerner and Hans-Josef Nie-
derehe). Amsterdam: J. Benjamins.
1984
Arab Grammatical Studies before Sbawayh. Matriaux pour une histoire
des thories linguistiques = Essays toward a history of linguistic theories
= Materialien zu einer Geschichte der sprachwissenschaflichen Teo-
rien. Edited by Sylvain Auroux, Michel Glatigny, Andr Jolly, Anne
Nicolas and Irne Rosier. Lille: Universit de Lille III. 227238.
Current Bibliography on the History of Arabic Grammar. Zeitschrif
fr arabische Linguistik 12, 8689.
Piginigado, kreoligado kaj Esperanto. Hungara Vivo 4, 127129.
Pidginization and Creolization: Te case of Arabic. (= Current Issues in
Linguistic Teory, 33). Amsterdam: J. Benjamins.
xviii bibliography kees versteegh
Reviews
Bakalla, Muhammad. 1983. Arabic linguistics. London: Mansell. Biblio-
theca Orientalis 41, 751754.
Carter, Michael. 1982. Arab Linguistics: An introductory classical text
with translation and notes. Amsterdam: J. Benjamins. Bibliotheca
Orientalis 41, 225230.
Nebes, Norbert. 1982. Funktionsanalyse von kna yafalu. Hildesheim:
Olms. Bibliotheca Orientalis 41, 754757.
1985
Current Bibliography on the History of Arabic Grammar. Zeitschrif
fr arabische Linguistik 14, 7981.
La Grande tymologie dIbn Jinn. La linguistique fantastique. Edited
by Sylvain Auroux, Jean-Claude Chevalier, Nicole Jacques-Chaquin
and Christiane Marchello-Nizia. Paris: Denol. 4450.
Studies in the History of Arabic Grammar. Papers Presented during a
Workshop held at the University of Nijmegen, April 1619, 1984. Editor
(with Hartmut Bobzin). Wiesbaden: O. Harrassowitz. (= Zeitschrif
fr arabische Linguistik 15).
Survey of Journals. Arab Journal of Language Studies / al-Majalla
al-arabiyya li-d-dirst al-luawiyya 2, 189197.
Te Development of Argumentation in Arabic Grammar: Te declen-
sion of the dual and the plural. Studies in the History of Arabic Gram-
mar. Edited by Hartmut Bobzin and Kees Versteegh. Wiesbaden:
Harrassowitz. 1985. 152173. [= Zeitschrif fr arabische Linguistik
15].
Translation
Salih, Tayyib. Seizoen van de trek naar het noorden. Uit het Arabisch
vertaald en van een nawoord voorzien. Amsterdam: Meulenhof.
[Dutch translation of Mawsim al-hijra il -aml].
Reviews
Sezgin, Fuat. 1982, 1984. Geschichte des arabischen Schriftums. VIII.
Lexikographie. IX. Grammatik. Leiden: E. J. Brill. Historiographia
Linguistica 12, 452461.
1986
History of Arabic Grammar. Nijmegen: Instituut voor Talen en Culturen
van het Midden Oosten, Katholieke Universiteit.
bibliography kees versteegh xix
Latinitas, Hellnisms, Arabiyya. Te History of Linguistics in the Clas-
sical Period. Edited by Daniel J. Taylor. (= Historiographia Linguistica
13, 425448). Amsterdam: J. Benjamins, 1987. 251274.].
Te Origin of the Romance Languages and the Arabic Dialects. Islo e
arabismo na pennsula ibrica: Actas do XI. congresso da Unio Euro-
peia de Arabistas e Islamlogos, Evora, Faro, Silves, 29 set.6 out. 1982.
Edited by Adel Y. Sidarus. vora: Universidade de vora. 337352.
Word Order in Uzbekistan Arabic and Universal Grammar. On the
Dignity of Man: Oriental and classical studies in honour of Frithiof
Rundgren. Edited by Tryggve Kronholm and Eva Riad. (= Orientalia
Suecana 3334). Uppsala: Almqvist and Wiksell. 443453.
1987
al-Arab f jibl al-Alp. al-Azmina 7, 2630.
Current Bibliography on the History of Arabic Grammar. Zeitschrif
fr arabische Linguistik 16, 130133.
Die arabische Sprachwissenschaf. Grundri der arabischen Philologie,
II. Edited by Helmut Gtje, Wiesbaden: L. Reichert. 148176.
Het Arabisch: Norm en realiteit (with Arie Schippers). Muiderberg: D.
Coutinho.
Marginality in the Arab Grammatical Tradition. Papers in the History
of Linguistics: Proceedings of the Tird International Conference on
the History of the Language Sciences (ICHoLS III), Princeton, 1923
August 1984. Edited by Hans Aarslef, Louis G. Kelly and Hans-Josef
Niederehe. Amsterdam: J. Benjamins. 8796.
Nahwiyyna wa-luawiyyna wa-mawqif Dozy iz at-turt an-nahw
al-arab. F l-mujamiyya al-arabiyya al-mus ira. Edited by Ahmed
El-Ayed and Ibrahim Ben Mrad. Tunis: Dr al-arb al-Islm.
401413.
Reviews
Sublet, Jacqueline. 1985. Cahiers donomastique Arabe, 19821984. Paris:
Centre national de la recherche scientifque. Bibliotheca Orientalis
44, 130133.
1988
De ontwikkeling van de technische woordenschat in Modern Standaard
Arabisch: Naar aanleiding van een recente publicatie. Sharqiyyt 1,
8085.
xx bibliography kees versteegh
Geschiedschrijving in de klassieke Arabisch-Islamitische samenleving.
Tussen traditie en wetenschap: Geschiedbeoefening in niet-westerse
culturen. Edited by R.B. van de Weijer, P.G.B. Tissen and R. Schn-
berger. Nijmegen: Katholieke Universiteit Nijmegen. 133145.
Pourquoi tudier la tradition grammaticale? Le Maroc et la Hollande:
tudes sur lhistoire, la migration, la linguistique et la smiologie de la
culture. Edited by Abdelmajid Kaddouri, Jilali Sab and Abdelmajid
Zeggaf. Rabat: Universit Mohammed V. 207217.
1989
A Sociological View of the Arab Grammatical Tradition: Grammarians
and their professions. Studia linguistica et orientalia memoriae Haim
Blanc dedicate. Edited by Paul Wexler, Alexander Borg and Sasson
Somekh. Wiesbaden: Harrassowitz. 289302.
Early and late Grammarians in the Arab Tradition: Te morpho-
nology of the hollow verbs. Zeitschrif fr arabische Linguistik 20,
922.
La tradition arabe. I : Le langage, la religion et la raison. Histoire des
ides linguistiques, I : La naissance des mtalangages en Orient et en
Occident. Edited by Sylvain Auroux. Lige: P. Mardaga. 243259.
Te Defnition of Philosophy in a Tenth-century Grammarian. Jerusa-
lem Studies in Arabic and Islam 12, 6692.
Vulgair Latijn en Koine-Grieks: De verhouding tussen standaardtaal
en volkstaal. Lampas 22, 7491.
Reviews
Bergter, Annette. 1989. Das Kapitel inna wa-ah awtuh aus dem Man-
haj as-slik des Grammatikers Ab H ayyn al-arnt (12561344).
Hildesheim: G. Olms. Historiographia Linguistica 16, 180184.
1990
Are Linguists Ridiculous? Notes on a heavenly discussion between
grammarians in the 11th century. History and Historiography of
Linguistics: Proceedings of the Fourth International Conference on the
History of the Language Sciences (ICHoLS IV), Trier, 2428 August
1987, I: Antiquitity17th Century. Edited by Konrad Koerner and
Hans-Josef Niederehe. Amsterdam: J. Benjamins. 147155.
Borrowing and Infuence: Greek grammar as a model. Le langage dans
lantiquit. Edited by Pierre Swiggers and Alphons Wouters. Leuven:
Peeters. 197212.
bibliography kees versteegh xxi
Freedom of the Speaker? Te term ittis and related notions in Arabic
grammar. Studies in the History of Arabic Grammar, II. Edited by
Michael G. Carter and Kees Versteegh. Amsterdam: J. Benjamins,
1990. 281293 [Arabic translation by Bouchaib Barramou: H urriyyat
al-mutakallim? Mus s alah al-ittis wa-l-mafhm al-murtabih a bihi f
n-nahw al-arab. Fikr wa-naqd 3:24 (1999). 99110].
Grammar and Exegesis: Te origins of Kufan grammar and the Tafsr
Muqtil. Der Islam 67, 206242. [Reprint in Te Early Islamic Gram-
matical Tradition. Edited by Ramzi Baalbaki. Aldershot [etc.]: Ash-
gate/Variorum. 2007. 3773].
Over taal en verandering. Nmegen: Katholieke Universiteit. Inaugural
lecture, University of Nmegen.
Studies in the History of Arabic Grammar, II. Proceedings of the 2nd Sym-
posium on the History of Arabic Grammar, Nijmegen, 27 April1 May,
1987. Editor (with Michael G. Carter). Amsterdam: J. Benjamins.
Te Arab Presence in France and Switzerland in the 10th Century.
Arabica 37, 359388.
Te Earliest Commentary on the Qurn: Muqtils Tafsr. Makalah-
makalah yang disampaikan dalam rangka kunjungan menteri Agama
R.I.H. Munawir Sjadzali, M.A. ke Negeri Belanda (31 Oktober
7 November 1988). Edited by Wim Stokhof and Nico Kaptein. Jakarta:
INIS. 213219. [with Indonesian translation: Tafsir Quran paling
awal: Tafsir Muqatil].
Reviews
Owens, Jonathan. 1988. Te Foundations of Grammar: An introduction
to medieval Arabic grammatical theory. Amsterdam: J. Benjamins. Die
Welt des Islams 30, 248250.
1991
Arabic Language Teaching and the Status of Standard Arabic. Proceed-
ings of the Symposium on Diferentiation in LSP, Learning and Teach-
ing, Leuven, 710 November 1990. Edited by Serge Verlinde. Leuven:
Instituut voor levende talen. 95102.
Greek Translations of the Qurn in Christian Polemics (9th century
A.D.). Zeitschrif der Deutschen Morgenlndischen Gesellschaf 141,
5268.
Man. 1. In grammar. Te Encyclopaedia of Islam: New edition. Edited
by C. E. Bosworth, E. van Donzel, B. Lewis and Ch. Pellat, assisted
by F. T. Dijkema and S. Nurit. Leiden: E. J. Brill. VI, 346.
xxii bibliography kees versteegh
New approaches towards the history of Arabic. Taqaddum al-lisniyyt
f l-aqt r al-arabiyya [Progrs de la linguistique dans les tats arabes].
Edited by Abdelkader Fassi Fehri. Beirut: Dr al-arb al-Islm.
199216.
Te substratum debate in creole linguistics. Diachronica 8, 5980.
Two Conceptions of Irreality in Arabic Grammar: Ibn Him and Ibn
al-jib on the particle law. De la grammaire de larabe aux gram-
maires des arabes. Edited by Pierre Larcher. Damas: Institut Franais
de Damas. 7792 [= Bulletin dtudes Orientales, 43].
Reviews
Holes, Clive. 1990. Gulf Arabic. London: Routledge. Bibliotheca Orien-
talis 48, 293296.
Wouters, Alfons. 1988. Te Chester Beatty Codex Ac. 1499: A Graeco-
Latin lexicon on the Pauline Epistles and a Greek grammar. Leuven:
Peeters. Bibliotheca Orientalis 48, 523527.
1992
De val van de Barmakiden volgens T abar. Elf wijzen van interpreteren:
Essays over het lezen van teksten uit het islamitisch cultuurgebied.
Edited by Ed de Moor. Nijmegen: Instituut voor Talen en Culturen
van het Midden-Oosten. 117126.
Grammar and Rhetoric: Jurjn on the verbs of admiration. Studies in
Semitic Linguistics in Honor of Joshua Blau. Edited by M. Bar-Asher
[et al.]. Jerusalem: Hebrew University of Jerusalem. 113133. [=
Jerusalem Studies in Arabic and Islam 15].
Te Debate about Latin and Early Romance. Diachronica 9, 259285.
Te History of Linguistics in the Low Countries. (= Studies in the His-
tory of the Language Sciences 64). Editor (with Jan Noordegraaf and
Konrad Koerner). Amsterdam: J. Benjamins. [Published in 1988 as
Historiographia Linguistica 15, 1+2].
Reviews
Malina, Renate. 1987. Zum schriflichen Gebrauch des kairenischen
Dialekts anhand ausgewhlter Texte von Sadaddn Wahba. Berlin:
Schwarz. Bibliotheca Orientalis 49, 526528.
Mitchell, T. F. 1990. Pronouncing Arabic, I. Oxford: Clarendon. Biblio-
theca Orientalis 49, 523526.
Sluiter, Ineke. 1990. Ancient Grammar in Context, Amsterdam: VU
University Press. Historiographia Linguistica 19, 401408.
bibliography kees versteegh xxiii
1993
Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden: E. J.
Brill.
Esperanto as a First Language: Language acquisition with a restricted
input. Linguistics 31, 539555.
Leveling in the Sudan: From Arabic creole to Arabic dialects. Interna-
tional Journal of the Sociology of Language 99, 6579.
Tree is a Crowd: lawyers and linguists on Qurn 4/11. Wolfdietrich
Fischer zum 65. Geburtstag. Edited by Hartmut Bobzin and Otto Jas-
trow. Wiesbaden: Harrassowitz. 302315 [= Zeitschrif fr arabische
Linguistik 25].
Reviews
Jastrow, Otto. 1990. Der arabische Dialekt der Juden von Aqra und Arbl.
Wiesbaden: O. Harrassowitz. Bibliotheca Orientalis 50, 858862.
Johnstone, Barbara. 1991. Repetition in Arabic Discourse: Paradigms,
syntagms, and the ecology of Language. Amsterdam: J. Benjamins.
Bibliotheca Orientalis 50, 855858.
1994
Arabic Outside the Arab World. Editor. New Delhi: Bahri Publications.
[= Indian Journal of Applied Linguistics, 20].
Te Arabic Linguistic Tradition in a Comparative Perspective. Arab
Journal of the Humanities 12 (no. 46), 282297.
Te Notion of Underlying Levels in the Arabic Grammatical Tradi-
tion. Historiographia Linguistica 21, 271296. [Arabic translation by
Bouchab Barramou: Mafm al-mustawayt at-tahtiyya f t-turt
an-nahw al-arab. Recherches Linguistiques 6:2 (2001), 81126].
Reviews
kesson, Joyce. 1990. Ahmad b. Al b. Masd on Arabic morphology =
Marh al-arwh , edited with translation, commentary and introduc-
tion. Part 1: Te strong verb as -s ahh . Leiden: E. J. Brill. Bibliotheca
Orientalis 51, 717719.
Broselow, Ellen, Mushira Eid and John McCarthy (Eds.). 1992. Perspec-
tives on Arabic Linguistics, IV. Amsterdam: J. Benjamins. Bibliotheca
Orientalis 51, 176179.
Comrie, Bernard and Mushira Eid (Eds.). 1991. Perspectives on Arabic
Linguistics, III. Amsterdam: J. Benjamins. Journal of the American
Oriental Society 114, 107109.
xxiv bibliography kees versteegh
Robins, Robert H. 1993. Te Byzantine Grammarians: Teir place
in history. Berlin: Mouton de Gruyter. Bibliotheca Orientalis 51,
719723.
1995
Te Explanation of Linguistic Causes: Az-Zajjjs Teory of grammar,
introduction, translation, and commentary. Amsterdam: J. Benjamins.
(Translation of Ab l-Qsim Abd ar-Rahmn b. Ishq az-Zajjj, Kitb
al-dh f ilal an-nahw).
Arabic Outside the Arab World. (= Indian Journal of Applied Linguistics,
20). (Co-editor). New Delhi: Bahri Publications.
1996
Arab Grammarians: Ab l-Aswad ad-Dual, 3 i-ii; Ab H ayyn, 4
i-5 ii; Ab Ubayd, 5 ii; Ab Ubayda, 6 i; al-Axfa al-As ar, 11 ii;
al-Axfa al-Awsat , 11 ii-12, i; al-As ma, 49 ii-50 i; al-Astarbd , 50
i-51 i; al-Azhar, 59 i-ii; al-Bat alys, 70 ii71 i; al-Frb, 288 ii-290
i; al-Fris, 290 i-ii; al-Farr, 291 i-292 i; al-Frzbd, 300 i-ii;
al-Jarm, 326 ii-327 i; al-Jawhar, 330 iii; al-Jurjn, 380 i-381 ii;
Xalaf al-Ahmar, 384 i-ii; Xall, 385 i386 i; Ibn Ab Ish q, 438 i; Ibn
jurrm, 438 i-ii; Ibn al-Anbr (Ab l-Bakr), 438 ii; Ibn al-Anbr
(Ab l-Barakt), 438 ii-439 ii; Ibn Aql, 439 i-440 i; Ibn Durayd, 441
i-ii; Ibn Fris, 442 ii-443 i; Ibn Jinn, 445 ii-447 ii; Ibn al-H jib, 447
i-ii; Ibn Xlawayh, 447 ii-448 i; Ibn Him, 448 i-ii; Ibn Kaysn, 448
ii-449 i; Ibn Mad, 449 ii-450 ii; Ibn Mlik, 450 ii-451 i; Ibn Manz r,
451 i-ii; Ibn Qutayba, 452 i-ii; Ibn as-Sarrj, 452 ii-453 i; Ibn Sdah,
453 ii-454 i; Ibn as-Sikkt, 454 i-ii; Ibn uqayr, 454 ii-455 i; Ibn Us fr,
455 ii-456 i; Ibn Walld, 456 i-ii; Ibn Ya, 456 i-ii; Ixwn as -S af,
458 i-ii; s ibn Umar at -Taqaf, 461 ii-462 i; al-Kis, al-Mzin, 617
ii- 618 i; al-Mubarrad, 651 ii-657 ii; an-Nahh s, 666 ii- 667 i; al-Ql,
766 i-ii; Qut rub, 767 ii- 768 i; ar-Rummn, 808 i-ii; Sbawayhi, 856
i-859 i; as-Srf, 863 i-864 i; as-Suyt , 895 ii-896 i; at -T alib, 904 i-ii;
Talab, 905 i-ii; Ukbar, 946 i-ii; Yahy ibn Ad, 1028 i-ii; az-Zabd,
1037 i-ii; az-Zajjj, 1038 i-ii; az-Zajjj, 1038 ii-1039 ii; az-Zamaxar,
1039 ii-1040 i; az-Zubayd, 1045 i-ii. Lexicon Grammaticorum. Edited
by Harro Stammerjohann. Tbingen: Niemeyer.
Jahilia. Het woordenboek van De duivelsverzen. Edited by Maarten
Asscher [et al.]. Amsterdam: Contact. 5253.
Linguistic Attitudes and the Origin of Speech in the Arab World.
Understanding Arabic: Essays in contemporary Arabic linguistics in
bibliography kees versteegh xxv
honor of El-Said Badawi. Edited by Alaa Elgibali. Cairo: American
University in Cairo Press. 1531.
Te Linguistic Introduction to Rzs Tafsr. Studies in Near Eastern
Languages and Literatures. Memorial volume of Karel Petrek. Edited
by Petr Zemnek. Prague: Academy of Sciences of the Czech Repub-
lic. 589603.
Western Studies on the History of Arabic Grammar, 19691994. Pro-
ceedings of the Colloquium on Arabic Linguistics, Bucharest August 29
September 2. 1994. Edited by Nadia Anghelescu and Andrei Avram.
Bucharest: University of Bucharest. I, 927.
Reviews
Fernndez, Mauro. 1993. Diglossia: A comprehensive bibliography
19601990, Amsterdam: J. Benjamins. Word 47, 288291.
Peled, Yishai. 1992. Conditional Structures in Classical Arabic. Wies-
baden: O. Harrassowitz. Bibliotheca Orientalis 53, 557560.
Wagner, Daniel A. 1993. Literacy, Culture, and Development: Becoming
literate in Morocco. Cambridge: Cambridge University Press. Interna-
tional Journal of the Sociology of Language 119, 129134.
1997
Bedouin and Townspeople: Ibn Khaldns view on the development of
Arabic. Te Diversity of Development: Essays in honour of Jan Klein-
penning. Edited by Ton van Naerssen, Marcel Rutten and Annelies
Zoomers. Assen: Van Gorcum. 332340.
Landmarks in Linguistic Tought, III: Te Arabic linguistic tradition.
London: Routledge.
S arf. Te Encyclopaedia of Islam: New edition. Edited by C. E. Bosworth,
E. van Donzel, W. P. Heinrichs and G. Lecomte, assisted by P. J. Bear-
man and S. Nurit. Leiden: E. J. Brill. IX, 5354.
al-Shaybn, Ab Amr Ish k b. Mirr. Te Encyclopaedia of Islam: New
edition. Edited by C. E. Bosworth, E. van Donzel, W. P. Heinrichs
and G. Lecomte, assisted by P. J. Bearman and S. Nurit. Leiden:
E. J. Brill. IX, 394395.
Te Arabic Language. Edinburgh: Edinburgh University Press. [2nd
revised edition 2001] [Arabic translation by Muh ammad a-arqw:
Al-lua al-arabiyya. Cairo: Al-Majlis al-al li-t -t aqfa, 2003].
Te Arabic Tradition. Te Emergence of Semantics in Four Linguistic
Traditions: Hebrew, Sanskrit, Greek, Arabic. Edited by Wout van
xxvi bibliography kees versteegh
Bekkum, Jan Houben, Ineke Sluiter and Kees Versteegh. Amsterdam:
J. Benjamins. 225284.
Te Emergence of Semantics in Four Linguistic Traditions: Hebrew, San-
skrit, Greek, Arabic. (with Wout van Bekkum, Jan Houben and Ineke
Sluiter). Amsterdam: J. Benjamins.
Wetenschap in de Islamitische samenleving. In het huis van de Islam:
Geografe, geschiedenis, geloofsleer, cultuur, economie, politiek. Edited
by Henk Driessen. Nijmegen: SUN. 278295.
Reviews
Bohas, Georges (Ed.). 1993. Dveloppements rcents en linguistique
arabe et smitique. Damas: Institut Franais de Damas. Bibliotheca
Orientalis 54, 788793.
Dialectologia Arabica: A collection of articles in honour of the sixtieth
birthday of Professor Heikki Palva. Helsinki: Finnish Oriental Society.
1995. Bibliotheca Orientalis 54, 800802.
Durand, Olivier. 1995. Introduzione ai dialetti arabi. Milan: Centro Studi
Camito-Semitici. Bibliotheca Orientalis 54, 799800.
Gille, Christiane. 1995. Das Kapitel al-Maus l (Das Relativum) aus dem
Manhaj as-slik des Grammatikers Ab H ayn al-arnt (12561344).
Hildesheim: G. Olms. Bibliotheca Orientalis 54, 793795.
Khan, Mohammad-Nauman. 1994. Die exegetischen Teile des Kitb al-
Ayn: Zur ltesten philologischen Koranexegese. Berlin: K. Schwarz.
Bibliotheca Orientalis 54, 806809.
Watson, Janet C. E. 1993. A Syntax of Sann Arabic. Wiesbaden:
O. Harrassowitz. Bibliotheca Orientalis 54, 795799.
1998
Reviews
Endress, Gerhard and Dimitri Gutas. 1994, 1995. A Greek and Arabic
Lexicon: Materials for a dictionary of the mediaeval translations from
Greek into Arabic, fasc. 2 and 3. Leiden: E. J. Brill. Journal of the
American Oriental Society, 118, 108109.
Ingham, Bruce. 1994. Najdi Arabic: Central Arabian. Amsterdam:
J. Benjamins. Bibliotheca Orientalis 55, 294297.
Marais, Jean (Ed.). 1996. Quebecs Aboriginal Languages: History, plan-
ning, development. Clevedon: Multilingual Matters. Applied Linguistics
19, 131135.
bibliography kees versteegh xxvii
1999
Loanwords from Arabic and the Merger of d /d. Compilation and
Creation in adab and lua: Studies in memory of Naphtali Kinberg
(19481997). Edited by Albert Arazi, Joseph Sadan and David
J. Wasserstein. Winona Lake, IN: Eisenbrauns. 273286. [= Israel
Oriental Studies 19].
Te Dutch/ArabicArabic/Dutch Dictionary Project: Aims and
methods (with Jan Hoogland). Revue de la lexicologie 1415,
1326.
Zayd ibn Als Commentary on the Qurn. Arabic Grammar and Lin-
guistics. Edited by Yasir Suleiman. London: Curzon Press. 929.
Reviews
Rizk, Attia. 1992. Ibn a-ajar. M ttafaqa lafz uh wa-xtalafa manhu.
Stuttgart: F. Steiner. Der Islam 76.184186.
2000
Ed de Moor (19362000). Sharqiyyt 12, iii-vi.
Editors Foreword. (With Sylvain Auroux, Konrad Koerner and Hans-
Josef Niederehe). History of the Language Sciences / Geschichte der
Sprachwissenschafen / Histoire des sciences du langage. I. Berlin:
W. de Gruyter. Volume I, HSK 18.1. XXVXXXV.
History of the Language Sciences: An international handbook on the
evolution of the study of language from the beginning to the present =
Geschichte der Sprachwissenschafen: Ein internationales Handbuch zur
Entwicklung der Sprachforschung von den Anfngen bis zur Gegenwart
= Histoire des sciences du langage: Manuel international sur lvolution
de ltude du langage des origines nos jours, I. Editor (with Sylvain
Auroux, Konrad Koerner and Hans-Josef Niederehe). Berlin: W. de
Gruyter. Volume I, HSK 18:1.
Language and logic in the Arabic grammatical tradition. History of the
Language Sciences / Geschichte der Sprachwissenschafen / Histoire des
sciences du langage, I. Edited by Sylvain Auroux, Konrad Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume I, HSK 18.1. 300306.
Tas rf. Te Encyclopaedia of Islam: New edition. Edited by P. J. Bearman,
T. Banquis, C. E. Bosworth, E. van Donzel and W. P. Heinrichs.
Leiden: E. J. Brill. X. 360361.
xxviii bibliography kees versteegh
Reviews
Enk, Gerrit J. van, and Lourens de Vries. Te Korowai of Irian Jaya:
Teir language in its cultural context. Oxford: Oxford University Press.
Word 51, 134137.
Lallot, Jean. Apollonius Dyscole, De la construction (syntaxe). I. Intro-
duction, texte et traduction. II. Notes et index. Paris: J. Vrin. Word
51, 304308.
2001
Arabic in Europe: From language of science to language of minority.
Lingua e stile 36, 335346.
Arabic in Madagascar. Bulletin of the School of Oriental and African
Studies 64, 177187.
History of the Language Sciences / Geschichte der Sprachwissenschafen /
Histoire des sciences du langage, II. Editor (with Sylvain Auroux,
Konrad Koerner and Hans-Josef Niederehe). Berlin W. de Gruyter.
Volume II, HSK 18:2.
Linguistic Contacts between Arabic and other Languages. Arabica 48,
470508.
Naphtali Kinberg. Studies in the Linguistic Structure of Classical Arabic.
Editor (with Leah Kinberg). Leiden: E. J. Brill.
Reviews
Kropp Dakubu, M. E. Korle meets the sea: A sociolinguistic history of
Accra. Oxford: Oxford University Press. Word 52, 163166.
Owens, Jonathan. 1998. Neighborhood and Ancestry: Variation in the
spoken Arabic of Maiduguri, Nigeria, Amsterdam: J. Benjamins. Journal
of African Languages and Linguistics 22, 101105.
2002
Dead or Alive? Te Status of the Standard Language. Bilingualism in
Ancient Society: Language contact and the written word. Edited by
J.N. Adams, Mark Janse and Simon Swain. Oxford: Oxford University
Press. 5274.
al-Zadjdjdj, Ab Ish k Ibrhm b. al-Sar. Te Encyclopaedia of Islam:
New edition. Edited by P. J. Bearman, T. Banquis, C. E. Bosworth,
E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI, 377378.
al-Zadjdjdj, Abu l-K sim Abd al-Rah mn b. Ish k. Te Encyclo-
paedia of Islam: New edition. Edited by P. J. Bearman, T. Banquis,
bibliography kees versteegh xxix
C. E. Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J.
Brill. XI, 378379.
al-Zamakhshar, Abu l-K sim Mah md b. Umar. Te Encyclopaedia
of Islam: New edition. Edited by P. J. Bearman, T. Banquis, C. E.
Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI,
432434.
Translations
Al Jahiz: Traktaat over de lof van de zwarten boven de blanken. De
taal der engelen: 1250 jaar klassiek Arabisch proza. Edited by Arnoud
Vrolijk. Amsterdam: Contact. 135139.
Hunayn ibn Ishaq: De behandeling van staar. De taal der engelen: 1250
jaar klassiek Arabisch proza. Edited by Arnoud Vrolijk. Amsterdam:
Contact. 279282.
Reviews
Abu-Manga, Al-Amin. 1999. Hausa in the Sudan: Process of adaptation
to Arabic. Cologne: R. Kppe. Word 53, 292296.
2003
Arab tradition. International Encyclopedia of Linguistics (2nd ed.). Edited
by William J. Frawley. Oxford: Oxford University Press. 202204.
De Arabische taalkunde in Nederland. Sharqiyyt 15, 2942.
Te Arabic component of the Indonesian lexicon. Rintisan kajian
leksikologi dan leksikograf. Edited by Lilie Suratminto and Muna-
war Holil. Jakarta: Fakultas Ilmu Pengetahuan Budaya Universitas
Indonesia. 216229.
Woordenboek Arabisch-Nederlands. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
Woordenboek Nederlands-Arabisch. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
2004
Approaches to Arabic Dialects: A collection of articles presented to Manfred
Woidich on the occasion of his sixtieth birthday. Editor (with Martine
Haak and Rudolf de Jong). Leiden: E. J. Brill.
From Classical Arabic to the Modern Vernaculars. Morphologie: Ein
internationales Handbuch zur Flexion und Wortbildung = Morphology:
an international handbook on infection and word-formation. Edited
xxx bibliography kees versteegh
by Geert Boo, Christian Lehmann, Joachim Mugdan and Stavros
Skopeteas in collaboration with Wolfgang Kesselheim. Berlin: W. de
Gruyter. II, 17401754.
Grammar and Lexicography.. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: MacMillan Refer-
ence USA. I, 279281.
History of the Arabic Language. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: Macmillan Refer-
ence USA. 5862.
Meanings of Speech: Te category of sentential mood in Arabic gram-
mar. Le voyage et la langue: Mlanges en lhonneur dAnouar Louca et
dAndr Roman. Edited by Joseph Dichy and Hassan Hamz. Damas-
cus: Institut Franais du Proche-Orient. 269287.
One Letter does not make a Word: Te doubly weak imperatives in
Arabic. Words, Texts and Concepts Cruising the Mediterranean Sea:
Studies on the sources, contents and infuences of Islamic philosophy
and Arabic philosophy and science dedicated to Gerhard Endress on his
sixty-ffh birthday. Edited by R. Arnzen and J. Tielmann. Leuven:
Peeters. 131141.
Phonological Constraints in Arabic Grammatical Teory: Te Iltiq
as-skinayn. Romano-Arabica 3, 221235.
Pidginization and Creolization Revisited: Te case of Arabic. Approaches
to Arabic Dialects: A collection of articles presented to Manfred Woidich
on the occasion of his sixtieth birthday. Edited by Martine Haak,
Rudolph de Jong and Kees Versteegh. Leiden: E. J. Brill. 343357.
Taalkunde en letterkunde in de kleine letteren. Gij letterdames en letter-
heren: Nieuwe mogelijkheden voor taalkundig en letterkundig onderzoek
in Nederland. Edited by Joost R. Swanborn. Amsterdam: Koninklke
Nederlandse Akademie van Wetenschappen. 189203.
Te Future of Creolistics. Linguistics Today: Facing a greater chal-
lenge. Edited by Piet van Sterkenburg. Amsterdam: J. Benjamins.
305319.
Whats Where and Hows What? Interrogatives in Arabic dialects.
Estudios de dialectologa norteafricana y andalus 8, 239251.
2005
Breaking the Rules Without Wanting to: Hypercorrection in Middle
Arabic texts. Investigating Arabic: Current parameters in analysis and
learning. Edited by Alaa Elgibali. Leiden: E. J. Brill. 318.
bibliography kees versteegh xxxi
Reviews
Talmon, Rafael. 2003. Eighth-century Iraqi Grammar: A critical explora-
tion of pre-H allian Arabic Linguistics. Winona Lake, IN: Eisenbrauns.
Jerusalem Studies in Arabic and Islam 30, 528535.
2006
A new treatise about the ilal an-nah w: Ibn al-Warrq on inna
wa-axawtuh. Grammar as a Window onto Arabic Humanism: A
collection of articles in honour of Michael G. Carter. Edited by Lutz
Edzard and Janet Watson. Wiesbaden: O. Harrassowitz. 5165.
Arabic Linguistic Tradition. Encyclopedia of Language and Linguistics:
Second edition. Editor in chief: Keith Brown. Oxford: Elsevier. I,
434440.
D d. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. I, A-Ed. 544545.
Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
General editor (with Mushira Eid, Alaa Elgibali, Manfred Woidich
and Andrzej Zaborski). I, A-Ed.
Introduction. (With Mushira Eid, Alaa Elgibal, Manfred Woidich and
Andrzej Zaborski. Eds.). Encyclopedia of Arabic Language and Lin-
guistics. Leiden: E. J. Brill. I, A-Ed. v-x.
History of Arabic language teaching. Handbook for Arabic Language
Professionals in the 21st Century. Edited by Kassem M. Wahba, Zeinab
A. Taha and Liz England. Mahwah, NJ: Lawrence Erlbaum. 312.
History of the Language Sciences / Geschichte der Sprachwissenschafen /
Histoire des sciences du langage. Editor (with Sylvain Auroux, Konrad
Koerner and Hans-Josef Niederehe). Berlin: W. de Gruyter. Volume
III, HSK 18.3.
Leerboek modern geschreven Arabisch. Grammatica, woorden, idioom
en oefeningen. (with Everhard Ditters and Lieke de Jong). Nijmegen:
Radboud Universiteit. 1, 2, 3.
Te Study of the non-Western Linguistic Traditions. History of the
Language Sciences / Geschichte der Sprachwissenschafen / Histoire
des sciences du langage. Edited by Sylvain Auroux, E. F. K. Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume III, 18.3. 27912802.
xxxii bibliography kees versteegh
Reviews
Crowley, Terry. 2004. Bislama Reference Grammar. Honolulu. University
of Hawaii Press. Bdragen tot de taal-, land- en volkenkunde 162,
572575.
2007
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. II: Eg-Lan.
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. III: Lan-Ser.
H araka. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. II: Eg-Lan. 232236.
Id fa. (With Karin C. Ryding). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 294298.
Idmr. (With Michael C. Carter). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 300302.
Il. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. II: Eg-Lan. 307308.
Illa. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
II: Eg-Lan. 308311.
Inna wa-axawtuh. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 354358.
Interrogative Pronoun. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 387389.
Isnd. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. II: Eg-Lan. 434437.
HISTORY
INSIDE THE SPEAKERS MIND:
SPEAKERS AWARENESS AS ARBITER OF USAGE
IN ARAB GRAMMATICAL THEORY
Ramzi Baalbaki
American University, Beirut
1. Introduction
Te Arab grammatical theory as represented by the later grammar-
ians is generally characterized by its focus on the formal aspects of the
utterance within well-defned, albeit complex and ofen controversial
rules of usage. Although it is very difcult to represent linearly the shif
which took place during the evolution of the theory towards formal
aspects at the expense of meaning since there have indeed been some
attempts to restore to meaning its primary position in linguistic analy-
sis , it may be safely argued that this shif started immediately afer the
frst major grammatical work, namely Sbawayhi s (d. 180/796) Kitb.
It is, of course, true that at the levels of morpho-phonology and mor-
pho-syntax Sbawayhi s formal considerations were adopted almost in
their entirety by subsequent grammarians . But Sbawayhi s method of
probing the relationship, at the syntactico-semantic level, between form
and meaning was continuously eroded by the grammarians attempt
to codify rules, systematize usage, and analyze structure largely on the
basis of formal considerations which govern its constituent elements.
Generally speaking, the continuous shif from Sbawayhi s method of
syntactico-semantic analysis culminates in works from the seventh
century A.H. onward, such as Alfyya commentaries and the extensive
sources (mut awwalt ), where pedantic formulae and rigid rules almost
fully replace the vivid and dynamic nature of Sbawayhi s analysis which
takes into account both the formal and the semantic aspects of citations
and utterances.
1
1
In his Muqaddima (10811084), Ibn Xaldn (d. 808/1406) praises Sbawayhi s
Kitb on the grounds that its author did not confne it to the formal rules related to
irb and that it is replete with proverbs and citations from poetry and speech. Tose
who study the Kitb are therefore likely to enhance their malaka (natural linguis-
4 ramzi baalbaki
One of the most signifcant aspects of Sbawayhi s method of gram-
matical analysis is the pragmatic role he assigns to the speaker (mutakal-
lim ), and by extension to the listener (muxt ab ), as part of the social
interaction which language represents for him. Tis aspect which lies
at the core of Sbawayhi s understanding of language as a form of social
behavior and which embodies his originality, and perhaps the essence
of his value in the history of linguistic ideas , has unfortunately been
severely diminished, if not totally annulled, by later authors. Although
this development is obvious in the works which immediately follow the
Kitbsuch as Mubarrad s (d.285/898) Muqtadab and Ibn as-Sarrj s
(d. 316/929) Us lthe shif of focus in linguistic analysis from social
interaction and context of situation to formal considerations becomes
more dramatic in later stages of the history of grammar . Equally unfor-
tunate is that modern scholarship has only recently paid any meaningful
attention to the role which Sbawayhi assigns to the speaker, the lis-
tener, and the context in which speech takes place. Carter (2004, 5657)
observes that one of the most striking features of Sbawayhi s analysis
is that it concerns itself almost exclusively with language as behavior :
speech is a set of actions, each named according to its intention, e.g.
istifm asking a question, tat niya making something dual, tanbh
drawing attention to something. He further observes (p. 57) that every
utterance takes place in a context of a speaker . . . and listener and that
this approach places great emphasis on the pragmatic roles of speaker,
listener and context and invites the analyst to propose psychological
explanations of linguistic phenomena .
2
Bohas et al. (1990, 38) convin-
cingly argues that, from a typological perspective, grammatical and lin-
guistic systems can be divided into two rough classes: on the one hand,
those which analyze utterances in terms of formal relationships between
their components; on the other hand, those which analyze them in terms
tic ability) , although some of them end up mastering grammar as a s ina (craf), but
not as a malaka. Contrarily, the books of the later authors (kutub al-mutaaxxirn) are
void of poetry and the speech of the Arabs , and contain nothing but grammatical rules
(al-qawnn an-nah wiyya). Readers of such works, according to Ibn Xaldn, can hardly
be expected to enhance their malaka and can only master the craf. See also Zakariyy
(1986, 23 f.).
2
See also Carter (2004, 9598) for further discussion of the speakers role and its
signifcance to Sbawayhi s reasoning. It would be particularly interesting to examine
in more detail the efect of the speakers choice (96) and the speakers intention (97) on
utterances in Sbawayhi s analysis of speech.
speakers awareness as arbiter of usage 5
of operations performed by the speaker in order to achieve a specifc
efect on the allocutee. Based on this distinction, they maintain that
Sbawayhi s approach basically belongs to the latter category, while the
classical grammarians, typically belongs to the former.
It must be stressed that what we shall call the linguistic awareness of
the speaker is, according to Sbawayhi , an essential component of the
competence which is demonstrated in successful interaction with the
listener. By linguistic awareness we refer to what he and a few other
authors perceive as the speakers alertness to the various tools which
language places at his disposal and his ability to use them to decide what
form best expresses the meaning which he intends to convey to the lis-
tener. Within this context, the purpose of the present paper is twofold.
First, we shall try to show that although Sbawayhi s method, which
assigns a central role to the meaning and to the speakers awareness , was
generally not followed by subsequent authors, some of them have indeed
attempted to restore that central role, albeit from diferent perspectives.
In a cursory look at the basic elements of the most important of these
attempts, it may be possible to establish that what they have in common
with Sbawayhi is a genuine concern for meaning and for the role of the
speaker. Te latter part of the paper looks into how Sbawayhi s assump-
tion of the speakers awareness is practically applied in his analysis of
particles whose etymology is essential in determining usage. For this,
we shall examine his analysis of those compound particles which may
be used as single words or split into their constituent elements.
2. Meaning and speakers awareness
Te relationship between meaning and the speakers awareness in
Sbawayhi s analysis cannot be overemphasized. As a general rule,
whenever Sbawayhi highlights the role of the speaker in the utter-
ance or his competence in producing correct speech, the reader should
expect meaning to be at the center of the authors argument. Te Kitb
abounds with examples which support this conclusion, and it is cer-
tainly beyond the scope of this paper to investigate the larger number
of formal, semantic and contextual elements which contribute to the
link between meaning and the speakers awareness in Sbawayhi s analy-
sis. Tis notwithstanding, it is important to establish the link between
meaning and one of the most basic concepts in the Kitb, namely taqdr
6 ramzi baalbaki
(suppletive insertion of elided elements) , mainly because, as a gram-
matical tool, taqdr embraces the formal, semantic and contextual ele-
ments that are at the core of correct speech according to Sbawayhi .
Another compelling reason for establishing this link is that it can help
us understand the sharp contrast between Sbawayhi s approach and
that of the later grammarians , for whereas they have largely adopted the
formal aspects of his taqdr, they have, more ofen than not, ignored the
semantic and contextual dimensions of the process of supplying missing
elements to structure. We shall not, however, make any detailed com-
parison between Sbawayhi and the later grammarians in this respect as
this would require an independent study. Rather, we shall point out two
of the most illuminating principles of the relationship between meaning
and taqdr in Sbawayhi s system of grammatical analysis .
3
Te frst principle is that in the process of taqdr , the proposed con-
struction should not contradict the meaning of the original construc-
tion, i.e. before the suppletion of the elements which are judged to be
elided. In an earlier study of the harmony which Sbawayhi tries to estab-
lish, through taqdr, in several types of constructions (Baalbaki 1979,
714), I discussed his method of breaking up one sentence into two,
both of which share a common feature and thus demonstrate an under-
lying harmony. For example, the two sentences Zaydan darabtuhu and
a-Abdullhi daraba axhu Zaydan are interpreted, at the level of deep
structure, as *darabtu Zaydan darabtuhu and *a-daraba Abdullhi
daraba axhu Zaydan respectively (Kitb I:81, 102). If a nominal sen-
tence is conjoined to a verbal sentence , as in raaytu Zaydan wa-Amran
kallamtuhu, he intervenes to restore the harmony by supplying a verb to
the nominal sentence, hence the proposed construction *raaytu Zay-
dan wa-kallamtu Amran kallamtuhu (Kitb I:88). In defending the res-
toration of a verb to produce a verbal sentence parallel to the frst one,
he argues that the introduction of the verb causes no contradiction in
meaning (l yanqud man; Kitb I:8889). Closely related to this argu-
ment is his assertion that in utterances which express amr (command)
or nahy (prohibition) , the verb may be uniformly elided, as in (idrib)
Zaydan, (l taqrabi) l-asada, and (xalli) t -t arqa (Kitb I:253254).
Sbawayhi s discussion of such constructions reveals that the element of
3
Carter (1991, 127128) notes that although Sbawayhi uses the term taqdr exceed-
ingly sparingly, only 24 times in fact (see Troupeau 1976, 167), he does give plenty of
advice on reconstruction without calling it taqdr.
speakers awareness as arbiter of usage 7
meaning which justifes the elision of the verb is present in the context
of situation since the listener would assume the virtual existence of a
verb of which he is the agent. Consequently, the grammarian s interven-
tion to restore this verb would only elucidate, rather than contradict
meaning, based on his understanding of what the speaker has in mind.
Te second principle which governs the relationship between taqdr
and meaning represents a further stage in Sbawayhi s analysis of con-
structions. While he remains faithful to the rule that taqdr should not
contradict meaning, he explores the niyya
4
(intention) of the speaker to
explain why he may well utter a certain part of the construction although
he has another usage in mind. An example of this is his belief that each
of the two constructions lam tika and l tka has the status (manzila)
of a noun at the level of the speakers niyya, and can thus be interpreted
as lam yakun itynun (Kitb III:2829). Te grammatical implication of
this niyya is only fully revealed when the uttered part is virtually replaced
by what the speaker intended to say. Tus, in Farazdaq s line: mamu
lays mus lih na aratan * wa-l nibin ill bi-baynin urbuh,
nibin is in the genitive although at the level of the actual utterance it
is conjoined to mus lih na, which, being the predicate of laysa, is in the
accusative . According to Sbawayhi , the recurrent use of the preposi-
tion bi- with the predicate of laysa (e.g., lays bi-mus lih na) reveals the
true intention of the speaker, and hence nibin is in reality conjoined
to a genitive noun which does not feature in speech, but is as valid as
an uttered noun. In other words, the preposition, which the speaker
has in mind, is syntactically valid and operational, and it determines
usage as if it were actually uttered (h att kaannahum qad takallam
bih f l-awwal; Kitb III:29). As far as meaning is concerned, Sbawayhi
asserts that the assumption of the preposition bi- in bi-mus lih na does
not alter the meaning
5
(l yuayyir al-man) because the preposition is
indeed frequently used with the predicate of laysa. Similarly, Sbawayhi
intervenes in constructions of the type marartu bihi fa-id lahu s awtun
s awta h imrin/s urxun s urxa t -t akl to supply a verb (i.e. yus awwitu,
4
According to Troupeau (1976, 208), the terms naw and niyya occur 13 and 27
times respectively in the Kitb. Te concept of intention, however, is ofen expressed
by much more frequent termsincluding an (136 times), man (891 times), arda
(1361 times), etc. (Troupeau 1976, 150; 102)or by expressions such as kaannahum
ql, tawahham, ixtra, etc.
5
Te signifcance of preserving the meaning of the construction in this line as well
as the meaning of other constructions within the context of Sbawayhi s analysis of the
taqdr of an afer f is discussed by Baalbaki (2001, 186209, esp. 188).
8 ramzi baalbaki
yas ruxu) before the accusative noun (Kitb I:355f.). Te intention of
the speaker is expressed here by the term tawahhama,
6
and the actually
uttered words lahu s awtun are said to have the same status as yus awwitu.
Syntactically , the outcome of taqdr is the use of the accusative in s awta
and s urxa, whereas at the semantic level , it is implicit that the introduc-
tion of yus awwitu and yas ruxu does not contradict the meaning because
these verbs share the same root with the nouns they govern.
Te study of the relationship between meaning and taqdr in the Kitb
strongly indicates that in his analysis of constructions, Sbawayhi tran-
scends the levels of grammatical correctness and the efect of the oper-
ants on case-endings to examine the speakers thinking and the mental
processes involved in the choices he makes. Tis linguistic awareness on
the part of the speaker becomes a real arbiter of usage and allows him,
for example, to use nibin where niban is expected, or to use s awta
h imrin in spite of the absence of a verb from the utterance. It is this
feature in Sbawayhi s analysis that one so much misses in the work of
the later grammarians . In spite of that, some authors were indeed inter-
ested in the role of the speaker and the efect of his awareness on the
speech he produces. From this perspective we shall briefy examine the
contribution of two leading fgures, Ibn Jinn (d. 392/1002) and Jurjn
(d. 471/1098), and show how their priorities are largely consistent with
those of Sbawayhis . In the light of the speakers awareness and his inter-
nal thinking, we shall then examine the analysis proposed by Sbawayhi
for constructions in which the etymology and word-class of certain par-
ticles are crucial for the speaker to achieve correct speech and for the
listener to comprehend what is meant.
3. Post-Sbawayhi authors
Te frst post-Sbawayhi author whose work refects serious concern for
the speakers awareness is certainly Ibn Jinn . Groomed in the gram-
matical tradition and himself author of several works which are in full
conformity with the general grammatical theory most notably Sirr
s inat al-irb , al-Luma f l-Arabiyya, and at-Tas rf al-mulkIbn
6
For the various senses in which the term tawahhum is used in the Kitb, see Baal-
baki (1982, 234237). In this particular case of Farazdaq s line, tawahhum refers to the
speakers mental restoration of elided parts in the utterance, resulting in their govern-
ment of parts actually uttered.
speakers awareness as arbiter of usage 9
Jinn stands out as a unique scholar whose innovative approach in his
most impressive work, al-Xas is , represents a major step in the direc-
tion of determining the theoretical bases of language usage. Te tradi-
tional phonetic , morphologic and syntactic data becomes in al-Xas is
the subject of study from a methodological and epistemological per-
spective. Within this framework, the speaker takes a central role in
Ibn Jinns interpretation of linguistic activity. His point of departure
for this is a frm belief in the intuitiveness
7
of native speakers of Arabic
as well as in their mental abilities, which are manifest in their produc-
tion of speech and subtle analysis of a host of linguistic phenomena .
A particularly illuminating example is that in which he tries to exam-
ine the linguistic awareness of a Tamm informer of his. When asked
why he would say darabtu axka, and not *darabtu axka, but would
still use axka in daraban axka, the Tamm expresses his astonish-
ment and wittily comments that each of the two expressions has a dif-
ferent perspective (ixtalafat jihat l-kalm; Xas is I:76; cp. 1:250). Ibn
Jinn confdently concludes that the Arab native speakers scrutinize the
syntactic positions of speech elements (adall ay al taammulihim
mawqi al-kalm) and that they knowingly and consciously (an mza
wa-al bas ra) assign to each element the position and case-ending it
merits. Similarly, he describes vowel mutation which results from the
speakers sense of lightness (istixff ) and heaviness (istit ql ) as proof
of keen insight and fneness of perception and observation (li-quwwat
naz arihim wa-lut f istiffhim wa-tas afuh ihim; I:78).
Ibn Jinn s interest in the intuitiveness of Arab native speakers and
in their mental awareness of the various phonetic , morphologic , and
syntactic processes involved in speech is best understood as part of his
attempt to reveal the h ikma which underlies Arabic and which consti-
tutes its intellectual basis. Whether the discussion relates to the reasons
(ilal ) , purposes (ard) and intention (qas d) associated with speech
(I:237, 245), or to the change which forms, etc. undergo due to recur-
rent usage (II:31), or to the temporal precedence of one part of speech
7
Ibn Jinn often expresses the notion of intuitiveness by derivatives of the root t -b-
(e.g., s un al-br subh nahu f an t abaa n-ns al hd ; tahjum bihim t ibuhum al
m yant iqna bihi; a-turhu l yuh sin bi-t abihi . . . hd l-qadr; Xas is II:117; III:273,
275 respectively). Cf. also the terms salqiyya and najr (I:76) for intuitiveness, and the
expressions min lut f al-h iss wa-s afihi wa-nas at jawhar al-fikr wa-naqihi; quwwat
nafsihi wa-lut f h issihi; I:239; III:75). See also the comments of Suleiman (1999, 6465)
on the intuition of native speakers and the rationality of Arabic within the more general
framework of Ibn Jinn s study of tall (causation) .
10 ramzi baalbaki
over others (II:33), or to the onomatopoeic nature of phonemes within
words (II:162, 164), or to the occurrence of two dialects side by side in
speech (III:317), the term h ikma is used as an inalienable trait of the
original wdi (creator) of Arabic and of his logic in deciding on what is
to be used or not and how it should be used. It can thus be argued that
the abstract notion of the wdi fnds its practical dimension in the role
of the speaker (or, more generally, the speech community) since Ibn
Jinn ascribes to him an awareness of the linguistic processes that are
the result of the original h ikma. In this respect, Ibn Jinns ideas are very
much in line with Sbawayhi s not only because both of them place the
speaker at the center of their linguistic analysis , but also due to their
mutual interest in the communication of meaning as the ultimate aim
of successful speech.
One of the most striking results of Ibn Jinn s focus on the speakers
role and of his frm belief that, at all levels of analysis, the linguistic phe-
nomena of Arabic are essentially rational is his discussion of grammati-
cal awmil or operants and their relatedness to meaning. Although he
adopts the traditional division of these operants into two types, lafz (for-
mal; expressed) and manaw (abstract) , his interpretation of the awmil
in the light of the speakers role redresses the imbalance between the
two types in the tradition. For example, Jurjn in his al-awmil al-mia
n-nah wiyya (8586; 312), labels ninety-eight of the awmil as lafz and
only two as manaw .
8
Tis imbalance, of course, predates Jurjn , and
it is interesting to note that Ibn Jinns own master, Ab Al al-Fris
(d. 377/987), is reported to have authored a book entitled al-awmil
al-mia (Sezgin 1984, 107). By insisting that all types of amal (rection )
i.e., raf , nas b, jarr and jazm , which cover the three nominal types
(nominative , accusative and genitive ) and the three verbal types (indica-
tive , subjunctive and jussive )are in reality ( f l-h aqqa) produced by
the speaker (Xas is I:109110), Ibn Jinn efectively reduces the tradi-
tional divide between lafz and manaw operants merely into a didactic
technique that tries to distinguish between amal which is accompa-
nied by an uttered operant and amal which lacks such accompaniment
(I:109). In other words, whether an operant is actually uttered or not,
rection is in all cases the result of the speakers internal thinking and
8
Note also the assertion of some grammarians that no mil may be classified as
manaw unless it cannot possibly be explained as lafz (l yudal il jal al-mil
manawiyyan ill inda taad d ur al-laf as -s lih ; Suyt , Ham I:159).
speakers awareness as arbiter of usage 11
a refection of the meaning he intends. Based on this and on the simi-
larly untraditional view that qiys lafz (formal analogical extension)
is not devoid of meaning, Ibn Jinn confdently formulates the general
principle that one cannot but bestow a manaw dimension on what is
lafz , whereas what is manaw may well do without a lafz dimension
(I:111).
It may be useful at this point, at the risk of disrupting the historical
sequence, to bring into the discussion the sixth-century Z hirite scholar,
Ibn Mad (d. 592/1196). Ironically, Ibn Jinn s attribution of amal to the
speaker is enthusiastically received by Ibn Mad . He quotes Ibn Jinns
statement fa-l-amal . . . innam huwa li-l-mutakallim nafsihi l li-ay
ayrihi and highlights his use of nafsihi as a corroborative to emphasize
al-mutakallim, followed by the assertion that rection is attributable to
nothing other than the speaker (l li-ay ayrihi; Radd, 77). Ibn Mad
quickly recognizes his odd position as a Z hirite embracing a Mutazilite
view, and thus hastens to resolve the situation by explaining that con-
trary to the Mutazilites , the doctrine of ahl-al-h aqq (i.e., the Z hirites )
stipulates that case-endings (here, as wt) are in reality produced by
God (innam hiya min fl Allh tal) but metaphorically attributed
to man. Irrespective, however, of this modifcation and of Ibn Mad s
argument that neither the uttered forms of the operants nor their mean-
ings cause rection, his merciless criticism of the grammarians focus on
amal and taqdr rests in part on their disregard for both the speaker and
the intended meaning. In fact, he ofen refers to the speakers intention
(cf. yanw, 89; yurd, 93) and assesses the relationship between meaning
and taqdr (80, 109). Tese two aspects of his theory obviously form his
best defense against traditional grammar and frmly place him, albeit
from the diferent perspective of his Z hirite doctrine, with the few
authors who challenged its shortcomings.
Afer Ibn Jinn , the most important attempt to reinstate a primary
role to meaning and the speaker who intends it is undeniably that of
Jurjn . Like his predecessor, Jurjn followed in the footsteps of the tra-
ditional grammarians in some of his works. In addition to al-Awmil
al-mia n-nah wiyya mentioned earlier, such works include al-Jumal,
most of which is a didactic summary of awmil (chapters 2 to 4), and
a commentary on Friss dh entitled al-muqtas id f arh al-dh .
However, Jurjns two major works in the feld of stylistics , primarily
Dalil al-ijz but also Asrr al-bala, represent a major shif from
the traditional syntactical analysis of the grammarians . In his Dalil, he
12 ramzi baalbaki
makes a zealous appeal for placing meaning at the center stage of gram-
matical analysis and repeatedly argues that naz m whose narrow sense
corresponds to word order , but which more generally refers to the com-
plex relations among the constituents of a structureis nothing other
than the proper adherence to the discipline of grammar (laysa n-naz m
ill an tada kalmaka l-wad allad yaqtadhi ilm an-nah w, 64). As such,
naz m should aim at what he calls the meanings of grammar (an-naz m
huwa tawaxx man n-nah w, 276, 282, 310, 403404; cf. Asrr, 65).
Tis means that syntactical rules , which govern the relationships among
the various parts of any utterance, can express the exact meaning
intended by the speaker since speech formation begins in ones mind ( f
n-nafs) and only then are words arranged to formulate the outcome of
this mental process according to a set of syntactical relationships (43f.).
Based on the conviction that the arrangement of meanings (at-tartb f
l-man) is prior to the arrangement of words, and that form is subsid-
iary to meaning (4445), it follows that any change in syntax is necessar-
ily accompanied by a change in meaning (86f.) Te speakers awareness
of the intricacies of syntactic relations hence acts as a virtual arbiter in
his choice of the naz m , which best expresses the intended meaning.
Jurjn s theory of naz m is to a large extent a reaction against tradi-
tional grammar in which formal aspects acquired greater prominence at
the expense of meaning. Even Sbawayhi , it has been suggested (Baalbaki
1983, 12f.), may have been the target of some of Jurjns critical com-
ments in which he accuses the grammarians of giving too little attention
to meaning. Tis notwithstanding, Jurjn is surely much closer in spirit
to Sbawayhi than to the later grammarians. Both authors strive to inves-
tigate the internal thinking of the speaker and examine its infuence on
actual utterances. On a wider scale, Sbawayhi , Ibn Jinn and Jurjn, the
three most original authors in the related felds of nah w (grammar), phi-
lology (ilm al-lua), and stylistics (bala) respectively, share the view
that meaning should be the main focus of linguistic analysis . A variety
of conceptssuch as the speakers intuitiveness, competence, intention
and awareness of the tools at his disposalfeature in the works of the
three authors as part of their study of meaning and the mental processes
to which it is related. Unfortunately, however, the three have one more
thing in common, for although their focus on meaning and the speakers
awareness represents the most signifcant and original aspect of their
contribution, that focus gave way in later writings to an ever-growing
shif towards formal considerations and pedantic formulae which rel-
speakers awareness as arbiter of usage 13
egate meaning and those concepts that are related to it to a much lower
position in their scale of linguistic analysis.
4. Te importance of the particles
Te importance which Sbawayhi attaches to the speakers competence
and his alertness to the means of successful communication of mean-
ing feature distinctly in his discussion of some particles .
9
More specif-
cally, Sbawayhi is interested in proving that the speaker can express the
intended meaning only if he is aware of certain peculiarities of these
particles. Te rest of this paper shall deal with etymology of particles
as one such peculiarity, the awareness of which by the speaker, accord-
ing to Sbawayhi , is crucial for correct speech. It should be noted here
that the later grammarians largely adopt Sbawayhi s analysis of particles
and their usage, but this does not at all mean that they have preserved
his method of syntactical analysis or embraced the psychological and
contextual explanations which he emphasizes so much. Rather, they
seem to dwell on the formal side of his analysis, or simply repeat his
views without incorporating them into their own methods of analysis
or developing them in any meaningful way.
Initially, it is essential to introduce the concept of h ikya , which
Sbawayhi applies to those particles which he discusses from the etymo-
logical perspective. Te original sense of the root h -k-y, to report; to
imitate, is preserved in Sbawayhi s use of h ikya for direct speech. For
example, in the construction qla Zaydun Amran xayru n-nsi (III:142),
Zayd s words are reported verbatim and therefore qla does not govern
Amrun, which remains in the nominative as in the original utterance.
Derived from this sense is the use of the term to refer to elements that
are intentionally not integrated into the syntactical build of the construc-
tion. Commenting on the sentence marartu bi-Zaydin, for example, one
may respond by saying man Zaydin, in the genitive (II:413). Tis strat-
egy in dialogue can result in the creation of forms that are only used in
9
Te term particle is used here in a general sense which includes not only what the
traditional grammarians classify as particles proper (h urf ), but also what they consider
to be verbs or nouns . We shall therefore refer, for example, to amm , interrogative m ,
and the halumma as particles although they are traditionally classifed as particle, noun ,
and verb respectively.
14 ramzi baalbaki
h ikya , as in the use of ayyayni/ayyna or manayni/manna in response
to raaytu rajulayni/rijlan (II:407408). Further removed from the
original sense of the term h ikya is its use to describe proper nouns that
are syntactically whole sentences or phrases and are uttered verbatim
regardless of their grammatical position. Tese include attested proper
nouns such as taabbat a arran (II:269; III:326), baraqa nah ruhu (II:269;
III:326), ba qarnh (II:85; III:207, 326), ibn jal (III:207), and forms
such as xayran minka, driban rajulan and min Zaydin (III:328329),
which are hypothetically proposed as proper nouns and serve as a test-
ing device for the full potential of h ikya. Sbawayhi s use of the term
h ikya in connection with proper nouns that are made of more than one
element is most probably what facilitated his generalization of the term
to apply to compound particles , that is, particles which he believes are
etymologically made up of more than one element.
As a preliminary example, we can consider the particle h ayt um
which Sbawayhi describes as h ikya . Tis sense of the term is clearly
distinguished in the Kitb from neighboring concepts, most notably
ism wh id and law . Te latter term, law, refers to elements of speech
which are otiose or redundant. Te negative particle l , for example,
is considered to be law when it does not signify negation , as in liall
yalama ahlu l-kitbi (Q 57:29), which is interpreted as li-an yalama in
the afrmative (IV:222).
10
As for the term ism wh id, Sbawayhi recur-
rently uses it in connection with the place name H adramawta , which he
cites as an example of two distinct nouns that coalesced into one (e.g.,
the four chapter titles in II:267; III:296, 374, 475). Although at times he
refers to particles which exemplify h ikya as ism wh id (II:417418),
h arf wh id (II:418), or kalima wh ida (III:115), Sbawayhi clearly
demarcates h ikya, as in h ayt um , from ism wh id in the sense which
H adramawta embodies. He achieves this by looking into the function of
-m in h ayt um and similar particles (see below) and comparing it with
the corresponding element -mawta in H adramawta. Te diference lies
in the relationship which each of -m and -mawta has with the preceding
element (III:331). Te introduction of -m to h ayt u-, he observes, does
not result in the retention of the fnal damma in h ayt u- since h ayt am,
with a fath a , is also attested. Moreover, and more importantly, the intro-
duction of -m causes a semantic change since to the adverbial sense of
10
Another example of law is m in mahm (interpreted as m- + -m), and afer
conditional in , adverbial id , and pronominal ayy (III:5960).
speakers awareness as arbiter of usage 15
h ayt u- is added to the conditional sense which h ayt um expresses. Sim-
ilarly, -m is introduced to an- in amm (amm <an- + -m) to cause
a grammatical change, namely the annulment of the rection of an-.
Because of these semantic and grammatical changes, which its intro-
duction brings about, -m in h ayt um and amm is called muayyira .
Te diference between -m and -mawta becomes clear semantically and
grammatically . Although Sbawayhi does not explicitly describe the dif-
ference, it is implied that h adra- can only have a fath a and not any other
vowel , and that one cannot speak of semantic or grammatical change
caused by -mawta, at least not in the sense which applies to particles .
Furthermore, the term muayyira draws a sharp distinction between
the nature of -m in h ikya and the concept of law which is essentially
antithetical to change.
Based on the above discussion, particles that are classifed as h ikya
can be described as typically made up of two constituent elements (e.g.,
h ayt um > h ayt u- + -m; innam > inna- + -m),
11
and the element
which is identifed as being introducedeither as a prefx , such as la-
in laalla, or sufx , such as -m in h ayt umis not law . But there is
another dimension to this issue since there are particles, such as excep-
tive ill , which may seem to be examples of h ikya according to the
above criteria, but which Sbawayhi interprets as non-compound par-
ticles . Since Sbawayhi does not use a special term for such particles,
we shall refer to them as non-h ikya or non-compound particles. Te
diferentiation between the two types is crucial for our understanding
of the role which Sbawayhi assigns for the speaker in using certain par-
ticles. Following is an illustrative list of h ikya particles , with the most
essential characteristics which Sbawayhi ascribes to each. Tese will be
analyzed and then compared to non-h ikya particles .
Te h ikya particles
12
comprise the following:
11
Only rarely does h ikya involve more than two constituent elements, as in d lika
which Sbawayhi mentions side by side with other h ikya particles that are made up of
two elements (III:332), and laallam in which -m is added to laalla- (IV:221), itself a
compound particle (la- + -alla; III:332).
12
Sbawayhi also uses the term h ikya in connection with personal and demonstra-
tive pronouns (e.g., anta, hd , huli, d ka, d lika, etc.; III:332; IV:218), but these are
not relevant to our discussion. Also beyond our scope are particles like lan which may
well be the result of merging two elements (III:5) but which speakers do not normally
recognize as compound particles .
16 ramzi baalbaki
1. innam : h ikya (III:331); h arf wh id (II:418; III:57); laysat -m bi-
law (III:57); -m changes inna- by imparting a new meaning to it
(IV:222; cf. II:138); -m changes (i.e., annuls) the rection of inna-
(IV:222).
2. annam : h ikya (III:331).
3. h ayt um : h ikya (III:331); -m does not prevent the fnal vowel of
h ayt u- from being either -u- or -a- (III:331); -m imparts to h ayt u-
the meaning of conditional (III:331; cf. III:59, 518; IV:222); -m is
muayyira (III:331; cf. III:59), is unlike -mawta in H adramawta
(III:331), and is not law (III:331); h arf wh id (II:418; cf. III:57);
has the status of in , imm (III:59) and ayna (IV:221).
4. imm : h ikya (III:331); -m is attached (madmma) to in- and
may be elided (III:331332; cf. I:266; III:141; IV:222); has the status
of amm in amm anta munt aliqan int alaqtu maaka (III:332);
see also h ayt um .
5. amm (in amm anta, as in 4 above); introduction of -m
prevents an- from governing the subjunctive and -m is hence
muayyira (III:331, 332).
6. halumma : h ikya in both H ijz and Tamm dialects (III:332; cf.
I:252; III:529, 534).
7. id m : -m imparts to id- the meaning of conditional (III:56); has
the status of innam and kaannam (III:57); laysat -m bi-law
(III:57); h arf wh id (III:57).
8. lawm and lawl : h ikya (III:333); -m and -l impart a new mean-
ing to law- (III:115; IV:222); -m is muayyira (IV:222223); h arf
wh id (III:115); kalima wh ida (II:180); as excitative (tah dd ) par-
ticles , they precede only verbs (III:115).
9. lamm (which governs the jussive ): -m is muayyira (i.e., it changes
the syntactic properties of lam-; IV:223).
10. kam : h arf wh id (III:116).
11. kad : h ikya (III:332; cf. III:151); ay wh id (III:171).
12. kaayyin : h ikya (III:332; cf. II:171; III:151).
13. kaanna : h ikya (III:332; cf. III:151, 164); ay wh id (II:171).
14. kaannam : h ikya (III:331); h arf wh id (II:418; III:57); laysat -m
bi-law (III:57); -m changes the rection of kaanna- (IV:221; cf.
II:138).
15. laalla : h ikya (III:332).
16. laallam : -m changes the rection of laalla- (IV:221; cf. II:138).
speakers awareness as arbiter of usage 17
17. rubbam : kalima wh ida , like qallam and other similar particles
(abhuhum; III:115); precedes verbs , unlike rubba- (III:116).
18. h abbad : kalima wh ida (II:180).
19. al and am (as interrogative particles ): h ikya (III:332).
20. all and hall : h arf wh id (III:5, 115); -l imparts a new meaning to
hal- (IV:222).
21. ill (as a conditional particle ): h ikya (III:332).
22. imm (as a conditional particle): h ikya (III:332); see also
h ayt um .
23. md and badam : see below.
Sbawayhi spares no efort to demonstrate the coherency of the above
group through cross-references in various and ofen distant parts of the
Kitb and by using a largely unifed terminology to describe the prop-
erties of the particles within the group. Furthermore, this coherency is
supported by the characteristics, which its members share particularly
at the level of meaning (nos. 1, 3, 7, 8, 20), syntax (8, 9, 17) and rec-
tion (1, 5, 14, 16). Tis notwithstanding, the particles within this group
considerably vary in their clarity, from the perspective of the speaker,
as to whether they are compound in nature or not. For example, one
can safely assume that id m , kad and laallam are much more easily
recognizable by the speaker as compound particles than, say, halumma
or amm which are far removed from their supposed origin and which
can even be the object of disagreement among grammarians concern-
ing what that origin really is (cf. the case of halumma in Suyt , Ham
II:106107). Te most revealing examples, as far as speakers aware-
ness is concerned, are those in which the particle may be either bro-
ken down into two elements or used as a single entity. To use the late
grammatical term lamh al-as l
13
(lit. recognition of origin) , one can say
that the speaker who splits such particles into two elements is aware
of their etymological origin as compound words, unlike the speaker
13
Te term lamh al-as l is used by the later grammarians mainly to refer to the
speakers recognition of word class as refected in usage. One example is that proper
nouns such as H rit and H asan, contrary to the norm, may be prefxed with the defnite
article al- since they are originally adjectives . Te article is hence called al-allat li-lamh
al-as l (Suyt , Ham I:174175), and the speakers recognition of the adjectival origin
(lamh as -s ifa) of such proper nouns justifes its prefxation to them (Ibn Aql , arh
91; Umn , arh I:8586). Due to the obvious similarity between the recognition of an
original word class and the recognition of a particle s etymology , the term lamh al-as l is
perfectly applicable to m d , man d , etc. when they are split into two units.
18 ramzi baalbaki
who treats them as single-word particles . Sbawayhi specifcally deals
with this issue in a chapter which he designates for md and mand .
Depending on whether d enjoys the independent status of the rela-
tive pronoun allad or is integrated with m into a single unit, which
Sbawayhi typically describes as ism wh id (II:417), the speaker respec-
tively uses either the nominative or the accusative in the noun which
follows md . In Labd s line al tasalni l-mara m d yuh wilu *
a-nah bun fa-yuqd am dallun wa-bt ilu, d is given an independent
status equivalent to allad and m by itself acts as interrogative . Te cor-
roborative noun nah bun thus acquires the nominative because it modi-
fes m, which has the grammatical function of the subject of a nominal
sentence (i.e., mubtada). Conversely, md could have been treated as
a single interrogative particle which grammatically serves as the direct
object of yuh wilu. In this case, corroborative nah ban would be, like
md which it modifes, in the accusative.
It is clear in Sbawayhi s discussion that the speaker has the choice of
using md as a single-word particle or splitting it into its constituent
elements. Te mental process which leads up to the speakers decision is
not restricted to the deconstruction of the particle into its elements or
its retention as one unit. It also has to do with the syntactical ramifca-
tions of the speakers choice since the case-endings of the noun which
follows the particle have to be consistent with that choice.
14
Te speakers
competence is thus demonstrable at two levels, namely his awareness
of the nature of md as a compound particle which admits two pos-
sibilities of usage, and his confrmation of this awareness by observing
the syntactical implications of each possibility. Furthermore, Sbawayhi
takes into consideration the listeners interpretation of the speakers use
of the particle in order to prove that successful communication also
depends on the listeners awareness of how the speakers choice between
two possibilities of usage afects his own response. Te correct response
to m d raayta, for example, would be matun h asanun in the nomi-
native since m itself is mubtada and hence nominative. On the other
hand, md raayta is equivalent to m raayta since the verb governs
14
Note that Sbawayhi does not refer to the fact that, in actual speech, stress may be
an essential factor in diferentiating md from m d . As a single word, md would
normally receive stress on its frst syllable . Conversely, the separation of the two ele-
ments would be indicated by a stronger stress on d than m, perhaps to underline the
likely demonstrative function of d . In short, the diference in meaning between the two
options can be best demonstrated by translating md faalta and m d faalta as What
have you done? and What is this that you have done? respectively.
speakers awareness as arbiter of usage 19
the single-word interrogative particle which precedes it in both cases,
and the answer should thus be in the accusative as in md anzala rab-
bukum ql xayran (Q 16:30).
15
As part of his discussion of md /m d and mand /man d ,
Sbawayhi mentions two other particles which the speaker can use as
a single word or as two separate elements. Tese are kaannam and
h ayt um (II:418; cf. also badam II:139). Sbawayhi ofen examines the
function of the sufxed -m in these and similar particles (see detailed
list above), including annulment of rection (cf. innam ) and shif in
the part of speech which the particle precedes. For instance, rubba and
qalla can only precede nouns as one cannot say *rubba yaqlu and
*qalla yaqlubut the sufxation of -m causes the new particle to
precede verbs only (axlas hum li-l-fl; III:115). In such cases, correct
speech is contingent on the speakers reanalysis
16
of such compound
particles with respect to their constituents as this would ensure proper
distinction between utterances like qallam yaf alu, innam yaf alu,
kaannam yaf alu, etc. and their counterparts with relative m ; i.e.,
qalla m yaf alu, inna m yaf alu, kaanna m yaf alu, etc.
17
For his
part, the listener is obviously expected to diferentiate between the two
types for successful communication to take place.
As we pointed out earlier, there is a group of particles which may
seem to be examples of h ikya but which Sbawayhi interprets as simple
or non-compound (i.e., non-h ikya) particles . In order to highlight the
contrast between these particles and the h ikya particles , Sbawayhi dis-
cusses both types side by side in one of his chapters (III:331332). It
is important, however, to determine at which level this contrast exists.
By classifying, for example, the exceptive particle ill with non-h ikya
particles , Sbawayhi does not want to specifcally deny that it is histori-
cally the result of a merger between in- and -l, in contrast with the
15
Sbawayhi cites the reverse usage, i.e., the accusative after m d and the nomina-
tive after md . But although this is grammatically explicable, he asserts that to use the
nominative after m d and the accusative after md is the most appropriate manner
of response (wajh; aqrab il an taxud bihi; II:418419).
16
Cf. the role of reanalysis in the use of t lam as discussed by Anghelescu (2004,
115116).
17
To illustrate this distinction in the case of innam , for example, we can replace it
by inna m in some Qurnic verses where this is syntactically possible. The resulting
constructions are grammatically sound, but they obviously differ in meaning from the
original constructions. Cf. innam/inna m ttuhu al ilmin ind (Q, 28:78); wa-
man yabxal fa-innam/fa-inna m yabxalu an nafsihi (Q, 47, 38); innam/inna m
tuxzawna m kuntum tamalna (Q, 66:7).
20 ramzi baalbaki
conditional particle ill (see no. 21 in the list of h ikya particles ). But
irrespective of whether or not he actually diferentiates between the two
types of ill at the etymological level, it is obvious that he intends to
establish the contrast between h ikya and non-h ikya particles at the
level of the speakers awareness of their compound versus non-com-
pound nature. In other words, the fact that linguistic analysis may well
prove the compound nature of exceptive ill and similar particles is of
no interest here to Sbawayhi since it has little to do with the speakers
awareness of their etymology . Based on the fact that proper usage and
comprehension of pairs of utterances which have two apparently identi-
cal particles is dependent on the distinction between the simple versus
the compound nature of each particle in a pair (e.g., exceptive versus
conditional ill), we can interpret Sbawayhi s interest in such pairs as
part of his overriding interest in the competence of the speaker in cor-
rectly communicating the intended meaning to his listener. As for why
the particles which he describes as non-h ikya are considerably fewer
in number than the h ikya particles , two reasons may be suggested. Te
frst is Sbawayhi s faith in the speakers ability to break down compound
particles into their componentsalthough some of them are more eas-
ily recognizable as compounds than othersand thus arrive at what
the later authors call lamh al-as l . Te second reason may have to do
with the grammarians own failure, due to their largely synchronic and
non-comparative approach, to ascertain the compound nature of a large
number of particles (e.g., lam , layta , lta , laysa , kayfa , ayna , etc.).
Within the above confnes, the particles which Sbawayhi explicitly
describes as non-h ikya (III:332) are the following:
1. ill : As an exceptive particle , it has the status of dif (oleander) i.e.,
it is a simple or non-h ikya particle and stands in contrast to the
conditional particle ill (see h ikya particles , no. 21).
2. h att : Like ill , it has the status of dif . It should be noted that h att
is the only member of Sbawayhi s group of non-h ikya particles
which is apparently not made up of two elements and which has no
counterpart in the h ikya group. We list it here, nevertheless, for the
sake of completeness.
3. amm : As an inceptive particle in constructions like amm Zay-
dun fa-munt aliqun, it is a non-h ikya particle which has the status
of arw (the like of a thing) and stands in contrast to amm as in
amm anta munt aliqan int alaqtu maaka (see h iyka particles ,
no. 5).
speakers awareness as arbiter of usage 21
4. al : When used inceptively, it is a non-h ikya particle which has the
status of words like qaf (back) and rah (quern), in contrast to inter-
rogative al which is a h ikya particle (no. 19 above).
5. am : Like al , it is a non-h ikya particle when used inceptively, in
contrast to interrogative am (no. 19 above).
Te distinction between non-h ikya particles and their h ikya counter-
parts in context is, of course, dependent on the speakers competence
in using the appropriate syntax , stress intonation, etc. But proper com-
municationas is implied in Sbawayhi s textis also to a large extent a
function of the listeners ability to identify each of two identical particles
one of which is simple and the other is compound. Tis etymological
distinction is to be assumed in the listeners comprehension of sentences
in which these particles appear. Cf., for example:
l tatn fa-tuh addit an ill izdadn fka rabatan (III:32) and wa-ill
tafr l wa-tarh amn akun min al-xsirna (Q 11:74; note that the
choice of these two examples is ours, since Sbawayhi does not provide
contrasting examples for ill );
amm Zaydun fa-munt aliqun (III:332), and amm anta munt aliqan
int alaqtu maaka (III:332); and
al innahu d hibun (IV:235) and al rajula imm Zaydun wa-imm
Amrun (I:289).
Te fact that the distinction between formally identical particles in
the above sentences on the basis of their simple or compound nature
is accompanied by certain syntactic peculiarities is not without paral-
lels in the grammatical corpus. Te mandatory use of f afer simple
or non-h ikya amm e.g., amm anta fa-munt aliqun versus amm
anta munt aliqan int alaqtu maakais strikingly similar to the man-
datory use of lm in constructions like in kna la-s lih an (cf. III:104).
Although Sbawayhi does not use a special term for this lm, it acquired
later the name of al-lm al-friqa because its presence indicates that in
is not a negative particle but the lightened form of inna , known as in
al-muxafafa. In other words, lm contributes to the distinction of two
formally identical particles both of which, unlike the case of amm ,
are not compound. In Sbawayhi s own words, this lm is mandatory
(alzamah l-lm) in order that the lightened form of inna not be con-
fused with in which has the status of m (li-all taltabis bi-in allat hiya
bi-manzilat m allat tanf bih; II:139).
22 ramzi baalbaki
5. Conclusion
Sbawayhi s interest in the speakers awareness of the etymology of par-
ticles and its efect on his usage and on the listeners response is yet
another proof of his method of grammatical analysis which aims at
examining the mental operations which the speaker performs and at
determining the formal and semantic efects of these operations. As the
study of md and mand versus m d and man d in the Kitb shows,
successful communication between the speaker and the listener (who in
turn assumes the role of speaker) is measured by their ability to associ-
ate each of the two possibilities of usage (i.e., simple versus compound
particle ) with the formal and semantic aspects which pertain to it. In
this particular case, what the grammarians refer to as etymology and
is intuitively discerned both by the speaker and the listener represents
the crucial factor upon which successful communication hinges. Te
close association between form, meaning, and speakers awarenessas
exemplifed in Sbawayhi s analysis of h ikya and non-h ikya parti-
cles has certainly been degraded, and at times even totally obliter-
ated by subsequent grammarians . On a wider scale, Sbawayhi s model
of linguistic analysis which, like that of Ibn Jinn s and Jurjn s, largely
rests on exploring the dialectics or interplay, so to speak, between
form and meaning, gave way to an alternative model which is heavily
tipped in favor of formal considerations. Proponents of this model con-
sequently failed to delve, as did Sbawayhi , into the mind of the speaker
in order to pursue the complex processes which result in his choice of
the form that most appropriately expresses the intended meaning and is
expected to have the desired efect and elicit the correct response from
the listener.
6. References
6.1 Primary sources
Ibn Aql , arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al Alfyyat Ibn Mlik.
Ed. by Ramz Munr Baalbak . Beirut: Dr al-Ilm lil-Malyn, 1992.
Ibn Jinn , Xas is = Ab l-Fath Ut mn Ibn Jinn, al-Xas is . Ed. by Muh ammad Al an-
Najjr . Cairo: Dr al-Kutub al-Mis riyya, 195256.
Ibn Mad , Radd = Ab l-Abbs Ah mad b. Abd ar-Rah mn Ibn Mad al-Laxm, ar-
Radd al n-nuh t. Ed. by awq D ayf . 3rd ed. Cairo: Dr al-Marif, 1988.
Ibn as-Sarrj , Us l = Ab Bakr Muh ammad b. Sahl Ibn as-Sarrj, al-Us l f n-nah w.
Ed. by Abd al-H usayn al-Fatl . Beirut: Muassasat ar-Risla, 1985.
speakers awareness as arbiter of usage 23
Ibn Xaldn , Muqaddima = Waliyy ad-Dn Ab Zayd Abd ar-Rahmn b. Muh ammad
Ibn Xaldn, al-Muqaddima. Beirut: Dr al-Kitb al-Lubnn, 1956.
Jurjn , Asrr = Ab Bakr Abd al-Qhir b. Abd ar-Rah mn al-Jurjn , Asrr al-bala.
Ed. By Helmut Ritter . Istanbul: Government Press, 1954.
Jurjn , Awmil = Ab Bakr Abd al-Qhir b. Abd ar-Rah mn al-Jurjn , al-Awmil
al-mia n-nah wiyya f us l ilm al-Arabiyya, bi-arh Xlid al-Azhar. Ed. by al-Badrw
Zahrn . 2nd ed. Cairo: Dr al-Marif, 1988.
Jurjn , Dalil = Ab Bakr Abd al-Qhir b. Abd ar-Rah mn al-Jurjn , Dalil al-ijz.
Ed. by Muh ammad Rad Rid . Repr. from the Cairo edition, Beirut: Dr al-Marifa,
1981.
Jurjn , Jumal = Ab Bakr Abd al-Qhir b. Abd ar-Rah mn al-Jurjn , al-Jumal. Ed. by
Al H aydar . Damascus: Dr al-H ikma, 1972.
Mubarrad , Muqtadab = Ab l-Abbs Muh ammad b. Yazd al-Mubarrad, al-Muqtadab.
Ed. by Muh ammad Abd al-Xliq Ud ayma . Cairo: Dr at-Tah rr, 196568.
Sbawayhi , Kitb = Ab Bir Amr b. Ut mn Sbawayhi: al-Kitb. Ed. by Abd as-Salm
Muh ammad Hrn . Cairo: al-Haya l-Mis riyya l-mma, 1977.
Suyt , Ham = Jall ad-Dn Ab l-Fad l Abd ar-Rah mn b. Ab Bakr as-Suyt , Ham
al-hawmi arh jam al-jawmi f ilm al-Arabiyya. Cairo: Mat baat as-Sada, 1327
A.H.
Umn , arh = Ab l-H asan Al b. Muh ammad al-Umn, arh al-Umn al
Alfyyat Ibn Mlik al-musamm Manhaj as-slik il Alfyyat Ibn Mlik. Ed. by
Muh ammad Muhy d-Dn Abd al-H amd . Cairo: Dr al-Kitb al-Arab, 1955.
6.2 Secondary sources
Anghelescu , Nadia . 2004. La langue arabe dans une perspective typologique. Bucharest:
University of Bucharest.
Baalbaki , Ramzi . 1979. Some Aspects of Harmony and Hierarchy in Sbawayhi s Gram-
matical Analysis. Zeitschrif fr arabische Linguistik 2, 722.
. 1982. Tawahhum: An Ambiguous Concept in Early Arabic Grammar. Bulletin of
the School of Oriental and African Studies 45, pt. 2. 22344.
. 1983. Te Relation between nah w and bala : A Comparative Study of the Meth-
ods of Sbawayhi and Jurjn. Zeitschrif fr arabische Linguistik 11. 723.
. 2001. Bb al-f [ f + Subjunctive ] in Arabic Grammatical Sources. Arabica 48.
186209.
Bohas , Georges , J. P. Guillaume and D. E. Kouloughli . 1990. Te Arabic Linguistic Tradi-
tion. London: Routledge.
Carter , Michael , G. 1991. Elision. Proceedings of the Colloquium on Arabic Grammar ,
Budapest 17 September 1991, Kinga Dvnyi and Tams Ivnyi: eds. Budapest: Et-
vs Lornd University and Csoma de Krs Society. 12133.
. 2004. Sibawayhi . London and New York: I.B. Tauris and Oxford University Press.
Sezgin , Fuat. 1984. Geschichte des arabischen Schriftums. IX. Grammatik. Leiden: E.J.
Brill.
Suleiman , Yasir . 1999. Te Arabic Grammatical Tradition : A Study in tall. Edinburgh:
Edinburgh University Press.
Troupeau , Grard . 1976. Lexique-index du Kitb de Sbawayhi . Paris: Klincksieck.
Zakariyy , Michel . 1986. al-Malaka l-lisniyya f Muqaddimat Ibn Xaldn . Beirut: al-
Muassasa l-jmiiyya li-d-dirst wa-n-nar wa-t-tawz.
PRAGMATICS AND CONTRACTUAL LANGUAGE IN
EARLY ARABIC GRAMMAR AND LEGAL THEORY
Michael Carter
University of Sydney
It is well known that Arabic grammar and Islamic law enjoy a peculiarly
close relationship. Te two sciences are united by a common purpose,
to control linguistic and general behaviour respectively, and they share
a common methodology, namely the inductive derivation of rules from
a linguistic corpus and the deductive application of these universal rules
to particular acts of the Muslim . Where they difer is in their sources.
Grammar relies on the natural, worldly speech of a select range of
human speakers (Bedouin ), law on the inspired texts of the Qurn and
the H adt , which are supernatural in origin and holy in status. All legal
systems are linguistic codes of one sort or another, spoken or written,
but the total dependence of Islamic law on a fnite body of revealed and
prophetic language is unique. Its modern secular analogue is the type of
law which is derived from a written constitution, and here too, the law
has to be discovered by an essentially linguistic process , whereby there
is ofen disagreement over the presumed intentions of those (invariably
dead) who framed the document.
Te development of a method for interpreting the language of the
Qurn and H adt took several centuries, and at risk of oversimplifca-
tion it can be said that the two sciences of grammar and law , aided by
imported Aristotelian logic , leap-frogged each other in an evolutionary
series, where the advances of one made further progress possible in the
other. Tis paper will review the general similarities in the approach
to language in early grammar and law , especially Sbawayhi s intuitive
pragmatism (here in the non-technical sense) and his awareness of the
legal implications of grammatical form . Tere follows a brief account of
some grammatical /legal problems discussed in an intermediate phase
in the 3rd4th/9th10th centuries, and the paper concludes by listing a
number of features of legal methodology which can be linked with ideas
frst noted in Sbawayhi , but which only acquired their fully developed
form afer the maturing of us l al-fqh as a discipline.
26 michael carter
Islamic legal hermeneutics proceeds from the axiom that, regard-
less of its supernatural origin, the language of the holy texts conforms
entirely to the principles of human discourse. Two short quotations, one
from Sbawayhi (d. ca 180/796) and one from a-f (d. 204/820, and
probably born about the same time as Sbawayhi ) will testify that this
notion was well defned at the very birth of the sciences of grammar and
law :
God s servants were spoken to in their own speech and the Qurn came
down in their language and according to what they mean
ibd[u] llhi kullim bi-kalmihim wa-ja l-Qurnu al luatihim wa-
al m yanna (Kitb Der. I:139/Bl. I:167).
In almost identical wording from a-f (the original Arabic could not
be checked: this is from Khadduri 1987, 94, and looks very like a quota-
tion from or paraphrase of the Kitb):
God has addressed his book to the Arabs in their tongue in accordance
with the meanings known to them.
Te context in a-f s case was the dispute over the possibility of for-
eign words in the Qurn , while for Sbawayhi it was a syntactical issue
of indefnite expressions such as salmun alayka; although a defnite
as-salmu would be expected, the indefnite is an old-established Arab
usage that must be accepted, especially when it appears in the Qurn.
A-f is regarded as the frst legal theorist to give proper weight
to the linguistic aspects of the law , since which time both grammar-
ians and lawyers have shown themselves to be remarkably strict and
uncompromising in subordinating the language of God to the linguis-
tic conventions of ordinary Arabic . Te latter, for Sbawayhi , was a dia-
logue between speaker and listener, both being required to conform to
what are basically ethical criteria to speak well (h asan ) and rightly
(mustaqm ). For the lawyers there could be no dialogue with God , only
the contemplation of the written record of what he and his Prophet said,
but over time they evolved a system of interpretation in which they
played the role of silent listeners to a speaker of their own tongue, under
the same conditions as natural speech.
For this they constructed an elaborate hermeneutical mechanism
(us l al-fqh ) which, as documented by Ali (2000), exhibits an impres-
sive congruence in many details with the modern branch of linguistics
pragmatics and contractual language in early arabic 27
known as Pragmatics (henceforth with capital P as a school of thought).
Te thesis of the present article is that a form of Pragmatics can be dis-
cerned in Sbawayhi s analysis of speech (kalm ) long before the the
us l al-fqh were codifed, and, further, that this approach to language
passed through at least two stages before it grew into the explicit type of
Pragmatics seen in Ali s sources.
For the purposes of this article us l al-fqh will be taken in the broad
sense of the principles of legal argumentation in deriving law from the
texts. As for Pragmatics , here is a recent defnition which so closely
refects the us ls position that it might have been written by one of
them, or even by Sbawayhi :
Te study of language from the point of view of the users, especially of the
choices they make, the constraints they encounter in using language in a
social interaction, and the efects their use of language has on the other
participants in the conversation [. . .] including aspects of deixis , impli-
catures , presuppositions , speech acts and discourse structure . [. . .] It has
been characterized as the study of the principles and practice of conversa-
tional performancethis including all aspects of language usage , under-
standing and appropriateness (Crystal 2000, s.v. Pragmatics).
Readers of the Kitb will fnd all these notions very familiar, and some
will be illustrated below. But frst it is necessary to introduce the essen-
tial elements of Pragmatics as laid down by Grice .
Grice (1989) treats speech (writing, signifcantly, does not ft easily
into his model, nor into Sbawayhi s, see below) as a cooperative activity
that is both purposeful and rational, in which the participants under-
stand each other by a logically structured process of conversational
implicature. Speech is a quasi-contractual matter, governed by four
maxims (Grice 1989, 26f):
Te maxim of Quantity :
1. Make your contribution as informative as is required (for the cur-
rent purposes of the exchange).
2. Do not make your contribution more informative than is required.
Te maxim of Quality :
1. Do not say what you believe to be false.
2. Do not say that for which you lack adequate evidence.
Te maxim of Relation :
Be relevant.
28 michael carter
Te maxim of Manner :
1. Avoid obscurity of expression.
2. Avoid ambiguity.
3. Be brief (avoid unnecessary prolixity).
4. Be orderly.
All these maxims can be deliberately ignored or fouted, sometimes for
personal reasons, e.g. a desire to mislead or deceive, sometimes for rhe-
torical purposes, e.g. exaggeration, irony, implying what one is reluctant
to say explicitly, intentional ambiguity etc. Tere can be and ofen is a
diference between the words actually spoken (sentence meaning ) and
what is really meant (utterance meaning ).
In Larcher (1998) an earlier form of Pragmatics is applied to the
category of in , i.e. performative utterances of the type qabiltu hd
n-nikh a I hereby accept this marriage [proposal]. Tis is the Pragmat-
ics of J. Searle and J. L. Austin of How to do things with words fame (bib-
liographical details in Larcher ), out of which Grice s maxims were later
elaborated. However, Grice has been invoked at least twice over the past
few years in articles on Arabic linguistics . Moutaouakil , who quotes the
above maxims in full (1990, 233), deals only with the Pragmatics of the
relatively late scholastic author as-Sakkk (d. 626/1229) and will not
detain us further, except to remark, following Simon (1993, 15) (where
further sources), that it was not until grammar and us l al-fqh had
themselves achieved systematic perfection that an independent science
of rhetoric could emerge, efectively the last of the Islamic sciences to
appear.
Te other is a direct comparison between Grice and Sbawayhi made
by Buburuzan . She interprets the constructions nima l-rajulu abdullhi
What a fne man Abdullh is! and abdullhi nima l-rajulu Abdullh,
what a fne man [he is]! (1993, 424, Kitb Der. I:259f/Bl. I:300f) as
compound expressions where the second part presupposes a question
from the listener. Te frst part, she says, has violated Grice s maxim of
quantity and requires completion in answer to the question who? or
what about him?
1
1
It is relevant here to recall a similar analysis of the syntax of the zaydun ja abhu
structure, Zayd, his father came, which Bravmann (1953, 136) explained as deriv-
ing from a self-answered question, What about Zayd? His father came; Sbawayhi loc.
cit. actually compares the nima construction to this same type, viz. abdullhi d ahaba
axhu Abdullh, his brother has gone. Bravmann s Isolated Natural Subject is not
pragmatics and contractual language in early arabic 29
Sbawayhi makes more or less the same statement about the syntax
of marartu bi-rajulayni muslimin wa-kfrin/muslimun wa-kfrun I
passed by two men, one Muslim , one pagan, where the speaker will
choose between the oblique case ( jarr ) in adjectival agreement or the
independent (raf ) case as if muslimun wa-kfrun were predicates of
elided subjects , because, as Sbawayhi observes (Der. I:182/Bl. I:214,
and see Der. I:215/Bl. I:252 for another, similar case), the speaker tries
to anticipate the question the listener might pose, either, what sort of
men? or, who were these two men?
Like Grice , Sbawayhi is very concerned with the listeners role in
conversation, and there are many linguistic events in which the listener
infuences the speakers choices. In what would be a neither [. . .] nor
[. . .] construction in English , it is qabh . i.e. structurally incorrect, to
say marartu bi-rajulin l frisin I passed by a man who was neither
a knight without completing it with wa-l ujin nor a valiant per-
son or the like, because it is an answer to someone who asked youor
whom you have put in the status of having askedwhether you passed
by a knight or a valiant man (Der. I:313/Bl. I:358, translators italics).
By the same token a listener who answers no to the disjunctive ques-
tion Is it Zayd who is with you or Bir? when one of them is known
to be there, has broken the communicative contract so gravely that his
answer is classifed as muh l [morally] wrong, [semantically] absurd,
i.e. an utterance which is self-contradictory and therefore meaningless
(Der. I:432/Bl. I:483). As if to reinforce the importance of the listener,
Sbawayhi comments that in talking to oneself, e.g. hall af alu why
dont I do this, you are like the listener (Der. I:114/Bl. I:136, kunta
fhi ka-l-muxt abi).
Tere is even a discussion of what looks like body-language when
Sbawayhi describes how, on seeing the fgure of an unknown person,
some sign (ya , the same word as for the verses of the Qurn !) appears
by which you identify him, so you exclaim, Abdullh! Good Lord!
Not only that, the same elliptical exclamation (that is, a predicate with-
out a subject ) can be uttered when the sign by which you identify a
person is his voice or perfume, or simply what you hear said about him
(Der. I:240/Bl. I:279).
intrinsically Gricean , but it accords well with the eminently Pragmatic principle stated
by Sbawayhi (Kitb Der. I:346/Bl. I:394, a notion he acquired from his teacher al-
Xall b. Ah mad ), that every subject must have a predicate because the listener is expect-
ing it.
30 michael carter
It will be apparent that as well as the speaker and listener, the physical
context is also linguistically relevant. Both Grice and Sbawayhi like to
present their data in the form of utterances set in a described situation.
In Grice it is undoubtedly fctitious and ofen involves broken-down
cars or sherry parties, in Sbawayhi it probably refects actual observa-
tion, e.g.:
an example of the suppression of the verb which could be expressed in
normal usage is when you see a man who has just returned from a journey
and you say, the best of returns xayra maqdamin, [. . .] where the depen-
dent (nas b ) form is as if [the speaker] had [syntactically ] constructed it on
the basis of having said may you return, viz. qadimta xayra maqdamin,
and even though he was not heard to say this expression, the arrival of the
other person and the sight of him have the same [linguistic ] status as the
speakers saying qadimta (Der. I:114f./Bl. I:136f.).
Te similarity with Gricean Pragmatics is unmistakable here: not only
does the speaker engage in a cooperative activity with the listener in a
real context, but that context itself can become an active constituent in
the grammatical form of the utterance (cf. Carter 2002, 7, where the
above item is discussed).
In the same way the vocative particle can be dispensed with when
the listener is standing right in front of the speaker (Der. I:104, 274/
Bl. I:125, 316); the object of a blessing does not have to be indicated
if the intended recipient is obvious from the context (Der. I:131/Bl.
I:157)the speaker however remains free to name the recipient for pur-
poses of emphasis). In a rather violent scenario (Der. I:107/Bl. I:128)
the speaker can dispense with the verb and merely shout the persons
name if he sees someone about to be killed or being abused. Tis is com-
mon in warnings, e.g. al-jidr(a) [mind] the wall!, al-asad(a) [dont
go near] the lion!, and the cry of at -t arq(a), at -t rq(a) [get out of] the
way! [Get out of] the way!
2
A striking feature of the Kitb is the sheer quantity of commercial and
contractual talk, much of it admittedly trivial, though it does tell us how
preoccupied the Bedouin in the Mirbad were with the price of sheep,
camels and wheat.
3
But many items are strictly legal in form and con-
2
Tese are always printed with dependent (nas b ) case endings , however, in the cir-
cumstances they are bound to be in pausal form .
3
Te Mirbad was not only a market but also a place where H adt scholars came to
check their vocabulary (EI
2
, art. Mirbad by C. Pellat ), and this may be one of the
reasons why Sbawayhi came to Bas ra to study t r or H adt . As it happens mirbad
pragmatics and contractual language in early arabic 31
tent, involving contracts, debts, sale of goods and property, testimony,
deposits, thef and the activity of fqh itself. Te following list appeared
in Carter 1972, 90f:
inna f alf dirhamin la-madrabun (Der. I:99/Bl. I:119); baya l-malat
l ahda wa-l aqda (Der. I:115/Bl. I:137); tabu d-dra h addun minh
kad wa-h addun minh kad (Der. I:137/Bl. I:165: note that the sale of
undefned property was not legal, at least among the H anafs , see Hamilton
1870, 257), lahu ilmun ilmu l-fuqahi (Der. I:151/Bl. I:181); lahu alayya
alfu dirhamin urfan (Der. I:160/Bl. I:190); bitu -a tan wa-dirha-
man, qmartuhu dirhaman f dirhamin, bituhu dr d iran bi-dirhamin,
etc. (Der. I:165/Bl. I:196); hd dirhamun waznan and others (Der. I:235/
Bl. I:275); inna alfan f darhimika bdun (Der. I:245/Bl. I:285); kam
minkum hidun al fulnin (Der. I:256/Bl. I:297); alayhi aru kalbayni
daynan (Der. I:257/B I:298); al-wadatu ayyuh l-biu (Der. I:284/Bl.
I:326); qadiyyatun wa-l Ab H asanayni = the caliph Al , (Der. I:310/Bl.
I:355); ti l-amra l yaqt au l-lis s a (Der. I:402/Bl. I:435); amm juhda
ray (Der. I:418/Bl. I:470). Te pre-emptive in a llhu to avoid being
bound by an oath may also be mentioned here: it is called istit n both by
Sbawayhi and in later legal terminology (Der. I:399/Bl. I:448, see also
Carter 1972, 90 n. 2).
Additional legal or commercial material which has come to light since
includes (and is not exhaustive: marginal items have been ignored such
as have you barley, wheat or dates?, though this may well refect a mer-
cantile context, Der. I:434f/Bl. I:485f):
Placement of merchandise (Der. I:64/Bl. I:76: even in those days it was
better to buy from the top of the pile rather than the bottom!); expres-
sions of time (important for contracts, Der. I:9093, 176/Bl. I:110114,
208); litigation (Der. I:94/Bl. I:114); price variations (Der. I:122, 170/Bl.
I:147, 200); proft, s adaqa and zakh , rendering accounts (Der. I:165f/Bl.
I:196f); weights and measures (Der. I:141f, 183/Bl. I:216, 292f); sub-
stitution of goods (Der. I:245/Bl. I:285); giving testimony (Der. I:421/
Bl. I:473); default masc. for mixed genders, i.e. slaves (Der. II:180/Bl.
II:174).
One specifcally contractual type of utterance will be discussed here in
some detail, involving the way prices are stated. In a long analysis (over
four chapters, 9295, Der. I:165168/Bl. I:1958) Sbawayhi explores
how deals are struck, beginning signifcantly with the grammar of
kallamtuhu fhu il fyya I spoke to him face to face and byatuhu
is mentioned once in the Kitb (Der. II:265/Bl. II:248) but only as an example of the
mif al pattern .
32 michael carter
yadan bi-yadin I traded with him hand in hand (i.e. for cash), where
the dependent (nas b ) forms are obligatory, since they are not literal, but
simply mean immediately, on the spot, no matter how physically close
the listener might be. In other words the legal (utterance) meaning is
diferent from the overt (sentence) meaning: by saying these words in
this form a legal obligation is created regardless of their literal meaning.
Tis leads to a whole string of commercial expressions, bitu -a tan
wa-dirhaman, qmartuhu dirhaman f dirhamin, bituhu dr d iran
bi-dirhamin, bitu l-burra qafzayni bi-dirhamin, axad tu zakta mlihi
dirhaman li-kulli arbana dirhaman, bayyantu lahu h isbahu bban
bban, in all of which the phrase indicating the unit and price (scil. tan
wa-dirhaman one sheep and one dirham) must be stated in its entirety,
otherwise the meaning will not be valid l yas ih h u l-man, i.e. legally.
By this Sbawayhi means only in the case of a contractual intent, since
the shortened expressions omitting the price are still meaningful but not
in any contractual sense, e.g. bitu tan tan I hereby sell my [col-
lective] sheep, sheep by sheep, bituhu dr d iran I hereby sell him
my house, one cubit, but this would lead the listener to believe that you
were selling your sheep one at a time or that your house was only one
cubit in size, and so on.
Nevertheless, as he observes, the price or the unit are frequently
omitted in ordinary speech, and people will say kna l-burru qafzayni
the wheat was [for sale] at two bushels, omitting the price, or al-burru
bi-sittna the wheat is for sixty [dirhams], omitting the unit of quantity.
Tey do this, Sbawayhi says, because in the frst instance they know
in their hearts ( f s udrihim) that bi-dirhamin is meant and that the
dirham is the standard price unit (allad yusaaru alayhi), so it is as if
they were answering the question, How much you get for a dirham?,
while in the second they and the listener both know what they mean,
as if someone had asked What is the price of a load? and received the
answer Te load is [for sale] at sixty [dirhams] (Der. I:166/Bl. I:196).
Sbawayhi advises us to follow the practice of the Arabs in this, though
al-Xall complicates the picture somewhat by pointing out alternative
formulations.
Sbawayhi s Pragmatic approach is self-evident here: he puts the
conversation in a real-life setting, which assumes all the Gricean max-
ims : he distinguishes between utterance and sentence meaning, and he
accounts for the grammatical features of the expressions in terms of the
extralinguistic situation and the intentions of the participants. If a state-
ment such as yajzu an taqla bitu d-dra d irun bi-dirhamin (Der.
pragmatics and contractual language in early arabic 33
I:166/Bl. I:197) were taken out of context there would be no way of
knowing whether it was from a legal work , scil. it is [legally] permit-
ted
4
to utter [the binding contractual formula] I hereby sell etc., or a
grammatical work, scil. it is allowed [by the rules of grammar ] to say, I
hereby sell etc.
5
Only when Sbawayhi goes on to discuss the possibility
of the independent (raf ) case in d irun here (as advocated by al-Xall )
do we learn that the topic is syntax . Te goal, however, is to determine
the legal consequences of the syntactic options. Why else would he raise
the matter at all?
In the 3rd/9th century a new weapon enters the methodological
armory , the concept, borrowed from Aristotle , that science has to be
logically structured and its enquiries carried out along systematic, that
is, logical lines. For about a century and a half there was a great deal of
experimentation before the sciences reached the point where they could
be defned and classifed, notably in such works as the Mafth al-ulm
of al-Xwrazm , written between 366/976 and 387/997. Two debates
from that period will be mentioned here as evidence of the transitional
stage in the development of grammar and law .
Ibn Walld (d. 332/943) reports a dispute which originated with
Mubarrad (d. 285/898) over the semantic status of commands and pro-
hibitions, whether they were logically complementary, i.e. whether a
command is equivalent to a prohibition from doing the opposite and
a prohibition is equivalent to a positive command to do the opposite.
Ibn Walld s wording (Intis r 42) is: kullu amrin amarta bihi fa-anta f
l-man nhin an xilfhi [. . .] fa-id nahayta amarta bi-xilfhi hence
the positive exclamation h ad araka beware! has the negative meaning,
do not approach. Te probable Stoic origins of this controversy (cf. Ver-
steegh 1977, 181) are not the main point of interest here:
6
what is impor-
tant is to observe how the intellectual environment stimulated such
arguments while the us l al-fqh were still in gestation. Al-Mubarrad
is among the earliest of the grammarians to reveal the direct infuence
4
Te old defnition of jiz by Bergstrer (1935, 32) is still the most informative:
zul ig im moralisch-religisen und zugleich rechtlichen Sinne, und daher rechtsgltig,
rechtswirksam. Replace legal by linguistic to see what the term meant to Sbawayhi :
permissible in a religious-moral and at the same time linguistic sense, hence linguisti-
cally valid, linguistically efective.
5
Sbawayhi s examples in this section are not unambigously performatives, and they
can be, and have been translated elsewhere as literal statements I have sold etc.
6
Te Stoic term pragmata for the things done which are represented by words is the
basis of our linguistic Pragmatics in the sense of doing things with words.
34 michael carter
of Greek ideas before there is any reliable documentation, suggesting
that the ideas circulated informally, perhaps from Christian to Muslim
scholars . Ibn Walld himself is very sceptical, and he fatly asserts that
there is nothing intrinsically positive or negative about commands: only
by inference (istidll ) can we determine whether an imperative verb is
ordering us to do something or prohibiting us from doing the opposite.
Te language and style of this argument are exactly what we encoun-
ter when the us ls take up the theme later in the century. In al-Bqilln
(d. 403/1013, and a strong supporter of the notion of complementarity),
the debate is given a thorough airing (Taqrb II:198207), and by this
time both the grammatical and the logical techniques are well advanced,
and the level of argumentation shows that the Aristotelian dialectic had
been fully absorbed. Tere is a good deal of sophistry from al-Bqilln s
opponents, who claim, amongst other things, that imperative and pro-
hibitive verbs have diferent forms, therefore cannot be complementary
to each other. He matches their sophistry with a reductio ad absurdum,
putting it to them that by their own criteria a positive act would have
to be defned as not refraining from doing the opposite ayru tarki
diddihi and he concludes with the practical argument that the mean-
ings of commands or prohibitions are understood naturally and grasped
immediately from the words themselves, without the need for any kind
of logical inference (while this may not look like Pragmatics , it is really a
question of difering opinions on conversational implicature: the meth-
ods of inference were a very contentious issue among us l s, see Ali
2000, especially ch. 5). Tere is thus a clear line of progression from
Sbawayhi s casual recognition of amr and nahy as grammatical catego-
ries (e.g. Der. I:105/Bl. I:1267), through Ibn Walld s early dialectical
treatment to the fully structured arguments of al-Bqilln and his fel-
low us l s.
In other words, there is now a theory where before there was only data.
Under these circumstances there is a universal tendency for theory to tri-
umph over data, and a very illuminating example is the treatment of the
word kad such and such [an amount]. For Sbawayhi, it was no more
than a dummy numeral (kinya , as he calls it, an allusion to any number,
just as fuln so-and-so is an allusion to any person), and it behaves like
interrogative kam, with a dependent (nas b ) complement, kad wa-kad
dirhaman so and so many dirhams (Der. I:256/Bl. I:297).
By the end of the 4th/10th century a more complicated system makes
its appearance, when ar-Rummn (d. 384/994, arh fol. 161r) proposes
that kad by itself has the value of 1120 (because these numbers are
pragmatics and contractual language in early arabic 35
regarded as single words) while kad wa-kd stands for compound
numbers from 21 upwards.
7
Ar-Rummn was famous for the exagger-
ated logical rigor of his experiments with language, but he did not go as
far as he could have done. Others did go further, and by the 7th/13th
century we fnd the fully elaborated system in Ibn Mut (d. 628/1231,
Fus l 23, further references in 2445), exactly mimicking the syntax of
the numerals :
kad wa-kad darhima = from 3 to 10 dirhams.
kad wa-kad dirhaman = from 11 to 99 dirhams.
kad wa-kad dirhamin = from 100 dirhams upwards.
8
Tis exquisitely artifcial and almost certainly unattested scheme is con-
nected with the H anafs and (which is not much diferent in this con-
text) with the Kfans , and it survived at least until the 19th century in
the work of Ns f al-Yzij (d. 1871, in Nr al-Qir, see Fleischer I:568
for references). What it shows is that as the linguistic and legal sciences
evolved they became increasingly abstract, passing beyond the limits of
actual usage. It does not seem likely that Ab H anfa himself explicitly
correlated the syntax of kad with that of the numerals in this manner,
but it fts the reputation of the H anafs for artifcial and over-systematic
reasoning, and can be seen as a fne specimen of what can happen when
a legal judgement (scil. how much is meant by kad in such and such a
case?) has to depend more on the methodology than on actual speech.
Classical Arabic was no longer a living language at this time, and the
us ls could only consult the rules of the dead language.
In doing so they nevertheless assumed that their one-way commu-
nication with the texts was a natural use of language. Here follow some
illustrations of the inherent similarities between Sbawayhi s concept of
speech and the lawyers approach to consultations with God .
It has already been said above that God s speech had to follow the
formal rules of Arabic grammar . Furthermore, as the us ls were well
aware, in order to make himself understood, God had to obey the con-
ventions of human communication. For this reason the usls , exactly
like Sbawayhi , strove to account for meaning in terms of the motives
of the speaker (man or God) and the real-life context of the utterance.
A good specimen from Sbawayhi of the dependence of meaning on
7
Trough a scribal error the single kad (scil. kad dirhaman) and paired kad (scil.
kad wa-kad dirhaman) are not distinguished in the manuscript.
8
Further references in Carter 2003a, endnote 27 referring to p. 180.
36 michael carter
motive and context is the statement m atn rajulun: it can mean no
man [at all] came to me, not one man came to me[but several], no
[real] man came to me [but a weakling], or no man came to me [but a
woman did] (Der. I:20/Bl. I:27). Tere are no formal linguistic clues
whatsoever as to which meaning the speaker has in mind: the clues lie
in the extra-linguistic context, in this case, what question might the
speaker be answering, and this is why Sbawayhi spends so much time
ofering psychological explanations for the speakers choices. Some cen-
turies later the lawyers had to do the same for God, trying to penetrate
his words to divine his purpose, which they could only do by assuming
that he spoke to his servants in the same way as a rational human being
would under normal circumstances.
Sbawayhi took this for granted. Te muxtabna in this next quota-
tion are the same people who are the muxt abna of everyday conversa-
tion, but this time it is God speaking to them:
When [God ] said your mothers are prohibited to you and so on to the
end of his speech (kalm), those being addressed (al-muxt abna) knew
that this was proscribed for them, and afrmed, but God went on to say
kitba llhi by written decree to add emphasis, tawkdan (Der. I:160/Bl.
I:191, on Q 4/2324).
Striking here is the attempt to explain away the mention of the written
decree: writing is ill-suited to Pragmatic analysis for the simple reason
that the recipient of writing (or in this case the sender!) is usually absent.
Tere can be no conversation, still less any conversational implicature ,
without the presence of both participants. Writing is mentioned only
occasionally in the Kitb, and is completely marginal and secondary.
For the us ls , too, it was not part of normal communication: it was a
special case, a kind of act of faith in the future, like farming, lending or
borrowing, and other actions which require a presumption of continu-
ity, istis h b al-h l , i.e. that the recipient would still be alive when the
letter was delivered (see Ali 2000, 80). We have to admire the us ls
for their commitment to the belief that the spoken words of God and
Muh ammad can be directly experienced through their written record:
the paradox was to a large extent resolved by the device of learning the
texts by heart so that, once implanted in the memory, they ceased to be
a document (mus h af , kitb) and became a virtual oral event (Qurn ,
h adt ), neurologically the same as a remembered discourse.
God as a speaker also has the same privilege as humans of presuming
knowledge in his listeners: in Q 3/180 (Der. I:347/Bl. I:395) he says let
pragmatics and contractual language in early arabic 37
not those who are miserly with the bounty that has come to them reckon
[being miserly] is better for them . . ., and Sbawayhi explains that God
here omits the word al-buxla being miserly which would be required
as the frst direct object of reckon, because the listener, al-muxt ab,
will know from the verb yabxalna they are miserly that miserliness
is meant. Anyone who has consulted a tafsr will be familiar with this
method of flling gaps in the Qurn , and the us l sources are no excep-
tion in the use of this procedure.
A second point of similarity between Sbawayhi and the us ls is the
presumption of sanity. Sbawayhi assumes, without spelling it out, that a
speaker will be mentally capable of formulating an idea and conveying
it successfullythis is implicit in his criterion of mustaqm right, used
for utterances which are fully understood in their intended meaning,
and in the term murd what is intended for the meaning of speech
acts. Te lawyers turned this into an overt legal principle . Tey had to
decide who was allowed to speak, and to eliminate those who were not
legal persons, and therefore had no voice, such as infants and the insane,
and they produced the following conditions of legally valid speech (here
paraphrased from Ali 2000, 42 based on al-mid , d. 631/1233):
1. It must be uttered intentionally.
2. It must be intended for a particular listener.
3. Te listener must be rational and understand it.
Te frst condition presupposes sanity, because only a sane person can
form an intention at all (legal or otherwise), or indeed be a Muslim for
that matter (sanity later formed part of the defnition of a Muslim); the
second criterion excludes soliloquy, and the third anchors speech in a
sane society, as well as giving us a hint as to how the us ls saw them-
selves in the dialogue. To recall the notion of leap-frogging introduced
above, it should be noted that these new legal defnitions of speech found
their way back into grammar , where the speech of the insane, or of those
talking in their sleep, or even of birds imitating humans, were excluded
for the sole reason that their speech could not be intentional (e.g. a-
irbn d. 977/1570, Nr 10, though it appeared before his time).
God himself comes under the same constraints: in order to commu-
nicate with humanity he must speak rationally. His attributes allow for
this: he has an intellect, a will and the power of speech, and his language
is that of the people he is addressing (there is some literature, which can-
not be looked at here, on the requirement that God address his prophets
in the language of their own people otherwise the revelation will be in
38 michael carter
vain: you cannot punish sinners for disobeying laws they are unable to
understand). For the us ls Gods mode of communication (bayn ) is
inevitably a major theme in their treatises: they either begin their works
with a detailed review of the nature of Arabic and of legal semantics ,
e.g. the introductory chapters of Ab l-H usayn al-Bas r (d. 436/1044),
Mutamad, or else the subject is raised afer the epistemological topics
have been covered, e.g. al-Bqilln (d. 403/1013), Taqrb, from I:316.
For Sbawayhi , and later the us ls , lexical meaning is arbitrary. Def-
nition by synonyms only leads to infnite regression (Der. II:339/Bl.
II:312), and meaning is nothing more than intention, hence the verb
arda to want and its derivatives are among the commonest terms in
the Kitb (1,362 times, plus 20 in the passive , according to Troupeau
(1976), s.v., and there are also synonyms). It is clear, too, that Sbawayhi
fully aware of the distinction between utterance meaning and sentence
meaning : he refers more than once to man l-kalm and man l-h adt
the [integral] meaning of the utterance i.e. not simply the sum of its
lexical parts, and man itself is almost exclusively used to denote the
meaning of speech acts, not of words, such as the acts of expressing sur-
prise, asking a question, giving an order etc., e.g. the meaning of swear-
ing an oath man l-qasam, the meaning of calling man n-nid, and
even of grammatical categories , the meaning of the dependent form
man n-nas b , the meaning of tanwn, etc.
A signifcant similarity between Sbawayhi and the lawyers is that they
both defne the meanings of the particles (h urf ) in terms of their dis-
course functions: thus wa- and is used to to bring one thing together
with another and join them without indication of order, and fa- and
[then] is the same except that you leave some scope for one to be afer
the other (Der. II:330/Bl. II:304), cf. Ab l-H usayn , Mutamad I:20,
very concisely, wa- is li-l-jam for joining while fa- is li-l-taqb for
arranging consecutively. Sbawayhi s defnition of naam yes is inter-
esting: instead of the expected agreement or consent he gives us a
rather legalistic defnition: naam indicates promise and belief , ida
wa-tas dq (Der. II:339/Bl. II:312), the former implying some kind of
contractual commitment (yes, I promise do it), the latter indicating
assent to a proposition (yes, I believe what you say), which in our con-
text could mean believing the sellers description of the goods or the
terms of a contract.
Sbawayhi never even asks where meaning originates, but the us ls
were obliged to agree on an answer before they could proceed to the
pragmatics and contractual language in early arabic 39
derivation of law from the texts. Here is a case where the debates of
the intervening century and a half carried the lawyers a long way from
Sbawayhi s agnostic position. Tey had to reconcile the potentially inf-
nite backward extension of meaning with the historical fact that Arabic
is not the oldest human language. Part of the solution, which will not be
discussed here, was to attribute to Adam the bridging role connecting
the supernatural Arabic he spoke in Heaven with the temporal world he
inhabited afer the Fall. He himself did not speak Arabic on earth which
appeared only later, evolving naturally until it reached perfection in the
time of Adams prophetic heir and descendant, Muh ammad .
9
Afer Sbawayhi there was considerable discussion of the origins of
language, which has been investigated for the grammarians by Lou-
cel (196364), with the general conclusion that the choice lay between
divine ordination (tawqf ) or human convention (tawdu ). Te lacuna
which Loucel pointed out, that there was no comparable study of the
origins of language in the legal sciences , remains unflled and will not be
dealt with here. For our purposes it must sufce to note that there was
overall preference for the view that language is in some way convention-
ally imposed (by wad) but the identity of the imposer, wdi, is lef
obscure, perhaps deliberately: it may be God , it may be the frst users of
Arabic , it may be all users of Arabic who agree amongst themselves on
the meaning of a word.
Te real dispute concerned whether meaning could be imposed inde-
pendently of a words being used. At one end of the spectrum (Mutazil ),
a word does have a meaning before it is used, and at the other (Ibn
Taymiyya ), a word cannot have a meaning until or indeed unless it is
used. Tese issues are well described by Ali (2000): what is important
for this paper is that none of these ideas, including the term wad in
this sense, are found in the Kitb. Te concept must have emerged later,
probably under the infuence of the Platonic debate over whether words
had meaning by their nature (physis, cf. Arabic t aba , replaced in Islm
by the creating God ) or by imposition (thesis, the same as the wad of the
grammarians and lawyers).
From the lawyers point of view it was important to detach meaning
from prehistory: in spite of disagreement about the origins of language,
lexical meaning was taken as given, either a priori or as recorded by lexi-
cographical experts (ahl al-lua ), or synchronically by mere usage. It
9
See Carter (2003b).
40 michael carter
could then be treated as purely conventional, and in this way the lawyers,
afer a long debate in which Sbawayhi took no part, eventually returned
to his Pragmatist position. Meaning, regardless of where it comes from,
is what you do with the language, or, put another way, language is only
meaningful for us ls when it has legal efects outside the actual utter-
ance. Tis restriction of meaning to habit and community usage enables
the law to control that communitys behaviour. Te result is the same for
both grammarians and lawyers: speakers are obliged to stay within the
habitual codes, whether linguistic or social.
We might say that Sbawayhi took a lawyers view of language and
lawyers a Sbawayhi an view. Consider his Pragmatist interpretation of
such verbs as ra etc. to see, consider, regard, be of the view that, of
which he says, even a blind man can use ra to see and say, I regarded
Zayd as the good man raaytu zaydan-i s -s lih a (Der. I:13/Bl. I:18).
10
Te equivalent lawyers position is stated by a-aybn (b. 132/749, d.
189/804, and therefore a contemporary of Sbawayhi ): a blind man who
has to feel the goods for sale when making a purchase is in the same
place (mawdi, i.e. legal situation ) as a sighted man ( Jmi 81).
Te primacy of usage and habit is asserted many times by Sbawayhi ,
and there is no better illustration (because one senses a tongue in cheek
here) than his discussion of expressions of praise and blame. Tey are
not unconstrained, he says, you must follow the speech habits of the
Arabs , so you cannot, for example, praise someone for being a tailor
or a seed-merchant, still less praise a person in terms normally used of
God , e.g. al-h amdu li-zaydin praise be to Zayd!. Sbawayhi , or pos-
sibly a commentator, allows himself a pun here by saying that would
be a grave sin, am, playing on tam magnifcation, the name for
this laudatory construction (Der. I:214f/Bl. I:251). And although it is
correct to use such attested idioms as he is as close to me as where my
waist-cloth is tied huwa minn maqida l-izri, you cannot say he is as
close to me as where the horse is tethered *huwa minn marbit a l-farasi
(Der. I:174/Bl. I:206). Ab l-H usayn discusses the interdependence of
meaning and use in a similar way (e.g. Mutamad I:17f, 2228), and it is,
of course the central problem of us l semantics , as Ali (2000) demon-
strates in great detail.
10
Te example is perhaps deliberately perverse, as the natural reading would be I
saw the good man Zayd, but this is in a chapter on verbs of the heart, so ra must have
the complete sentence Zayd [is] the good man as its direct object .
pragmatics and contractual language in early arabic 41
Finally Sbawayhi s recognition of the performative and illocutionary
uses of speech can be compared with the far more elaborate and self-
conscious expressions of the same ideas in the legal works , written afer
the lawyers had had time to absorb Aristotle s categories of sentences (a
post-Sbawayhi an innovation), which include vocatives , requests, com-
mands and entreaties.
Tus when Sbawayhi defnes the vocative noun in y abdallhi as
made dependent (nas b ) through the suppression of the verb which [in
this context] is not expressed (Der. I:262/Bl. I:303, nas bun al idmri
l-fli l-matrki ihruhu), he does leave it open for others to supply a
verb such as und I call out to, though his choice of the verb here
is signifcant, and in fact there are situations where a verb cannot be
restored, e.g. subh na llhi glory be to God , for which there is no verb
available to be suppressed. Tis is enough (especially when taken with
Sbawayhi s other observations on elliptical utterances, see above with
xayra maqdamin) to permit the assumption than he was aware of the
performative aspects of y and other speech elements. In the us l treat-
ment, those who did regard y as representing an elided verb such as I
call are roundly rebuked by Ab l-H usayn (Mutamad, I:20). He frmly
rejects this on the grounds that merely to say the word y and its noun
conveys in itself the information that an act of calling has occurred. Here
he makes explicit what is lef implicit in Sbawayhi , as so ofen happens.
By Ab l-H usayn s time it had become a polemical issue, and he accuses
those who would restore a verb in this situation of either oversimplify-
ing for pedagogical purposes (how condescending!) or misinterpreting
what Sbawayhi meant by h arf nid a particle of calling, a performative
element by nature, which needs no verb, as Ab l-H usayn observes.
What this article has tried to show is that Sbawayhi , perhaps by his
own genius, perhaps encouraged and inspired by the community of
pioneering Islamic scholars around him, chose to treat language in a
manner which shows a remarkable afnity with modern Pragmatics .
Te nine points which Ali (2000, 3f) regarded as essential pragmatic
insights are taken from us l sources , but equivalents to most of them
can easily be found in the Kitb. Indeed it might have made a tidier ver-
sion of this paper simply to match them seriatim with material from the
Kitb.
Tese ideas were, so to speak, embryonic in Sbawayhi , but the mar-
riage of Aristotelian logic and Arab linguistic studies in the 4th/10th
century led to the birth of a much more self-conscious and program-matic
version of the same attitude in legal scienc e, reaching its peak of
42 michael carter
development in the 7th/13th century works which provide the bulk
of the material for Ali (2000). Symbolically we may point to the term
qnn law as an indication of the change of direction, though the term
has never carried much weight in medieval Islm . It would not have
meant much to Sbawayhi , who had half a dozen terms for the correct
use of Arabic , one of them sunna , nor is it prominent in medieval legal
contexts , but perhaps because it was circulating in the 4th/10th century,
e.g. in Frb (d. ca 339/950) Ih s 10, it may have fostered the notion
that a systematic code of behaviour along the lines of Greek ethics was
desirable and possible. As with grammar from that point on, works on
us l al-fqh display an impressive mastery not only of the large body of
historical Islamic data but also the methods of logical enquiry and dia-
lectical disputation. Te two disciplines henceforth continue in paral-
lel, grammar becoming increasingly legalistic to stop the language from
changing, and jurisprudence becoming more and more a grammatical
analysis of the unchangeable texts.
One result is that it was possible to formulate a set of fve principles of
communication which, as set out by Ali (2000, 64), present an instruc-
tive analogue to Grice s four maxims , here briefy paraphrased:
1. Te speakers disposition to make his intention manifest.
2. Te speakers truthfulness.
3. Tat what is said should have its due efect.
4. Tat what is said should be grasped immediately.
5. Tat the existing conventions should be maintained.
With great diligence and subtlety the classical us ls applied these and
other principles to the interpretation of God s speech in search of
Gods law as it has been put. Tis had not been Sbawayhi s goal, but
his exhaustive description of how the Arabic language works between
speaker and listener prefgured the Pragmatics of the us ls and pro-
vided a basis (fltered through the subsequent grammatical tradition )
for their scholarly exertions.
In a sense there was always an implicit Pragmatics in Arabic grammar
and Islamic law . On the macro-level Islm itself is a covenant with God ,
and Islamic society is a kind of social contract in which everything said
between Muslims has a contractual dimension, exactly as in Grice s per-
ception of speech. Tat is one reason why Muslims are enjoined not to lie
to each other, particularly when transmitting the religious knowledge on
which the survival of their faith depends. Sbawayhi could aford to take
truthfulness for granted, since it does not afect linguistic form, but the
pragmatics and contractual language in early arabic 43
us ls had to stipulate truthfulness as a condition of a valid legal utter-
ance precisely because there was no way to tell from the words them-
selves whether the speaker was lying, i.e. breaking Grices unenforceable
maxim of Quality, where lying is likewise formally undetectable.
Te last point to make is that all this legal matter found in the Kitb
goes to confrm two other aspects of Sbawayhi s life and work which tie
him closely to the lawyers. Firstly, from what little we know of his biog-
raphy we can deduce that he associated not only with revered authorities
on Arabic but also with early legal and religious scholars , and secondly
it is evident from the Kitb that these scholars were the inspiration for
much of his technical vocabulary and methodology. While this does not
solve the problem of the origins of grammar completely, it does make
it more likely that the Kitb is the creation of a single mind, an unpre-
cedented description of Arabic in all its domains, religious, poetic , pub-
lic and private, in a theoretical framework which drew deeply upon the
principles of the nascent legal system and owes almost nothing to exter-
nal traditions. Tis view may not meet with the approval of the dedica-
tee of the present volume, but it is a tribute to his belief in the right of
dissent that such heresies can be published without fear of legal action.
References
2.1 Primary sources
Ab H usayn , Mutamad = Ab l-H usayn al-Bas r, Kitb al-Mutamad f us l al-fqh .
M. H amdullh with M. Bekir and H . H anaf, eds . Damascus, 1964.
al-Bqilln , Taqrb = Ab Bakr Muh ammad b. at -T ayyib al-Bqilln , at-Taqrb wa-l-
ird as -s ar. Abd al-H amd b. Al, Ab Zayd, ed . Beirut, 1998.
al-Frb , Ih s = Ab Nas r Muh ammad b. Muh ammad, Ih s al-ulm. A. Gonzles
Palencia, ed . Madrid, 1932.
Ibn Mut , Fus l = Ab l-H usayn Yahy b. Abd an-Nr, Ibn Mut , al-Fus l al-xamsn.
Mah md Muh ammad at -T anh , ed . Cairo, 1976.
Ibn Walld , Intis r = Ah mad b. Muh ammad b. al-Wald, Ibn al-Walld, Kitb al-Intis r.
Monique Bernards , ed. Changing Traditions. Al-Mubarrads Refutation of Sbawayh
and the Subsequent Reception of the Kitb, Arabic pp. 1212. Leiden, New York, Kln,
1997.
ar-Rummn , arh = Ab l-H asan Al b. s ar-Rummn , arh Kitb Sbawayhi. MS
Feyzulla 1984.
a-aybn , Jmi = Muh ammad b. al-H asan a-aybn, al-Jmi as -s ar f l-fqh. Mar-
gin of Ab Ysuf , Kitb al-xarj. Blq, 1884.
Sbawayhi , Kitb = Ab Bir Amr b. Ut mn Sbawayhi, al-Kitb. (1) Hartwig Deren-
bourg, ed . Le livre de Sibawaihi. Paris, 18811889. Repr. Hildesheim , 1970. (2) Kitb
Sbawayhi. Blq Press, 18981900. Repr. Baghdad, 1965. (3) Kitb Sbawayhi. Abd
as-Salm Muh ammad Hrn, ed . Cairo, 196877 (cross-paginated with the Blq
edition). References to the Kitb are usually in the form Der/Bl.
44 michael carter
as-Srf , arh = Ab Sad as-Srf, arh Kitb Sbawayhi. MS Atif Efendi 2548.
a-irbn , Nr = Muh ammad a-irbn al-Xat b, Arab Linguistics, an introductory clas-
sical text with translation and notes [Nr al-sajiyya f h all alfz al-jurrmiyya].
Michael G. Carter, ed . Amsterdam. 1981.
2.2 Secondary sources
Ali Mohamed M. Yunis . 2000. Medieval Islamic Pragmatics. Sunni legal theorists models
of textual communication. Richmond.
Bergstrer, G . 1935. G. Bergstrers Grundzge des islamischen Rechts: bearbeitet und
herausgegeben von Joseph Schacht. Berlin, Leipzig.
Bravmann, Meir . 1953. Studies in Arabic and General Syntax. Cairo.
Buburuzan, Rodica . 1993. Exclamation et actes de langage chez Sbawayhi. Revue Rou-
maine de Linguistique 38, 421437.
Carter, Michael G. 2002. Patterns of reasoning: Sibawayhis analysis of the h l. Pro-
ceedings of the 20th Congress of the Union of European Arabists and Islamicists, Part
One, Linguistics, Literature, History [= Te Arabist, vol. 2425]. K. Dvnyi , ed. Buda-
pest. 315.
. 2003a. Legal Schools and Grammatical Teory. Arabistikai islamoznanie. Tom 2.
Studi po sluchai 60godishnata na dots. d.f.n. Penka Samsareva, Simeon Evstatiev , ed.
Sofa. 177183.
. 2003b. Talking with and about God, Adam and the Arabic language. Majz,
culture e contatti nellarea del Mediterraneo. It ruoli dell Islam (21st Congress of the
Union of European Arabists and Islamicists, Palermo 2002) [= La Memoria vol. 15].
Antonino Pellitteri , ed. Palermo. 197208.
Crystal, David . 2000. A Dictionary of Linguistics and Phonetics. 4th ed. Oxford.
Fleischer, Heinrich Leberecht . 188588. Kleinere Schrifen, gesammelt durchgesehen und
vermehrt von A. Huber, prof. Torbecke, und F. Bhlau. Leipzig. Repr. Osnabrck.
1968.
Grice, H. Paul . 1989. Logic and Conversation. Studies in the Way of Words, 2240.
Cambridge, Mass. and London.
Khadduri, Majid . 1987. Al-Imm Muh ammad ibn Idrs al-Shfs al-Risla f us l al-
fqh. Treatise on the foundations of Islamic jurisprudence. Translated with an introduc-
tion, notes and appendices. 2nd ed. Cambridge.
Larcher, P. 1990. lments pragmatiques dans la thorie grammaticale arabe post-clas-
sique. Studies in the History of Arabic Grammar II. Kees Versteegh , Michael G. Carter ,
eds. Amsterdam. 193214.
. 1998. Une pragmatique avant la pragmatique: medivale, arabe et islamique.
Histoire, Epistmologie, Langage 20, 101116.
Loucel, Henri . 1963, 1964. Les origines du langage daprs les grammairiens arabes.
Arabica 10, 188208, 253291; 11, 5772, 151187.
al-Marnn, Burhn ad-Dn . 1870. Te Hedaya or Guide: a commentary on Musulman
laws, trans. by Charles Hamilton , 2nd edition. Standish G. Grady, ed. London (refer-
ence is to the reprint 1963).
Moutaouakil, Ahmad . 1990. La notion dactes de langage dans la pense linguistique
arabe ancienne. Studies in the History of Arabic Grammar II. Kees Versteegh , Michael
G. Carter , eds. Amsterdam. 229238.
Simon, Udo . 1993. Mittlelalterliche arabische Sprachbetrachtung zwischen Grammatik
und Rhetorik. Heidelberg.
Troupeau, Grard . 1976. Lexique-index du Kitb de Sbawayhi. Paris.
Versteegh, Kees [C.H.M.]. 1977. Greek Elements in Arabic Linguistic Tinking. Leiden:
E.J. Brill.
ID MR IN THE MAN OF AL-FARR:
A GRAMMATICAL APPROACH BETWEEN DESCRIPTION
AND EXPLANATION
Kinga Dvnyi
Corvinus University, Budapest
1. Introduction
Kees Versteegh stimulated the discussion on the history and develop-
ment of Arab grammatical thinking in a number of his publications.
In one of his books (Versteegh 1993, 150), he refects upon my earlier
analysis of al-Farr s linguistic methods in his Man (Dvnyi 1990).
He pointed out the insufcient analysis of idmr and its related terms
in this authors work. In another chapter of the same work (Versteegh
1993, chapter fve), he assembled data in an attempt to present the inter-
relationship between grammarians , readers and commentators who
worked in the 2nd/8th century.
Te present contribution would like to pick up these two threads and
examine, on the one hand, in some detail the role of idmr in the Man
l-Qurn of al-Farr and, on the other hand, analyze the role and place
of this grammatical commentary of the Qurn from the point of view of
other grammars (mainly Sbawayhi s Kitb) and other exegetical works ,
like for example those of al-Axfa and at -T abar .
It is a well-known fact that al-Farr (d. 207/822) held in great esteem
Sbawayhi s (d. 180/796) Kitb, which more than twenty years predated
his composition. Teir starting points and approaches were, however,
widely diferent. Versteegh (1993, 180) has already pointed out that the
interests of the two authors lay elsewhere. To this, we can add that al-
Farr and Sbawayhi, though working within the framework of prac-
tically one grammar or one grammatical idealhad widely diferent
aims. While al-Farr, in his Man, used his grammatical knowledge
for the analysis of an existing corpus which he described from the point
of view of the listener to this text, Sbawayhi aimed at creating, from
the point of view of the speaker, a comprehensive grammar in which he
used poetical and Qurnic excerpts only by way of illustration.
46 kinga dvnyi
2. An overview of al-Farr s methods
In order to have a brief overview of the methods used by al-Farr , a
few examples from the beginning of the Man will be presented frst.
Tat the corpus al-Farr is working on is the text of the revelation has
special importance. On the one hand, al-Farrlike grammarians and
later rhetoricians considers that irb is a necessary prerequisite for
the understanding of any text and so the text of the Qurn . On the other
hand, since the Qurnic text is usually understood without relying on
the irb endings , there is a strong tendency to analyze diferent end-
ings at a given place without entailing a change in the meaning.
1
Tis
method aims at eliminating the problems posed by the diferent qirt .
Although in most of the cases the grammatical analysis only underlines
and systematizes the interpretation given by the frst exegetes of the
Qurnic text , in several cases, however, it is the grammatical analysis
which helps to disclose the meaning of the ya .
2.1 Man I:3 regarding Q 1:2 al-h amdu li-llhi
Te task here is to determine the vowel ending of the word al-h amd .
Step (1): Te examination of the readings (qirt ):
According to al-Farr the readers are in total agreement concerning the
raf ending.
2
It is also interesting to note, that al-Farr only mentions
this ending but does not present a grammatical explanation for it.
3
Step (2): Te elicitation of extra-textual linguistic source:
Te Bedouins (ahl al-badw) say three things:
(a) al-h amda li-llh,
(b) al-h amdi li-llh,
(c) al-h amdu lu-llh.
1
A notable exception, where diferent readings refect a diference in meaning is e.g.
Q 5:6. For the analysis of this ya , see Dvnyi 198788 and Burton 1988.
2
Makram and Umar (1985, I:5) also list al-h amda and al-h amdi among the readings
of this verse .
3
Tis is in contrast with al-Axfa (Man I:9 f.) whose analysis at this place is rather
similar to that of al-Farr , but who also provides a detailed grammatical analysis of the
raf ending .
ID MR in the MA
N of al-farr
47
Step (3): Te explanation of the variants:
For (a), i.e. the nas b ending , al-Farr
(i) gives a grammatical rule according to which h amd is a mas dar
in place of which a verb could also have been used. Tis is an
example that fxes the rule;
(ii) supports it with other similar instances in the Qurn , like:
Q 47:4: fa-id laqtumu llad na kafar fa-d arba r-riqbiinstead
of which one could say in kalm : fa-d rib r-riqba
Q 12:79: mad a llhi an naxud a . . .which is the same as:
nad u bi-llhi;
(iii) and props it with Bedouin usage (qawl al-arab) where saqyan
laka may be used instead of saqka llh.
Te (b) variant, i.e. al-h amdi li-llh is treated as one word. As such, it is
compared to ibil , where two i vowels follow each other. Tis is a descrip-
tive explanation making reference to usage. Other parallels are pre-
sented as well, among them the (c) variant: al-h amdu lu-llh. Al-Farr
approached this corpus from the point of view of the listener to the text.
And it seems from the second step employed by al-Farr that the listener
could hear some variants. And because al-Farr was interested not only
in the text of the Qurn , but obviously placed it in the context of the
Arabic language as a whole, he analyzed these versions. What is even
more, he proceeded to explain these variants in detail thoughaccord-
ing to his knowledgethese variations were not Qurnic readings .
2.2 Man I:78 regarding Q 1:7 [. . . anamta alayhim] ayri
l-madbi [alayhim wa-l d-dllna]
Te task here is to determine the vowel ending of the word ayr. Tere
are neither variant readings nor extra-textual variants.
Step (1): Grammatical analysis:
Al-Farr explains the i in ayri as a nat to allad na, mentioning that it is
defnite (marifa) because of the following word (al-madbi).
4
He also
4
It is important to note in this respect, that while the analysis of al-Axfa goes along
the same lines (Man I:18) (with the usual diference in terminology ), but with one
basic diference, i.e. that he does not defne negatively the ending of the word, in other
words he does not say what it is not, rather contents himself with saying what it is. Tat
is to say, he does not deal with the refutation of grammatically incorrect endings or with
48 kinga dvnyi
categorically refuses to connect ayri to the preceding word, alayhim.
Doing this he seems to argue against the view of those among his con-
temporaries who carry the surface descriptive analysis or the analysis
based upon proximity (itb) to the extremes.
Step (2): Semantic analysis:
Turning to the semantic side of the explanation, he states that ayri is
connected to wa-l d-dllna which follows it and its meaning is l. He
also adds that if it meant siw, it could not have been followed by l. So
it can be established that al-Farr connects the two types of analysis, the
semantic and the grammatical .
2.3 Man I:1112 regarding Q 2:2 [d lika l-kitbu l rayba fhi]
hudan li-l-muttaqna
Te question is whether hudan is in raf or nas b . Tough in the case
of this particular word both endings are realized in the same surface
form, the question should be decided both from the point of view of the
semantic interpretation of the structure and for the sake of other similar
structures in the Qurn where even the surface realization is diferent.
In the case of similar phrases in the Qurn a number of qirt have
been preserved with readings in both raf and nas b, as e.g. Q 31:13 (
lm tilka yt al-kitb al-h akm hudan wa-rah matun/an li-l-muh sinn)
and Q 11:72 (a alidu wa-hd bal ayxun/an).
Step (1): Grammatical statement:
Both endings can be explained in diferent ways (wajh ). If the interpreter
of grammatical structures does not deviate from the intended mean-
ing, he may freely choose between raf and nas b or the diferent ways.
Tis freedom is expressed by al-Farr in the use of the 2nd person: id
aradta . . ., wa-in jaalta . . . rafata . . . (Man I:11, 12f.).
Step (2): Te detailed grammatical analysis:
(i) raf
(1) It can be the xabar of d lika l-kitb whichin this caseis ana-
lyzed as the mubtada . A paraphrase is given: d lika hudan.
the refutation of grammatically not permissible analyses. Tis diference will remain
characteristic throughout the two books.
ID MR in the MA
N of al-farr
49
(2) It can be a nat (modifer) of the xabar : l rayba fhi. A Qurnic
parallel is quoted: wa-hd kitbun anzalnhu mubrakun (Q
6:92, 6:155).
(3) It can also be considered the beginning of a new structure which
comes afer a complete phrase (al stinf li-tamm m qablahu).
5
A Qurnic parallel is quoted: wa-hd bal ayxun (Q 11:72).
6
(ii) nas b al l-qat
(1) It is either cut of from al-kitbu, which in this case would be
analyzed as the xabar of d lika.
(2) Or it is cut of from the -hi in l rayba fhi. A paraphrase is given
as: l akka fhi hdiyan.
A ready-made grammatical rule is given as an explanation for both pos-
sibilities: li-anna hudan nakira ittas alat bi-marifa qad tamma xabaruh
fa-nas abtah li-anna n-nakira l taknu dallan al l-marifa.
2.4 Man I:1415 regarding Q 2:16 fa-m rabih at tijratuhum
Te task is to defne the reason why certain structures are permissible in
contrast to other seemingly similar structures that are not.
Step (1): Parallels from kalm al-arab :
hd laylun nimun
Step (2): Parallels from the Qurn :
Q 47:21 fa-id azama l-amru
Step (3): Semantic defnition
Te permissibility of specifc structures greatly depends upon their
communicational value, i.e. they can only be permitted if they can be
understood unambiguously (ulima manhu I:14, 17). In these exam-
ples, the reference of the verbs and the adjective is unambiguously not to
5
In connection with this third analysis, attention should be called to the difculties
of grammatical and semantic analysis arising from the lack of punctuation. No wonder
that writings on qat and istinf developed into a special branch of the Qurnic sci-
ences . On the use of istinf and related terms in early grammar and exegesis in general,
see Versteegh 1993, 132136 and the literature cited there.
6
Te reading in nas b can be found at this place in todays printed editions, and this
was al-Farr s reading as well (Man I:12, 2). Te raf ending was read by Ibn Masd
and Ubayy among others (see Makram and Umar 1985, III:125).
50 kinga dvnyi
the other word in the phrase which is a common noun (tijra, layl, amr)
but to the people who act behind these words. Similar structures are
not permitted in the case of possible ambiguity, e.g. *qad xasara abduka
is not permitted because of the ambiguity of the word abd (he can both
be trader and the object of trade), as al-Farr puts it, its meaning cannot
be known: fa-l yulamu manhu (Man I:15, 3).
2.5 Man I:16 regarding Q 2:1718 wa-tarakahum f ulumtin . . .
s ummun bukmun umyun
Te task is to explain the endings of s umm, bukm and umy.
(i) Te raf ending is explained by istinf and the completeness of the
preceding clause in itself (li-anna l-kalm tamma). An additional
reason is that the whole phrase is split between two ya s, but this is
not a prerequisite of istinf . It should, however, be mentioned that
istinf is frequently found at the beginning of yas.
(ii) Te nas b reading (summan bukman umyan) is explained in two
ways:
(1) according to the meaning (al l-man), i.e. by referring back
to the verb in the preceding clause (tarakahum);
(2) by their being expressions of blame (d amm).
A parallel Qurnic passage quoted by al-Farr is Q 9:111112:
wa-man awf bi-ahdihi mina llhi
. . . at-tibna l-bidna l-h midnaexplained as istinf
or the reading in nas b
. . . at-tibna l-bidna l-h midnaexplained as qat
Te use of the explanatory terms istinf and qat shows clearly that for
al-Farr the most important issue was to understand the text as a whole.
And one of the frst issues to be dealt with was the correct segmentation
of the text. Te term qat expresses more clearly the syntactic structure ,
whereas the term h l can only refer to the meaning.
7
2.6 Man I:17 regarding Q 2:19 yajalna as biahum . . . h ad ara
l-mawti
Meaning, however, can also play a decisive role in the explanation of an
irb ending and the determination of the syntactic structure , as it is the
7
For the diference in al-Farr s usage between hl and qat , see Kinberg 1996, 194.
ID MR in the MA
N of al-farr
51
case, e.g. in Q 2:19 (Man I:17, 1f.): yajalna as biahum . . . h ad ara
l-mawti. Te word h ad ara is in nas b not because the verb yajalna
afects it (*yajalnah h ad aran) but by way of specifcation (tafsr ). It is
the same as e.g. in the phrase at aytuka xawfan where fear is not given
but something is given because of it.
3. Te explanatory technique of idmr in the Man
Te above examples were presented to show the steps followed by
al-Farr in his explanations.
One of the basic explanatory techniques used by al-Farr in the
Man is the term idmr (together with its related forms). In the course
of the explanation of the text of the Qurn , al-Farr is basically con-
fronted with three types of problems:
(i) Te meaning of the ya needs clarifcation;
(ii) Tough the meaning is clear, there is a problematical irb - ending ;
(iii) Te qirt of similar structures at diferent places in the Qurn
should be harmonized and explained.
It will be seen that the technique of idmr features as a prominent solu-
tion in all of these cases. It is outside the scope of the present paper
to examine all the occurrences of idmr and its related terms in the
Man. In the following, however, we shall try to present a typology of
the most signifcant types of occurrences of this term.
3.1 idmr is supported by another place in the Qurn where the
suppressed element occurs in the text
3.1.1 Man I:13 regarding Q 2:7 xatama llh al qulbihim wa-al
samihim wa-al abs rihim iwatun
Tis ya has a variant reading, iwatan, which entails no difer-
ence in the meaning of the ya (wa-manhum wh id).
8
Tis read-
ing is explained by otherssays al-Farr by the idmr of jaala. Tis
8
It might be interesting to note that al-Axfa did not mention the existence of difer-
ent readings here (see al-Axfa, Man I:34).
52 kinga dvnyi
is supported by Q 45:23 where the phrase actually occurs as wa-jaala
al bas arihi iwatan. Al-Farr accepts this explanation, i.e. the use of
another Qurnic place to support the explanation by idmr, though he
emphasizes that were it not for the Qurnic parallel, the idmr in this
sentence could not have been accepted. Tis is because according to
al-Farr, idmr is acceptable only in those utterances (kalm ) which
are coherent (yajtami ), i.e. where the beginning refers (yadullu al) to
the end. Te idmr is acceptable (yah sunu) if it is well known (urifa; or
at other places malm).
A kalm example is brought in to illustrate what is meant by the
term yadullu al, and how it operates in structures. Te example is as
follows: qad as ba fulnun al-mla fa-ban d-dra wa-l-abda wa-l-
ima wa-l-libsa l-h asana. It can be seen from this example, proceeds
al-Farr , that the notion of building (bin) cannot be extended to
slaves and clothes. By quoting this example from kalm, al-Farr appeals
to the listeners linguistic insight and competence in the Qurnic text .
He explains that the frst part of the utterance (as ba) refers to the
slaves, etc., i.e. qad as ba fulnun al-mla fa-ban d-dra wa-[as ba]
l-abda wa-l-ima wa-l-libsa l-h asana. Without saying that it should
be repeated, instead he uses the term dalla, yadullu al. Sbawayhi would
rather say explicitly that as ba should be repeated before al-abda.
Al-Farr only hints at the idmr of as ba.
In the following (Man I:14), Q 56:22 is mentioned as a similar
example: yat fu alayhim wildnun muxalladna bi-akwbin . . . wa-
h rin nin.
9
Al-Farr accepts the -in reading on the basis of proxim-
ity (itb ) but mentions that those who put it in raf (wa-h run nun)
do this on the basis of the meaning, since the cups cannot be co-ordi-
nated with the beautiful companions, and by the idmr of a word like
indahum, fh, maa d lika. Again, as in the previous example from
kalm , al-Farr does not use the term idmr here, but only hints at it.
Te limits of idmr are imposed by understandability and coherence,
or rather their lack. Tis is called by al-Farr qillat al-ijtim. So. e.g. the
idmr of qataltu is not permissible in the following utterance: darabtu
fulnan wa-fulnan [*qataltu]. Te reason is that there is no reference
9
Mainly because he considered it the accepted reading. Tis was the reading of a
great number of readers, among them al-Kis (cf. Makram and Umar 1985, VII:65),
but in todays printed mus h af the -un reading can be found.
ID MR in the MA
N of al-farr
53
(dall ) to the meaning here, the speakers intention cannot be known by
the listener.
At -T abar adds (Tafsr I:87) that xatama cannot refer to the eyes
because they are never described by xatama either in the Qurn or in
the kalm .
3.1.2 Man I:35 regarding Q 2:72: wa-id qataltum nafsan
fa-ddratum fh
Te jawb of id is present neither in this ya nor several others simi-
lar to it in the Qurn . Al-Farr explains the structure by the idmr of
ud kur before id, and supports this explanation by similar utterances
in the Qurn where this word is present, as e.g. in Q 8:26 (wa-d kur id
antum qall . . .). Suppression of ud kur with id can be inferred from
places like Q 8:26 and Q 7:86. al-Farr adds that if ud kur was not
mentioned in Q 7:86, you could still infer that this is what is intended
because it occurs previously.
A parallel case of idmr is also mentioned: Q 7:73 (wa-il Tamda
axhum Slih an) where the known meaning allows the explanation of
this structure by the idmr of arsaln, i.e. arsaln Slih an. Tis exam-
ple in itself can be regarded as a sub-case of the acceptance of idmr on
the basis of parallel Qurnic passages, since though the verb arsala is
used in the Qurn , it is mainly the content of the ya which determines
the selection of the suppressed element.
10
3.1.3 Man I:166 regarding Q 2:246 wa-m lan all nuqtila
Al-Farr starts his analysis with comparing similar utterances from dif-
ferent parts of the Qurn , some of them with an and some other with-
out it, like e.g.:
Q 14:12: m lan all natawakkala al llh
Q 57:8: wa-m lakum l tuminna bi-llhi wa-r-raslu yadkum li-
tumin bi-rabbikum
Dropping (ilq) an here cannot be considered defciency (illa) accord-
ing to al-Farr because of its frequent usage in the arabiyya . Te usage
of an relies on the meaning of manaa. According to this explanation,
10
On the meaning of this term in the Man, see Kinberg 1996, 377.
54 kinga dvnyi
the utterance m laka l tus all f jama means: m yamnauka an
tus alliya. Te an may be inserted into the m laka phrase because the
latters meaning is identical with manaa. And a last evidence is put
forward from the text of the Qurn itself: Q 7:12: m manaaka all
tasjuda id amartuka runs parallel to Q 15:32: m laka all takna maa
s-sjidna.
3.2 idmr is supported basically by kalm
3.2.1 Man I:32 regarding Q 2:48 wa-ttaq yawman l tajz nafsun
an nafsin ayan
Al-Farr explains this passage by the idmr of fhi afer tajz, insisting
on the necessity of distinguishing by reference to the meaning adver -
bial structures allowing idmr from verbal structures with s ifa (prepo-
sitional phrase ) complement, where idmr is not possible. For example,
anta llad takallamtu fhi should be distinguished from anta llad takal-
lamtu. In doing this he argues with al-Kis and others who understand
too rigidly the formal description and identify the two, formally simi-
lar but semantically diferent types. On the basis of this identifcation
al-Kis refutes both structures with idmr while others permit both
of them. Te same holds true, says al-Farr , for the fhihu alternation
within adverbial structures of time and place but not in verbal prepo-
sitional phrases . Te phrase tka yawma l-xams is interchangeable
with tka f yawmi l-xams, since the s ifa and h here agree in meaning
(muttafaq manhum). But when the meaning difers it is not allowed
to suppress f in place of h and vice versa: yuh ibbuh does not equal
with yuh ibbu fh.
at -T abar , who quotes extensively (Tafsr I:203, 11f.) al-Farr s inter-
pretation, adds that the idmr in this ya is possible because it is well-
known (malm), this being the generally accepted explanation for
idmr by the exegetes (as h b at-tawl ).
3.2.2 Man I:113, 9, Q 2:185: wa-li-tukmil l-idda
Te li- + verb cannot be art
11
(cause) of a previous verb because the ww
blocks its impact. Al-Farr discusses in many places that some particles ,
11
On the meaning of this term in the Man, see Kinberg 1996, 377.
ID MR in the MA
N of al-farr
55
especially the ww, do not allow us to suppose a grammatical link with
a preceding regent. Tis phenomenon is explained on the basis of kalm
examples: jituka li-tuh sina ilayya cannot be transformed to jituka
wa-li-tuh sina ilayya only if you mean (jituka) wa-li-tuh sina ilayya
jituka. Tis second occurrence of the same verbal phrase is, naturally,
suppressed (al-arab tudxiluh f kalmihim al idmr fl badah).
Tis rule stated, al-Farr cites many Qurnic places where the
same causal structure wa-li and the idmr of a verbal phrase can
be observed signifying the result of this cause, e.g. wa-kad lika nur
Ibrhma malakta s-samawti wa-l-ardi wa-li-yakna min al-mqinna
(Q 6:75) where the suppressed (mudmar) phrase is a repetition of the
verbal phrase before ww: araynhu.
At -T abar (Tafsr II:88, 3f.) quotes two opinions. One is a verbatim
quotation of al-Farr s view without mentioning his name (qla bad
nah wiyy l-Kfa). Te other is the view of bad ahl al-arabiyya which
was rejected by al-Farr. Tis supports our analysis that whenever
al-Farr rejects a seemingly only theoretical possibility of analysis, he
rejects the opinion of a certain group of people without attributing their
view to anybody.
3.2.3 Man I:271 regarding Q 4:46 mina llad na hd yuh arrifna
l-kalim
Te meaning is: man yuh arrifna l-kalim. Tis is supported by kalm
examples where man is suppressed in the beginning of the utterance
(mubtada al-kalm ): minn [man] yaqlu or minn [man] l yaqlu.
Tis is possible according to al-Farr , because min, i.e. the frst part
(awwal ) of the utterance refers to the meaning of what has been lef out,
being its part (bad). Al-Farr considers it important to note that man
cannot be lef out (idmr ) as a rule, but only in those cases where the
prepositional phrase refers to it. Tis latter sometimes may be f too: fh
s lih na wa-fh dna d lika.
3.3 idmr is supported by the immediately preceding context (the
frst part of the utterance or the previous utterance): Man I:141142
regarding Q 2:220: wa-in tuxlithum fa-ixwnukum
With idmr of the rf , i.e. the nominal subject : hum, or if it is nas b , with
idmr of the second occurrence of the verbal phrase : fa-ixwnakum
tuxlit na. In both cases the suppressed element is given in the frst
56 kinga dvnyi
part of the utterance or, sometimes, in a previous utterance, as in
Q 2:239: fa-in xifum fa-rijlan. Tis is called ijtim al-kalm . Here,
says al-Farr , it would not be correct to suppose huwa because it is not
an (everlasting) state (dim ) but an action. Te previous ya (h f
al s -s alawti . . . wa-qm li-llhi . . .) gives a clue for the explanation.
Hence the meaning can be restored as: in xifum an tus all qiyman
fa-s all rijlan.
At -T abar (Tafsr II:208) follows al-Farr s argument, and refers to the
same Qurnic passage, etc. but everything is more detailed. It is as if he
worked from a fuller version of the Man.
3.4 idmr is supported primarily by the readings of Abdallh and
Ubayy: Man I:156 regarding Q 2:240: wa-llad na yutawafawna
minkum wa-yad arna azwjan was iyyatan
Here al-Farr presents two readings, nas b (was iyyatan) and raf
(was iyyatun), and both of them are explained on the basis of idmr . Te
frst idmr relies on the rule of the absolute object , the suppressed verb
being a jussive of the same root (li-ys . . . was iyyatan). He refuses, how-
ever, to accept the supposition that the previous verb in the yad arna
azwjan phrase could have put was iyya into nas b. Tis may have been
held by the supporters of the proximity-theory who always preferred the
explanation on the basis of the neighboring word.
Te second, raf , reading (was iyyatun) may also be explained by
idmr . Te source of this idmr is the reading of Ibn Masd or Ubayy:
12
fa-matun li-azwjikum or kutiba alaykum al-was iyyatu li-azwjikum,
both containing raf , the frst also a lexical variation.
12
From among the 64 Qurn readers whom al-Farr quotes by name, Ibn Masd
is by far the most frequently quoted, with 411 references. Ubayy is the seventh most
frequently quoted reader, with 92 occurrences. Cf. Dvnyi 1991, 160161.
ID MR in the MA
N of al-farr
57
3.5 Te use of major, meaning-triggered grammatical rules in
supposing idmr
3.5.1 h l
3.5.1.1 Man I:24 regarding Q 2:28: kayfa takfrna bi-llhi
wa-kuntum amwtan
Te meaning is established as qad kuntum, since the second verb is a
h l to the frst one and it refers back to a state prior to the past action
denoted by the frst verb.
Afer the defnition of the meaning, al-Farr formulates the following
rule: this type of h l necessitates qad , either overtly (ihr ) or in a sup-
pressed way (idmr ). If this condition is not fulflled, i.e. the frst verb
does not refer to the past (like e.g. in the case of kda, as) then the
second cannot contain qad in either way.
Tis rule is reinforced by another type of proof, a parallel place in
the Qurn where qad is being suppressed and where it is generally
understood to be necessary for the correct meaning (Q 12:27: in kna
qams uhu qudda min duburin fa-kad ibat).
At -T abar (Tafsr I:146, 18f.) follows al-Farr s arguments but he
adds that the explanation for the suppression of qad lies in the fact that
if the verbal form faala takes the place of h l it is self-evident (malm)
that it requires qad. It is to be emphasized that al-Farr also considers
that only those elements can be suppressed which are well-known to the
speakers.
3.5.1.2 Man I:372 regarding Q 7:4: fa-jah basun baytan aw
hum qilna
Te meaning is established: [wa]-hum qilna is a h l , parallel to
baytan. Its grammatical rule is: wa-huwa filun, accordingly the idmr
of wa- is compulsory. Te wa- should be present either overtly or in a
suppressed way. All this is supported by an example from kalm where
both ways (ihr and idmr ) are correct: ataytan wliyan aw wa-an
mazl or awan mazl.
13
13
Az-Zajjj (Man II:317) refutes this explanation without mentioning al-Farr
and says that it is not necessary to suppose wa- in h l in general.
58 kinga dvnyi
3.5.2 Te delimitation and rules of idmr in quotation (h ikya )
3.5.2.1 Man I:38 regarding Q 2:58 wa-ql h it t atun
Considering h it t atun as a h ikya entails the idmr of the rf (i.e. the
nominal subject ) which is supposed to be m umirtum bihi or simply
hiya. Tough not called idmr in this place, it is analyzed as such, as will
be seen below.
It is a quotation (h ikya ) which presupposes that it should be correct
(s aluh a) if the rf /ns ib or xfd is suppressed. Al-Farr exemplifes
this rule as follows: qultu l ilha ill llh fa-yaql al-qil: qultu kali-
matan s lih atan. But kalimatan s lih atan cannot stand alone without a
ns ib.
To decide whether an utterance is quotation or indirect speech is very
important for the Qurnic text where change(s) of speaker and hearer
can occur even within the same ya .
3.5.2.2 Man I:40 regarding Q 11:69: ql salman qla salmun
Te word salm is used here in two meanings (al manyayni). In the
frst step the two endings (nas b and raf ) are explained. Te explanation
of the endings relies on the grammatical rule diferentiating indirect
speech from quotation (h ikya ) . Having fxed this al-Farr adds that
in the case of h ikya the idmr of alaykum should be supposed. In a
third step he demonstrates that this idmr is possible because salm
frequently occurs in kalm alone, i.e. without alaykum.
3.5.2.3 Man I:93 regarding Q 2:154: wa-l taql li-man yuqtalu f
sabli llh amwtun
It is another instance of h ikya where the raf ending is explained by the
idmr of hum.
14
Te irb ending and the diference between quotation
and indirect speech is demonstrated by substitutional analysis (huwa bi-
manzilat . . .). He says that the nas b ending is not permitted here because
amwt is an ism and not a qawl , i.e. it is not a nominalized phrase .
He illustrates the possible and impossible usages as follows:
qultu laka xayran, i.e. kalman h asanan.
qultu laka xayrun, is similar to: qultu laka mlun.
14
Az-Zajjj (Man I:229) also accepts this analysis.
ID MR in the MA
N of al-farr
59
In the case of xayr both structures are correct whereas their substitutes
cannot be used in the other structure. Since there are no diferent read-
ings here it might be supposed that by illustrating these structures in
detail al-Farr may possibly refute analyzing techniques that take into
account only the surface structure.
3.5.2.4 Man I:296, 7 regarding Q 4:171: wa-l taql t alt atun
Tis and similar examples show that it is a general grammatical rule that
necessitates the use of idmr , scil. if there is a marf afer the qawl (as
a quotation afer the verb qla), there must be a rf (subject) as well, in
the above example: wa-l taql hum t alt atun
General rule: if something is in raf afer qla (yaqlu, qawl ), there
should be a rf either overtly or in a suppressed way (idmr ). Here al-
Farr gives a formal explanation of an irb ending .
3.5.3 Te id in the relative clause: Man I:157 regarding Q 2:246
ibat lan malikan nuqtil f sabli llhi
Te verb may not be nuqtilu (in raf ) referring to the preceding noun
as a s ila , because there is no reference back to malikan. However, if one
accepts the yuqtil reading, it can be explained as either s ila (yuqtilu)
or jaz (yuqtil ) afer an imperative (amr) , as in the case of nuqtil.
Al-Farr shows that the form yuqtilu can also be used when there is
no antecedent , with idmr : ibat lan llad yuqtilu. Ten he presents a
kalm example: allimn ilman antafu bihi and allimn llad antafu
bihi. But if bihi is dropped, then only the jazm is correct, since there is
no reference back (id ).
3.5.4 jaz: Man I:178, 4, Q 2:265: fa-in lam yus ibh wbilun
fa-t allun
Al-Farr here refers to the grammatical rule of jaz that requires md
forms , thus he supposes the suppression of kna before t allun. Te
interesting point of his analysis is, however, when he says: udmirat kna
fa-s aluh a l-kalm, because it sheds light on the meaning of idmr : It is
not simply suppression or deletion but the supposition of a suppressed
element which corrects the utterance and makes it ft the grammatical
rules .
60 kinga dvnyi
3.5.5 amm . . . fa-: Man I:228 regarding Q 3:106: fa-amm llad na
swaddat wujhuhum a-kafartum
Tis is a good example of a well-known grammatical rule (the obliga-
tory use of fa- afer amm ) which makes the grammarian suppose the
idmr of fa-. At the same time the meaning of the ya also requires the
supposition of a suppressed phrase yuql. Afer dropping yuql, how-
ever, the fa- is also dropped. In reconstructing the meaning one needs to
reconstruct the utterance as well: fa-amm llad na swaddat wujhuhum
fa-yuql a-kafartum.
In the above fve examples idmr is necessitated by some basic gram-
matical rules considered by al-Farr to be generally accepted and evi-
dent to such an extent that he does not even try to explain them with the
exception of the last example.
3.6 Te impossibility of idmr: Man I:195197 regarding Q 3:15: qul
a-unabbiukum bi-xayrin min d likum li-llad na ttaq inda rabbihim
janntun
Jannt is in raf because of the infuence of lm. Tis intervening lm
prevents referring back to the beginning of the utterance (awwal
al-kalm).
He then draws the conclusion: the xfd (here bi-) cannot be sup-
pressed (lam yudmar), therefore one cannot suppose that the bi- (in bi-
xayrin) is suppressed before jannt causing it to be in xafd . Tis also
means that the irb -ending xafd cannot be used independently. Te
same principle is stated in connection with Q 6:96 (Man I:346) where
a h l stands in the way of the efect of a preceding verbal noun (jil):
wa-jilu l-layli sakanan wa--amsa wa-l-qamara h usbnan. Since
a-amsa and al-qamara cannot be in xafd , therefore in al-Farr s view
they take nas b ending according to their meaning.
4. Concluding remarks
4.1 A summary of idmr types
To sum up the method of idmr , it can be established that al-Farr
derives the explanations of specifc Qurnic passages from the follow-
ing main types of sources:
ID MR in the MA
N of al-farr
61
(i) the end of an utterance is related to its beginning, the preceding
context (sometimes on the basis of the diferent segmentations of
the text), whereas al-Farr (Man I:13, 1415) puts itwa-
innam yah sunu l-idmru f l-kalmi llad yajtamiu wa-yadullu
awwaluhu al xirihi, i.e. the context is coherent. It is contrasted
with inqat aa, where the context is incoherent.
(ii) another place in the Qurn ;
(iii) an extra-textual linguistic source : kalm al-arab/badw or some-
times a poem , though it should be pointed out that poems do not
have their central signifcance they have in the Kitb;
(iv) a grammatical rule ofen quoted without example or proof. Its main
types are:
(a) only the rule is mentioned;
(b) a rule-like example is given (kam taql . . .);
(c) a grammatically incorrect example is used to support his rea-
soning (l yajz . . .);
(v) a qira ;
(vi) the codex of Abdallh b. Masd or, less frequently, that of Ubayy
b. Kab or the codices of their followers which sometimes contain
non-canonical variations.
4.2 Te meaning of idmr
Te meaning of idmr is not simply h ad f (deletion), but rather the
obligatory supposition of an element. Tus when al-Farr writes that l
budda min idmr kna li-anna l-kalm jaz (Man I:178, 4) it means
just the opposite of obligatory h ad f (the obligatory deletion); it means
that it is obligatory to suppose the existence of an element either present
in the utterance or suppressed.
Te use of the terms uskitat and ulqiyat also sheds light on the main
characteristic of idmr , i.e. its obligatory nature. For example at Man
I:163 regarding Q 2:246 there are two options to express the same mean-
ing: either with all or only l. Te dropping of an, however, cannot be
considered idmr, since both variants are equally correct Arabic (al
wajh al-arabiyya), the structure with only l not being brought back to
the one with an.
62 kinga dvnyi
4.3 Two kinds of traditions in the Man
It can be established that al-Farr relied on two kinds of traditions in
his Man.
I Te grammatical tradition
A) al-Farr and Sbawayhi
Disregarding the terminological diferences it can be stated that on the
whole he followed the same grammatical tradition that became exem-
plifed by Sbawayhi s Kitb.
a) He takes sides with a Sbawayhi -like grammatical analysis on the basis
of the mil theory , without, however, using the same terminology.
15
b) Frequent reference can be found to grammatical rules considered
self-evident and thus in no need of further explanations.
c) We can see the main diference between the Man of al-Farr and
Sbawayhi s Kitb not in the method of grammatical analysis but in
their approach. Al-Farr, starting from a complete text, always has
in mind the text as a whole, as a series of utterances, and he proceeds
accordingly. Whereas Sbawayhi cites only specifc examples to illus-
trate the grammatical rules he wishes to present.
B) al-Farr and al-Axfa
Te diference between al-Farr and al-Axfa lies in the fact that the
latter basically writes about grammar , while the former deals essentially
with Qurnic exegesis on a grammatical basis. Tis may account for
al-Axfa s lack of interest in eliciting all the known readings at a given
place whereas al-Farr seems keen on mentioning whatever can be
heard (see, e.g. al-h amdu/al-h amda/al-h amdi above), whereas does not
seem to attach great importance to the explanation of structures or end-
ings that he considers self-evident.
15
According to Talmon (2003, 309312) Sbawayhi s main concern in syntax is irb
carried out by amal efect. Talmon also postulated that in the Kitb al-h udd, al-Farr
seemed to focus in his syntactic description on sentence-types and the determination
of syntactic relations. We can also experience in the Man that while al-Farr dealt
with irb-endings in a somewhat fexible way, he did not make allowances in the case
of syntactic structures.
ID MR in the MA
N of al-farr
63
C) al-Farr s opinion on the linear interpretation of the text
a) At least half of al-Farr s analysis of idmr deals with the interpreta-
tion of diferent irb endings which he considers fundamental in
clarifying the meaning of the Qurnic text . Terefore he is in con-
stant struggle with those views according to which the short vowel
endings are merely phonetic or morphological phenomena and not
irb endings . Te most frequent of these solutions is the choice of
the fnal vowel on the basis of proximity, itb .
16
b) He also introduced some formal descriptive interpretations in which
he might have followed Kfan exeg etical methods. He accepts these
explanations if they ft into his concept of the linear descriptive text
analysis . Te best examples to demonstrate this are those places
where he insists that certain particles , such as wa-or the preposi-
tions , may block the impact of the previous grammatical structure
(e.g. Q 2:185 as analyzed in 3.2.2 above).
c) al-Farr , however, refuses to accept superfcial formal analyses which
run contrary to the meaningful interpretation (e.g. Q 2:48 when he
refuses identifcation of f in temporal adverbs with f in prepositional
verbs ). In such cases he does not say whose opinion, whose analy-
sis is rejected by him. Sometimes he uses the passive (yuql), some
other times it can only be supposed that he is at variance with some
people who only looked upon the text as a linear string of utterances
and derived the explanations quasi mechanically from the preceding
string.
II Te exegetical tradition
Al-Farrs grammatical activities in the feld of Qurnic exegesis cannot
be overestimated. It becomes especially evident if we compare his expla-
nations to the relevant passages in at -T abar s Tafsr. It goes without say-
ing that the scope of at -T abars commentary is much wider than that of
al-Farr s work. But if we limit our comparison to the linguistic exege-
sis , we fnd that al-Farrs explanations are most of the time taken over
verbatim without his name being mentioned. One can also suppose that
16
Al-Farr accepts these variant forms in case of diferent existing sam in the col-
lected corpus, as e.g. in the case of al-h amda/i/u (Q 1:2). If, however, the diference
appears not in the given vowel ending but in the explanation, then he always advocates
the explanation based on iml , as e.g. in Q 1:7 where he does not accept the expla-
nation according to which the ending of ayri would be determined by the preceding
alayhim.
64 kinga dvnyi
at -T abar had access to a fuller version of the Man, it is, however, also
possible that al-Farr was only one prominent exponent of a common
thinking about grammatical issues in the Qurn and the analyses he
dictated from his memory belonged to a common stock of knowledge, a
long line of grammatical exegetical tradition , with which at -T abar was
still familiar. It is interesting to note, however, that the two other Man
works, Ab Ubayda s Majz and al-Axfa s Man do not seem to have
been incorporated into what became the defnitive commentary of the
Qurn for centuries. So it might be concluded that while the grammati-
cal tradition as it was shaped in Bas ra outshone the Kfan , its trace in
the grammatical analysis of the Qurn is not signifcant.
17
5. References
5.1 Primary sources
Ab Ubayda , Majz = Ab Ubayda Mamar b. al-Mut ann at-Taym, Majz al-Qurn.
Muh ammad Fud Sazgn [Fuat Sezgin], ed. 2 vols. Cairo: al-Xnj, n.d.
al-Axfa , Man = Ab l-H asan Sad b. Masada al-Muji al-Axfa al-Awsat , Man
l-Qurn. Fiz Fris, ed . 2 vols. Amman: Dr al-Bar, 1981.
Farr , Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ah mad
Ysuf Najt and Muh ammad Al an-Najjr, eds . 3 vols. Cairo: al-Haya al-Mis riyya
l-mma li-l-Kitb, 1980
2
.
Sbawayhi , Kitb = Ab Bir Amr b. Ut mn Sbawayhi, al-Kitb. Abd as-Salm
Muh ammad Hrn , ed. 5 vols. Beirut: lam al-Kutub, 19661977. n.d.
at -T abar , Tafsr = Ab Jafar Muh ammad b. Jarr at -T abar, Jmi al-bayn f tafsr
al-Qurn. 30 vols. Cairo: al-Mat baa al-Maymaniyya, n.d.
az-Zajjj , Man = Ab Ish q Ibrhm b. as-Sar az-Zajjj, Man l-Qurn wa-irbuhu.
Abdaljall Abduh alab, ed . 5 vols. Beirut: lam al-Kutub, 1988.
17
An interesting feature of al-Farr s explanatory method is his usageat certain
placesof the 2nd person singular in the case of irb endings and other grammatical
interpretations. Te use of 2nd person singular is on the one hand the usual practice
in Sbawayhi s Kitb, but Sbawayhi presents his linguistic analysis from the point of
view of the speaker, the producer of diferent utterances, whereas al-Farr deals with
a concrete text, and what is even more, the text of the revelation. Te use of the 2nd
person singular in this case might indicate that al-Farr considered that the sacred text
was the one without the short vowel endings or that it had been revealed according to
what is termed as sabat ah ruf which leaves the reader of the text some freedom in the
vocalic realization. It should, however, be noted that al-Farr similarly to other com-
mentators, or in fact Sbawayhi when he deals with the Qurn (e.g. Kitb II:155, 10 ad
Q 5:69)does use the 3rd person singular or the passive when he deals with diferent
irb endings in the Qurn .
ID MR in the MA
N of al-farr
65
5.2 Secondary sources
Burton, John . 1988. Te Qurn and the Islamic Practice of wud . BSOAS 51:1:2158.
Dvnyi, Kinga . 198788. Mujwara: A Crack in the Building of irb. Quaderni di
Studi Arabi 56, 196207.
. 1990. On Farrs linguistic methods in his work Man l-Qurn. Studies in the
History of Arabic Grammar II. Proceedings of the 2nd Symposium on the History of Ara-
bic Grammar, Nijmegen, 27 April1 May 1987, Michael G. Carter and Kees Versteegh,
eds . Amsterdam: J. Benjamins.
. 1991. Al-Farr and al-Kis: References to Grammarians and Qurn Readers
in the Man l-Qurn of al-Farr. Te Arabist. Budapest Studies in Arabic 34, 159
176.
Kinberg, Naphtali . 1996. A Lexicon of al-Farrs Terminology in his Qurn Commentary.
Leiden: E.J. Brill.
Makram, Abd al-l Slim and Ah mad Muxtr Umar . 1985. Mujam al-qirt al-
Qurniyya. 8 vols. Kuwait: Jmiat al-Kuwayt.
Talmon, Rafael . 2003. Eighth-century Iraqi Grammar. A Critical Exploration of Pre-
Xallian Arabic Linguistics. Winona Lake, Indiana: Eisenbrauns.
Versteegh, C.H.M. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden:
E.J. Brill.
ARABIC ALLAD AS A CONJUNCTION:
AN OLD PROBLEM AND A NEW APPROACH
Werner Diem
Cologne
Sunt aliquot quoque res, quarum unam dicere causam
non satis est, verum pluris, unde una tamen sit.
Lucretius: De rerum natura VI 703f.
1. Introduction
Tis paper is devoted to the Arabic sub-standard phenomenon of the
relative pronoun allad in the function of a conjunction predominantly
meaning that; because. Te historical interpretation of this phenom-
enon, which had been noted by Arab purists for the formula al-h amdu
li-llhi llad Praise be to God that as early as the 9th century C.E.,
has been a topic for Arabists during the last ffy years. Considering the
intensity with which the historical dimension of allad that; because
has been discussed, it may seem superfuous to want to take it up yet
again. Tis, however, would be a rash conclusion as no communis opinio
concerning the origin of the conjunctional allad has ever been agreed
upon.
Te aim of this paper is threefold. First, I shall discuss the theories
as to the diachronic aspects of allad as a conjunction in chronological
order. Ten I shall present additional early evidence from documen-
tary sources dating from the 11th12th centuries C.E. and later. Afer
that, I shall present my own theoretical approach combining important
insights of my predecessors with hitherto neglected aspects. Finally, I
shall deal with the origin of the formula al-h amdu li-llhi llad .
A point that I shall not consider in this paper is the general function of
the H amdalah in formulaic expressions used by todays Arab Muslims .
Te important role of this formula, simple and expanded, in everyday
Arabic speech is borne out by numerous examples in Piamenta (1979,
247f., Index) and (1983, 209, Index), among them examples of the
68 werner diem
al-h amdu li-llhi llad type with allad as a conjunction meaning that;
because (Piamenta 1979, 89, 172f., 176).
2. Teoretical approaches
2.1 Meir M. Bravman (1953)
In the chapter Te development of the psychological (logical) subject -
predicate relation in his Studies in Arabic and General Syntax (1953),
Meir M. Bravmann , afer stating that in constructions of the al-h amdu
li-llhi llad type and similar expressions allad can be interpreted as
that, maintains that the construction is already found in Classical Ara-
bic , for which he adduces a verse by Imra al-Qays . Tis verse need not
be discussed here as it allows of other interpretations. More important
is Bravmann s general theory about the origin of allad as a conjunction
(1953, 41):
[. . .] we have to assume [. . .] relative clauses which do not link up with
the immediately preceding expression (as al-h amdu li-llhi etc.), but are
parts of an independent new sentence or, more exactly, predicates whose
subject known from the preceding sentence or from the situationis
mentally supplied but not linguistically expressed [. . .]. [. . .] the sentence
al-h amdu li-llhi llad lam amut would mean: Tank God. [I am one]
who has not died etc.
Spitaler (1963) rejected Bravmann s theory (see below 2.3), and it was
not discussed by other scholars. I tend toward Spitaler s rejection of this
theory as it does not explain satisfactorily the phenomenon in question,
and therefore I shall disregard it.
2.2 Joshua Blau (1961)
In his grammar of Judaeo-Arabic , Joshua Blau treats the conjunctional
allad in 346, which begins with the words allad introducing noun
clauses .
1
Te paragraph falls into fve sections according to the syntactic
status of the allad clauses : (a) attributive clauses , (b) subject clauses ,
(c) predicate clauses , (d) object clauses , (e) clauses in which allad has a
causative function . Most of the examples refer to past events, but there
1
Tis and the following translations from the Hebrew of Blau are mine.
arabic allad as a conjunction 69
are also some referring to future events with the head s of the clause s
expressing prevention and necessity.
As to the origin of the conjunctional allad , Blau begins section (a)
with the following words (1961, 226):
Relatively usual are attributive clauses (apposition ), which represent the
transition from relative clauses,
a statement which is continued in a note with the remark (words in
brackets are my additions):
However, this does not mean that allad introducing noun clauses has
originated from attributive clauses only. It is, for example, possible that
from sentences such as al-h amdu li-llhi llad anan Praise be to God
who helped me allad = k, e- [that; because] originated, marking
object clause s or mipne e [because].
Tese are quite general statements, which hint to the direction where,
according to Blau , a historical interpretation should be looked for. In his
opinion, as can be grasped from both remarks cited above, the origin of
allad as a conjunction has to be seen (a) in attributive relative clauses
and (b) in the al-h amdu li-llhi llad type.
2.3 Anton Spitaler (1963)
Spitaler s approach difers fundamentally from Bravmann s and Blau s
as it is monocausal . He starts from the al-h amdu li-llhi llad type and
assumes that this construction was frst re-interpreted (Spitaler does not
use this word) as meaning gottlob, da, afer which it was generalized.
Te following quotations will illustrate Spitaler s approach:
al-h amdu li-llhi llad ist der Anfang eines normalen, von dem Wort Allh
abhngigen Relativsatzes , aber aus seinem eigentlichen Zusammenhang
gelst und sekundr in eine neue Konstruktion bertragen. (1963, 101)
Es ist nun ganz eindeutig, dass das stereotyp wiederkehrende al-h amdu
li-llhi im Lauf der Zeit einer Funktionsschwchung unterlegen ist [. . .].
Dadurch verlor das nachfolgende allad zwangslufig den lebendigen
Zusammenhang mit seinem Beziehungsnomen und der ganze Ausdruck
wurde zu einem starren, wenn auch mit einem ganz bestimmten Affekt-
gehalt geladenen Syntagma, bei dem allad nur mehr als berleitendes
Element, als verbindende Partikel , letztenendes eben als Konjunktion
empfunden wurde. Und nunmehr war es natrlich gleichgltig, welche
syntaktische Form der anschlieende Satz hatte. (1963, 235)
70 werner diem
Afer presenting some examples semantically corresponding to al-h amdu
li-llhi llad praise be to God that which, however, contain fnite forms
of h md I and kr I instead of the nominal al-h amdu li- (1963, 105), and
afer rejecting Bravmann s above-mentioned theory (2.2), Spitaler con-
tinues with the words:
Mit der Herauslsung aus seinem ursprnglichen legitimen Relativsatz-
gefge und der bertragung in eine ganz neue Konstruktion hat nun aber
unser Typus seine Entwicklung nicht abgeschlossen; es zeigt sich nmlich,
dass die Formel al-h amdu li-llhi als solche zurcktritt und durch andere
Ausdrcke ersetzt wird, die ebenfalls der usserung des Dankes, der
Freude, frohen berraschung, Befriedigung usw. dienen. [. . .] Die dabei
gegebenen phraseologischen Mglichkeiten sind sehr mannigfaltig [. . .].
Sie werden aber noch dadurch bereichert, dass die Konstruktion nunmehr
[. . .] auch fr die damit kontrastierenden Gefhle des Bedauerns, der Reue
usw. verwendet wird. (1963, 109)
Tere follow in Spitaler s article numerous examples of these latter con-
structions both from post-classical writings and modern dialects , all of
them referring to events in the past. Additionally, Spitaler presents some
examples of allad that afer verbs from other semantic felds , among
them some where the allad -clause , as in some of Blau s examples, refers
to future events expressed by the Arabic imperfect .
While Spitaler s approach to the problem is defnitely a monocausal
one, he was well aware of the fact that a multicausal approach would
also in principle have been possible, all the more in view of Hebr .
a
er
(on its origin see now Rubin 2005, 49f.) and Aram . d-, which, besides
their original function as relative particles , developed the meaning
that; because (1963, 106f.). In his fnal remarks, Spitaler emphatically
defended his own theory on the grounds: (a) that the Arab grammarians
noted the conjunctional allad only for the H amdalah , which therefore
seems to be central, (b) that the al-h amdu li-llhi llad type possesses the
chronological priority within the evidence of the conjunctional allad ,
and (c) that it would be implausible to assume two diferent starting
points of the conjunctional allad converging into one phenomenon
(1963, 111f.).
Tere is no reference to Blau s Diqduq (1961) in Spitaler s article,
probably due to the fact that he was unaware of it when he wrote the
article. In fact, Blau in some respect anticipated Spitaler s theory about
the central role of al-h amdu li-llhi llad .
arabic allad as a conjunction 71
2.4 Joshua Blau (1965)
In his Emergence and Linguistic Background of Judaeo-Arabic (1965),
which appeared four years afer Diqduq, Blau is essentially of the same
opinion as in Diqduq (1961) about the origin of allad that, and even
the wording reminds one of that used in Diqduq although it is more
explicit:
Since a relative clause may ofen be mistaken for a substantive clause , the
transition of allad into a conjunction introducing substantive clauses was
easily accomplished. (1965, 109)
Afer this general statement, Blau gives two examples: wa-kam alimta
min h adt i bni Dwrdy jama m faalahu llad abida s -s anama and li-
llhi l-h amdu llad aslamahu f aydkum, in Blau s translation and like-
wise, you know the story of B. D., everything that he had done, that he
had served the idols and praise be to God that he has delivered him
into your hands. According to Blau , it is in both cases only the word
order that indicates that allad introduces a substantive clause and not
a relative clause , i.e. that allad is a conjunction and not a relative pro-
noun . While the second example indeed illustrates this transition, as
allad can be interpreted both as a relative particle and a conjunction
(praise be to God, who delivered him into your hands and praise be to
God that He delivered him into your hands), this, in my opinion, is not
possible with the frst example as it admits of the interpretation of allad
as a conjunction only. From a semantic point of view, the allad clause in
this sentence is an apposition to m faalahu, which means that it cannot
be interpreted at the same time as an attribute to Ibn Dwrdy. Tus, this
example does not represent the starting point of the re-interpretation of
allad but allad in its new conjunctional function .
Spitaler s theory is not taken into consideration by Blau , but his article
is mentioned in a footnote (1965, 109).
2.5 Joshua Blau (1967)
In contrast with Emergence (1965), where Spitaler s article (1963) is
mentioned in a footnote only, having no impact on Blau s theoretical
approach, it is fully considered in the third volume of Blau s monumen-
tal Grammar of Christian Arabic , which appeared in 1967. Blau writes
there:
72 werner diem
alladh opens substantive clauses . Beginnings of this construction already
appear in CA
2
and it becomes rather frequent in MA . A. Spitaler [. . .]
has brought out in full relief the early and very interesting history of the
type al-h amdu li-llhi-lladh thank God that and its development. Te
use of alladh for introducing substantive clauses in ASP , insofar this is
not due to literal translation of Aramaic d-, is mainly from this type and
its developments. Tis is all the more remarkable since in JA other sub-
stantive clauses (especially apposition clauses ) introduced by alladh are
quite common. It stands to reason that, in accord with Spitaler s fndings,
this use of alladh frst developed in the type al-h amdu li-llhi. Early ASP
exhibits mainly this stage, whereas in later JA other kinds of alladh intro-
ducing substantive clauses have arisen. (1967, 526f.)
Among Blau s examples there is none of the nominal al-h amdu li-llhi
llad type. As to his examples of the corresponding verbal type that con-
tains a reference to God, such as h amida llhi llad or akara llhi llad ,
it must be stated that none of them unequivocally represent the new
type with allad as a conjunction because each can also be interpreted
as a relative clause referring to Allh . Of Blau s remaining examples,
many are likewise syntactically ambiguous, while some do represent the
new type. In several cases, Blau himself draws attention to the alterna-
tive interpretation as relative clause s. Apart from that, as Blau remarks
himself, some examples might be calques on the Aramaic d-, which, as
has been mentioned above, is both a relative particle and a conjunction
meaning that; because. Tus unequivocal evidence of the new type is
less for Ancient South Palestinian than might be inferred from the num-
ber of Blau s examples at frst glance.
2.6 Manfred Woidich (1980)
Manfred Woidich s article illi als Konjunktion im Kairenischen (1980)
is an investigation into illi , which goes back to allad , as a conjunction in
modern Cairene Arabic . In his masterly analysis, which shows an inter-
est and a competence in syntax not common for Arabic dialectology ,
Woidich gives a thorough picture of illi as a conjunction from a syn-
chronic perspective. Woidich s examples are mostly drawn from written
sources, but he also asked native speakers of what is called Educated
Cairene Arabic . It would be beyond the scope of this paper to comment
2
CA = Classical Arabic , MA = Middle Arabic , ASP = Ancient South Palestinian , JA
= Judaeo-Arabic .
arabic allad as a conjunction 73
extensively on what Woidich found out about illi as a conjunction in
Cairene Arabic . Terefore I shall concentrate on some points that are
also important from a historical perspective.
Woidich begins with the statement that the A (bergeordnete Aus-
drcke ), i.e. the head s on which the illi clause s as uS (untergeordnete
Stze ) depend, are:
uerungen des Dankes, der Freude, der Befriedigung, des Bedauerns usw.
[. . .] Der Sprecher nimmt jeweils emotional Stellung zu einem im unter-
geordneten Satz (uS) geschilderten Sachverhalt und wertet ihn in diesem
Sinne. [. . .] Wir teilen das gesammelte Material nach der Art der AA ein,
von denen sich zwei Typen unterscheiden lassen. Die einen (Gruppe A)
werten den Sachverhalt des uS direkt, indem sie [. . .] emotional Stellung
nehmen, die anderen (Gruppe B) werten auf eine indirekte Weise, indem
sie angeben, wie eine Person sich oder andere wertet aufgrund des Sach-
verhalts des uS, oder zu welcher affektisch bestimmten Handlung eine
Person durch diesen veranlat wird. (1980, 225f.)
Examples of type A are il-h amdu li-llh illi ultaha b-nafsak Gott sei
Dank, da du es selbst gesagt hast! and kuwayyis illi wiit f gbak inta
Gut, da es in deine Tasche gefallen ist! Characteristic of this type is
according to Woidich the absence of resumptive pronouns :
Eine Verknpfung des A mit dem uS durch pronominale Rck- und Ver-
weise kann hier nicht stattfnden, da im A keine Pronomen aufreten.
(1980, 226).
Tough this statement is corroborated by the above-cited examples, the
last two examples of Woidich s type A do have a pronominal concatena-
tion , among them ayiz ni lli dayyat il-bazburt-i bti Es rgert mich,
da ich meinen Pa verloren habe. Woidich considers examples of this
type einen bergang zur Gruppe B (1980, 227). In my opinion, such
examples do not belong to group A but rather to group B, as expressions
like ayiz + object pronoun are a kind of pseudo-verbs , the pronoun
being the logical subject , and there are similar examples in group B.
As for type B, examples are d-ana batnaddim illi gt Ich bereue, da
ich gekommen bin and inta karihni lli bal il-h a Du kannst mich
nicht leiden, weil ich die Wahrheit sage. Concerning the syntactic char-
acteristics of this type B, Woidich says:
Logisches Subjekt des A ist [. . .] eine Person, die im uS als Subjekt auf-
tritt. Dadurch kommt eine Verknpfung der beiden Teilstze durch pro-
nominalen Verweis zustande, der ein Charakteristikum fr diese Gruppe
darstellt. (1980, 227)
74 werner diem
Contrary to this statement, the person in the A (i.e., the head ) to whom
the illi clause refers is not always the logical subject , as is, e.g., borne
out by the above-mentioned example inta karihni lli bal il-h a, where
the Person, die im uS als Subjekt aufritt is the (logical) object , not
the (logical) subject of the sentence. Furthermore, there are examples of
type B declared by Woidich to be marginal where there is no pronomi-
nal concatenation between the illi clause and an element of the head at
all, e.g. ana mabst illi ma-g I am glad he did not come (1980, 229).
Considering this evidence, it seems to me that the feature of coreferenti-
ality is questionable for Woidich s type B in Cairene Arabic , and it is evi-
dent from Blau s, Spitaler s and my evidence that it cannot be sustained
in a general diachronic and diatopic perspective. A certain degree of
coreferentiality is per se bound to exist for the simple reason that when
in the head s of sentences emotions are spoken of the causes of these
emotions as expressed in substantival clauses depending on those head s
are in most cases connected with the persons mentioned in the head s of
the sentences as having these emotions.
Important insights in Woidich s article are that in type B the element
of the head to which the illi clause refers must be human (1980, 230),
that the predicates are faktiv, that is, refer to real facts, mostly past
events (1980, 231), and that this kind of illi can always be replaced by
inn - that, with the diference that illi is considered by the informants
as being strker than inn -, that is, more afective (1980, 234).
2.7 Manfred Woidich (1989)
While Woidich s article of 1980 is essentially descriptive, his contribu-
tion illi dass, illi weil und zayy illi als ob: zur Reinterpretation von
Relativsatzgefgen im Kairenischen, which was published in 1989,
is diachronically oriented. As announced in the title, Woidich distin-
guishes between three diferent kinds of illi: illi that , illi because and
zayy illi as if .
(a) illi that
Te type illi that corresponds to type A of Woidich (1980), e.g., il-h amdu
li-llh illi sabitak gottlob, da sie dich verlassen hat and kuwayyis illi
gt gut, da du gekommen bist (1989, 110f.). As for the origin of this
type, Woidich follows Spitaler (1963) in saying that il-h amdu li-llh in
arabic allad as a conjunction 75
sentences like il-h amdu li-llh illi waaik fyya Lob sei Gott, der dich
mit mir zusammengebracht hat:
unterlag einer Funktionsschwche [. . .]. Als starres Syntagma wurde
diese Formel nicht mehr als analysierbar aufgefat, was zur Folge hatte,
da insbesondere N = allh nicht mehr als Nomen gesehen wurde, dem
ein syndetischer Relativsatz angeschlossen werden konnte. (1989, 111)
While in the assumption of Funktionsschwche Woidich explicitly fol-
lows Spitaler , he difers from him in assuming that the al-h amdu li-llhi
llad formula was re-interpreted because there are similar sentences in
Cairene Arabic having the same marked structure of Rhema-Tema,
that is, comment-topic, such as h ilwa di prima ist die! or b ikkalm
da eine Schande sind solche Worte! He sums this up by saying
da die Reinterpretation von illi als Relativpronomen zu illi da ausgelst
wurde durch den Umstand, da die ursprngliche syntaktische Struktur
mit dem eingebetteten Relativsatz nicht der thematisch-kommunikativen
Funktion dieser Stze entsprach. Sie war markiert und wurde durch Rein-
terpretation der bei diesen Stzen blichen funktionalen Satzstruktur
Prd.Subjekt angeglichen. Dadurch konnte dem illi die Funktion einer
Konjunktion zugeordnet werden, die Subjektsstze einleitet. (1989, 115)
Here we have to ask what Woidich intends to demonstrate from a
general Arabic perspective. Does he want to demonstrate a historical
development valid for Cairene Arabic only, or a general development
in Neo-Arabic the results of which are palpable in Cairene Arabic also?
A reconstruction of the frst kind would be fawed by the fact that the
Cairene conjunctional allad cannot be detached from the common
history of Neo-Arabic , and a reconstruction of the second kind by the
fact that it would be problematic to reconstruct a common Neo-Arabic
development on the basis of the specifc evidence of one modern dialect ,
while disregarding other evidence, older and newer. Since we are con-
cerned here with the second perspective only, which is tantamount to a
general reconstruction of the conjunction allad in Neo-Arabic , we have
to ask what Woidich s theory means for the history of the conjunctional
allad in early Neo-Arabic .
In Neo-Arabic the equivalents of Woidich s above-mentioned two
Cairene examples would probably be something like h ilwa hd and ayb
hd l-kalm. Comparing these sentences with a sentence like al-h amdu
li-llhi llad jita, it seems very improbable that the two types of sen-
tences should have been mentally connected by any speaker of Arabic ,
76 werner diem
and thus Woidich s theory seems to me to be farfetched. Nevertheless,
his drawing attention to the Rhema-Tema (or comment-topic ) struc-
ture, which syntactically corresponds to a predicate-subject structure, is
an important insight, to which I shall come back below when discussing
my own theory (4.3).
Goldenberg comments on Woidich s theory with the words Tis form
is structurally parallel to the classical Arabic constructions of clef sen-
tences with m having the same order of constituents (1994, 16/261),
but I doubt the validity of this alleged parallel, as the clef sentences
mentioned by Goldenberg , apart from their Rhema-Tema structure,
difer fundamentally from the sentences mentioned by Woidich .
(b) illi because
illi because corresponds to type B of Woidich (1980), and examples
are ana farh n illi ufak I am happy because (that) I saw you (1989,
116) or itnaddimit illi gat maya sie bereute es, da sie mit mir gekom-
men war (1989, 117). As for the meaning of illi in the second example,
it remains to be shown that in Cairene Arabic illi in such sentences can
be replaced by the causal conjunctions an and li-ann -. Interestingly,
no transformations of sentences containing expressions of this type are
among those adduced by Woidich (1980, 235) for the replacement of illi
by the causal conjunctions an and li-ann -.
In view of the semantic diference between the two kinds of illi in
Cairene Arabic as seen by him and also in view of the feature Refe-
renzidentitt (that is, coreferentiality) in the second type, which he
thinks is essential, Woidich assumes for the second type (type B in
Woidich 1980) an origin independent of the al-h amdu li-llhi type (type
A in Woidich 1980). Tis origin he fnds in sentences such as ana h mr
illi dafat il-h isb ich bin ein Esel, der ich die Rechnung bezahlt habe!,
which contains a direct relative clause depending on ana. Since in sen-
tences of this kind there exists a causal connection between the relative
clause and the head , Woidich assumes that illi could be re-interpreted
as a causal conjunction , ich bin ein Esel, da ich die Rechnung bezahlt
habe!, adding that this re-interpretation is also true of alle anderen
Stze dieser Struktur (1989, 118). It goes without saying that if we are
to assume this kind of re-interpretation for early Neo-Arabic in general
(and not only for Cairene Arabic ), we would have to assume that allad
had been generalized as a relative particle by then, an assumption which
poses no problem.
arabic allad as a conjunction 77
Te fact that Woidich declares his example (13) itnaddimit illi gat
maya sie bereute es, da sie mit mir gekommen war not to belong
to the original type, confrms that, in his opinion, it is in nominal sen-
tences that this type originated (1989, 119). Against this it can be argued
(a) that verbal sentences (such as itnaddamit illi) seem to prevail over
nominal sentences (such as ana h mr illi), and (b) that it seems highly
improbable to me that speakers should have used sentences like ana
h imruni llad addaytu d-dannra I am an ass, (I) who have paid the
dinars, an farh nuni llad najawtu I am glad, (I) who have escaped
or an mutaassifuni llad taaxxartu I am sorry, (I) who was late at all.
For this theory to be accepted, it would be prerequisite to fnd unam-
biguous relative clause s of this kind in Classical Arabic . I have checked
more than a thousand items with allad in Tradition , many of them in
dialogues, without fnding even one example of such constructions, and
I dare say that it is very improbable that they occurred at all.
To sum up, as long as the syntactic type ana h imr illi dafat il-h isb,
which Woidich presumes to be the starting point of this type, is not
shown to have been a normal construction in pre-Neo-Arabic , I con-
sider sentences of this kind in Neo-Arabic the result of a specifc devel-
opment, which should be explained otherwise, rather than the origin
thereof.
(c) zayy illi as if
In interpreting this third type, Woidich starts with cases such as ir-rgil
frih bna zayy illi laa lya der Mann freute sich ber uns, wie einer, der
einen Fund gemacht hat, in which the syntagma zayy illi wie jemand,
der; like somebody who could be interpreted as meaning als ob; as if .
In cases representing unequivocally the result of this re-interpretation,
only the interpretation of zayy illi as as if is possible, such as ma-kan-i
f tagwub, zayy illi kunt-i bakallim h agar es gab keine Resonanz, es war,
als ob ich mit einem Stein sprche.
As for earlier evidence of this type, the ka-llad mentioned in gram-
mars , which structurally corresponds to zayy illi , has a diferent func-
tion, namely that of a conjunction of comparison with real facts, such as
xudtum ka-llad xd You have plunged as they have plunged Qurn
9:69, already adduced by Reckendorf (1921, 192, 4). For more evi-
dence of this ka-llad , see Blau (1967, 527f.), Hopkins (1984, 238) and
Goldenberg (1994, 27f./276).
78 werner diem
An example of ka-llad that functionally corresponds to zayy illi as a
particle of comparison with unreal or hypothetical facts, as described by
Woidich , was adduced by Spitaler (1962, 108) in a footnote: fa-l yazlu
kad lika ka-llad yuh ibbu an yasxara min s h ibih
3
and he continues to
behave in this manner like one (or as if) intending to mock his friend,
sc. the tomcat (as-sinnawr) who plays with the mouse (al-farah) afer
having caught her (al-Jh iz , H ayawn V, 202). Spitaler s second example
concerns the well-known other ka-llad type: fa-s allaw ka-llad kn
yaf alna and they prayed as they always had. Another example of
ka-llad as a particle of comparison with hypothetical facts is wa-man
axad ahu bi-irf nafsin lam yubrak lahu fhi ka-llad yakulu wa-l
yabau and who takes it (sc. certain money) with haughtiness of mind
will not be blessed regarding it, like one (or as if) eating without getting
replete (al-Buxr , Sah h , Kitb az-zakh, N
o
1379), and more evidence
of this kind can be found in Tradition . Since in examples of this kind
allad can still be interpreted as a relative pronoun , they represent ka-
llad in the meaning as if in statu nascendi. Obviously, Spitaler did not
realize that his two examples belong to two fundamentally diferent types.
Te frst ka-llad is a conjunction corresponding to kam (accord-
ingly it should perhaps be transcribed as kallad ), while the second
ka-llad consists of the particle ka- and the relative pronoun of the 3rd
masc. sing. In Cairene Arabic , this second type was grammaticalized as
zayy illi , with zayy standing for ka -, which is unusual in Cairene Arabic .
Future research should pay attention to the existence of two diferent
types of ka-llad .
It is, then, evident that ka-llad has to be separated from allad that,
which is the topic of this paper. Terefore, these remarks must sufce,
and I shall not return to this special type.
2.8 Gideon Goldenberg (1994)
In his article allad al-mas dariyyah in Arab grammatical tradition
(1994), which was reprinted in 1998, Gideon Goldenberg begins with
general observations, then mentions Spitaler s article (1962) and Woid-
ich s second article (1989). Goldenberg describes the aim of his paper
as follows:
3
Tere follow four further an-clauses dependent on ka-llad yuh ibbu. Spitaler s quo-
tation has the variant bi-s h ibih.
arabic allad as a conjunction 79
About the distribution of infnitival allad in Arabic writings in general
I have nothing to add; in the present paper I just wish to adduce some
further evidence for a fuller understanding of the relevant structures as
described and treated in Arabic grammatical literature . Such examination
is important, because syntactical constructions that are considered gram-
matical by the great masters of Arabic grammar cannot easily be discarded
as inadmissible or non-Classical . (1994, 16f., 262)
Some of the examples that the Arab grammarians commented on are
from the Qurn , while other examples were made up by the grammar-
ians themselves. A survey of the examples in the Concluding remarks
shows that the examples belong to disparate types. In some of them,
allad is combined with a preposition , such as ka-llad as , al llad for
that , bada llad afer, in which allad stands for the more usual m. In
other cases, it is again either m , or an , which allad stands for. It might
even be doubted that some of the sentences on which the grammarians
dwelt in length ever occurred in normal speech, such as allad mararta
mamarrun h asanun It is a good passing that you passed, apart from
the fact that the syntactic interpretation of allad in this sentence is far
from being unequivocal.
As far as I can see, Goldenberg makes no attempt to draw historical
conclusions from his examples within the general discourse of conjunc-
tional allad , nor can I do this myself. So, in spite of Goldenberg s thor-
ough and learned approach, the historical dimension of his examples
has yet to be demonstrated.
3. Additional early documentary evidence
In the following, I shall give documentary evidence of allad as a con-
junction , which is the result of a wide and systematic reading of pre-
modern Arabic documents. Te major part of the evidence comes
from Judaeo-Arabic documents found in the Genizah in al-Fust t (Old
Cairo ). Considering that I likewise have read Arabic documents written
by Muslims as well as Christians , the preponderance of the Judaeo-Ara-
bic evidence is noteworthy, being due to the more sub-standard charac-
ter of Judaeo-Arabic writings as compared with Muslim and Christian
ones. Te value of documentary examples lies in that they are chrono-
logically well-defned, which is not necessarily the case with literary
examples, especially when these occur in sub-standard texts which have
come down to us through late manuscripts.
80 werner diem
Meanwhile, many of the Judaeo-Arabic documents which I had read
in their editiones principes have been re-edited by Moshe Gil in his two
comprehensive collections of Genizah documents. Where this is the case
I shall cite both editions, partially in order to give credit to the original
editors and partially because there are occasionally diferences between
Gil s re-editions and the original versions. While some interesting or
difcult cases of dialectal divergences from the literary language will be
commented upon in footnotes, substandard elements as such are not
systematically marked. Some of my examples have already been cited by
Blau (1961), as became clear to me while preparing this article. I have
retained these examples for the reason that they belong to an exclusively
documentary corpus, whereas Blau draws his examples from both liter-
ary and documentary sources, with the majority being literary.
My examples will be presented according to a linguistic typology
which mirrors the historical development of allad as I see it. My divi-
sion owes much to Woidich s division of his Cairene material into his
types A and B (see above 2.6 and 2.7), even if his division and mine do
not fully overlap from a historical perspective as I assume more sub-
types to have existed than he does. Many examples contain vulgarisms,
which I shall not mark systematically.
A. Te head of the sentence expresses praise of or gratitude to God
(a) Nominal type
(1) li-llhi l-h amdu llad knati l-qibatu li-xayrin
4
Praise be to God that the result (of the afair) was good
Goitein , Arkiyon, N
o
39r, 21 = Gil , Texts, N
o
215r, 20 (letter
to al-Fust t , early 11th c. C.E.)
(2) fa-li-llhi l-h amdu wa-l-minnatu llat
5
knati l-aqbatu
h amdatan
So to God be praise and gratitude that the result (of the
afair) was praise worthy
6
4
Goitein reads [il l-]xayri, Gil li-xayrin. Gil does not mention Goitein s reading.
5
allat, which seemingly refers to al-minnah, is due to a kind of hypercorrection .
6
Another example with allat that, or, more exactly, li-llat in order that, is li-
anna qawm mina l-maribati qad s r yus all indan li-llat (!7%%) yut bahum in
Goitein , Kneset, N
o
4v, 4f. Indeed, Goitein rendered !7%% as the Hebrew fnal conjunc-
tion *.)%. However, Gil s re-edition in Texts, N
o
328, has %% instead of !7%%. So what
we have here is the simple prayer Allhu yut bhum May God reward them!sc. the
Maghrebis for praying in the synagogue of the Jerusalemites in Old Cairo .
arabic allad as a conjunction 81
Diem , Geschfsbriefe Wien, N
o
45r, 10 (letter to a merchant
in Egypt , 12th13th c.s C.E.)
(b) Verbal type
(3) fa-h amidn llha al d lika llad naz ara ilayn wa-lam
yumit bin
and we praised God for this (happy outcome),
7
namely that
He cared for us and did not have us mocked at
8
Goitein , Iggeret r, 16 = Gil , Texts, N
o
616r, 17 (letter of
al-Mahdyah to al-Fust t , 11th c. C.E.)
(4) wa-h amidtu llha]
9
an wa-ax llad masayn
10
lah iqn al
xayr wa-fyah li-Ms r
11
and we praised God], I and my brother, that we (fnally)
reached Egypt on foot well and in good health
Toledano, Teudot, N
o
2, 1 (letter from Egypt , 1540 C.E.)
B. Te head of the sentence does not express praise of or gratitude to God
(a) Te head of the sentence contains an abstract noun expressing a
non-personal emotional evaluation of the contents of the allad
clause
(5) wa-m wajadn azan li-qulbin ayra annahu l-waylu
lan nah nu llad in li-hd ihi l-mas ibi wa-narab hd ihi
l-akwsa l-murrata
We did not fnd consolation for our hearts but (all we can
say is) that woe is us that we have lived (long enough to go
through) these disasters und (that we) have to drink these
bitter cups!
Gottheil and Worrell , Fragments, N
o
27r, 79 = Gil , Texts, N
o
501r, 79 (letter from Jerusalem to al-Fust t , 1065 C.E.)
7
Te details of the happy outcome of the afair in question are imparted immedi-
ately before the passage cited.
8
Gil in his translation considers d lika an antecedens of allad in the sense of the
German darber, da. However, the allad clause is an apposition to d lika, which in
its turn refers to details mentioned before. In other words, the sentence is an expansion
of the usual fa-h amidn llha al d lika.
9
Te addition is mine. Other additions are possible, but praise of God is the most
probable one.
10
Dialectal form for maayn.Te language of the letter is substandard and exhib-
its features of Moroccan Jewish Arabic .
11
Te grapheme renders Mas ar < Mas r < Mis r.
82 werner diem
(b) Te head of the sentence contains a verb or a participle express-
ing the emotion of a person caused by the contents of the allad
clause
Joy
(6) wa-an bi-h amdi llhi mutabit un bi-llad ttas altu ilayhim
I am, thank God, rejoicing that I joined them
12
(in
marriage)
13
Ashtor , Documentos, N
o
2r, 11 = Gil , Documents, N
o
457r,
12 (letter from Jerusalem to Toledo , 1057 C.E.)
14
(7) wa-qad sarran llad anfad ta lahu rah lahu
It pleased me that you sent him his merchandise
Gottheil and Worrell , Fragments, N
o
9v, margin, 1113 (let-
ter from al-Fust t to Aden , probably 12th c. C.E.)
(8) fa-qad radtu laka bi-llad uxid tum fa-huli nsun
muh taimn wa-llhi z-zabbln m rad bi-llad uxid tum
I was content concerning you that you (both)
15
were pun-
ished <as well as certain other persons were content about
it> for these are decent people. Only the street-sweepers
16
were, by God, not content that you (two) were punished
Diem , Geschfsbriefe Wien, N
o
10, 8 (letter of a jealous wife
to her husband, 12th c. C.E.)
Wonder
(9) wa-t umma inn ajabu minka llad lam tus b man yaktubu
laka kitb ill daf atan
Furthermore, I am astonished at you that you (allegedly)
found only once somebody writing a letter down for you
17
Goitein , Saloniqi, N
o
1r, 37f. (letter from Saloniki to al-
Fust t , 11th c. C.E.)
12
Sc. the male relatives of the bride.
13
Ashtor conceives of allad as standing for allad na, translating Estoy contento de
aquellos con que [he emparentado por matrimonio].
14
According to Ashtor , the date is 1053. Te diference consists in the reading of the
last letter in the date according to the Jewish era on the margin (ttyd vs. ttyh )
15
Sc. the addressee and his mistress, the plural standing for the dual in vulgar
language.
16
Tat is, the mob.
17
Sc. a letter to be sent to the writer of the letter, who is the addressees father.
arabic allad as a conjunction 83
Anger
(10) wa-an adbnu alayka y-ax kat r allad was al[ta il]
Mis ra wa-lam tas il il Adana
I am very angry about you, O brother, that you came to
Egypt (or Cairo ), whereas you did not come to Aden
18
Braslavski , Mish ar, r 12 (Letter from Aden to al-Mahdyah ,
c. 1149 C.E.)
Reproach
19
(11) m baqiya alayn ayun ill llad lam tuarrifn kayfa
knat was yatu xlika naxud uhu (!) llad lam tuarrifn in
kna was ala laka ayun mina l-kutubi
Nothing remains for us (to say) except that (until now)
you have not informed us as to how your uncles last will
was. What we reproach him (!)
20
with is that you did not
inform me as to whether you received any of our letters
Gottheil and Worrell , Fragments, N
o
9v, 40f. (letter from
al-Fust t to Aden , probably 12th c. C.E.)
Grief
(12) wa-azza alayn d lika kat r allad lam yakn ah adun
minn indaka yuwinuka f-m jar alayka f taabika f
m yaxus s u amra xlika
and we were very much grieved by that,
21
(namely) that
none of us was with you to assist you in that which you
had to endure in your concerns regarding the afair of (the
illness and death of) your uncle
Gottheil and Worrell , Fragments, N
o
9r, 12f. (letter from
al-Fust t to Aden , probably 12th c. C.E.)
(13) wa-dqa s adrun kat r allad lam yakn laka maahu
kitbun yut amminun
and we were very much distressed that he had no letter
of yours with him setting our minds at rest (with regard
to you)
18
Te writer intends to say that continuing the travel from Egypt to Aden would have
been easy for the addressee, his brother.
19
Only the second allad in (11) is an example of allad following a verb expressing
reproach. For the frst allad , see (17).
20
Scribal error for you.
21
Reference to the illness and death of the addressees uncle, the details of which are
recapitulated by the writer before.
84 werner diem
Gottheil and Worrell , Fragments, N
o
9r, 28f. (letter from
al-Fust t to Aden , probably 12th c. C.E.)
(14) nulimukum ann bi-xayrin f fyatin wa-qad azza alayya
kat r allad lam takn h dir h att tuxallis a laka rah laka
I inform you (herewith) that I am well (and) in good health,
but (that) I am much grieved that you are not present so that
you might free your luggage (from the authorities)
Amari , Diplomi, 53, 2f. (letter from Tunis to Pisa , 12th
13th c.s. C.E.)
(c) Te head of the sentence contains a verb or noun not expressing
an emotion
(15) wa-qad akartu tafaddulahumh arasahum llhullad
qad d akarn f kitbihim bi-s-salmi wa-bi-fli l-jamli f
bb
I am also grateful for their
22
kindnessmay God protect
them(consisting in) that in their letter they gave greet-
ings to me and (also consisting) in performing good deeds
to me
Assaf , Meqorot, 51, line 20f. = Gil , Documents, N
o
298r,
20f. (letter from Jerusalem to al-Qayrawn , 1039 C.E.)
(16) wa-jb lan kitbaka wa-nah nu laysa indan xabarun
h att jbahu r-rajulu l-warrqujazhu llhu ann xayr
allad nalamu
and he brought us your letter while we (still) had no news
of you until that man, the book-seller, brought itmay
God requite him in our stead with good that we (now)
know (how you are)
Gottheil and Worrell , Fragments, N
o
9v, 368 (letter from
al-Fust t to Aden , probably 12th c. C.E.)
(17) See (11), frst allad .
I am the frst to admit that 17 examples of allad that are a somewhat
meagre result of a decade-long reading of texts, but this result simply
shows how rare this phenomenon is in texts whose writers intended lit-
erary Arabic . Surprisingly scarce is especially evidence of the type A(a)
al-h amdu li-llhi llad .
22
Reference to two addressees, whose letter the writer answers.
arabic allad as a conjunction 85
Asking whether we can conclude from the scarcity of this type in doc-
uments, especially letters, that it was likewise marginal in the spoken
language, we must allow for some reservations. Te H amdalah occurs
frequently in letters, mostly at the beginning but also in other parts.
Usually the H amdalah is mentioned in the context of news which are
deemed praiseworthy, as, for example, the writers or other persons
good health or a good outcome of a difcult situation. However, this
kind of the H amdalah is usually preceded by the report of the fact to
which it refers, as, e.g., katabtu ilayka . . . wa-an wa-man qibal bi-
xayrin wa-fyatin wa-l-h amdu li-llhi (al d lika) I am writing you . . .,
while I and my family are in good healthPraise be to God (for this).
Tis conventional structure prevents the al-h amdu li-llhi llad type
from frequently occurring in letters, which in its turn must also lead
to lower frequency of the re-interpreted al-h amdu li-llhi llad in let-
ters than may have been the case in the spoken language. Documentary
texts where the al-h amdu li-llhi llad type usually does occur are, e.g.,
waqf documents, appointments of high ofcials and marriage contracts
between persons belonging to the upper class. In these kinds of texts, an
introductory al-h amdu li-llhi llad is expanded into long complicated
passages praising God for bounties related to the topic of the text, but
they are of so elaborated a style that they lack any deviation from the
literary language . To sum up, the rare occurrence of the re-interpreted
al-h amdu li-llhi llad in documents is of no relevance for the question
of its potential frequency in the spoken substandard language.
Scarce is also my evidence of the type B(a), where an abstract noun
expresses a non-personal evaluation of the contents of the allad clause ,
especially when compared with the abundant evidence of this type in
Cairene Arabic as collected by Woidich (1980 and 1989). Again it is pos-
sible that the scarcity of early evidence of this type in documents is due
to specifc circumstances. Expressions of this type are mostly exclama-
tory and thus have a clear Kundgabefunktion, which makes it more
likely for them to be used in the spoken language than in writing.
4. Historical typology
4.1 Introductory remark
Since the question of whether, and if so, how the diferent subtypes of
allad that are connected with each other is still open at this stage
86 werner diem
of this investigation (even if the answer to this question is admittedly
anticipated by the arrangement of my own material in the preceding
paragraph), I shall deal with them one by one. In doing so I shall con-
sider whether there is reason to assume that the subtype in question
had an origin of its own independent of the other subtypes or whatever
other origin there might be. Te sequence of the subtypes corresponds
to decreasing syntactic and semantic markedness and specifcity.
4.2 Types A(a) al-h amdu li-llhi llad Praise be to God that and type
A(b) h amidtu llha llad I praised God that
Tere seems to be nobody so far denying the validity of Spitaler s theory
that in al-h amdu li-llhi llad , to which my type A(a) corresponds, the
relative particle came to be re-interpreted as a conjunction meaning
da; that. In Spitaler s view, this development is due to a Funktions-
schwchung of allad , whereby das nachfolgende allad zwangslufg
den lebendigen Zusammenhang mit seinem Beziehungsnomen [ver-
lor] (for details see above 2.3), and this opinion is whole-heartedly
shared by Woidich (see above 2.7).
In my opinion, this approach is fawed by the fact that, besides the
nominal type al-h amdu li-llhi llad , attested in the Qurn and else-
where, there existed a corresponding verbal type , as e.g., in the saying
ascribed to Muh ammad , y-maara l-arabi h mad llha llad rafaa
ankumu l-ura O company of the Arabs , praise God, who took the
tithes from you! Ibn H anbal , Musnad, Musnadu l-aarati l-mubairna
bi-l-jannati, N
o
1566. Tis verbal type is likewise attested in the form of
the re-interpreted verbal type A(b), along with the re-interpreted nomi-
nal type A(a). Te existence of the re-interpreted verbal type A(b), whose
verb varies according to the syntactic context , clearly proves that when
the re-interpretation of al-h amdu li-llhi took place its constituents, in
spite of its formulaic character, must have been present in the minds of
the speakers and cannot have been a de-etymologized complex as, e.g.,
add how much? (< *qadr ayyi ayin) in Syrian Arabic or izzayy
how? (< * zayy) in Cairene Arabic . Both the nominal al-h amdu
li-llhi and the verbal h amidtu llha are equally attested as early as in
pre-Islamic poetry,
23
which proves that they had existed side by side
from the frst.
23
An example of al-h amdu li-llhi occurs in the Muallaqah of Imra al-Qays , verse
124, in Ahlwardt s edition, and for examples of the verbal type see Brockelmann (1922,
arabic allad as a conjunction 87
Rather, the re-interpretation took place because the implicit logi-
cal structure of many items of the Qurnic al-h amdu li-llhi llad is
causal. A typical example is al-h amdu li-llhi llad najjn mina l-qawmi
z -z limna Q 23:28, in Arthur J. Arberry s translation: Praise belongs to
God, who has delivered us from the people of the evildoers. Tis literal
translation of the relative clause is of course correct. However, when
considering the propositional structure of the Arabic sentence we can
easily discern that there is between the relative clause and the head of
the sentence an intrinsic connection consisting in the implication that
God is to be praised because He delivered the believers from the evil-
doers.
24
It should be noted that this causal structure does not automati-
cally exist in all al-h amdu li-llhi llad sentences but, it seems, only or
mainly in those which refer to certain individual bounties of God. In this
respect, it may, e.g., be doubted that the relation in al-h amdu li-llhi llad
xalaqa s-samwti wa-l-arda Praise be to God, who created the heav-
ens and the earth Qurn 6:1 or al-h amdu li-llhi llad wasia samuhu
l-as wta Praise be to God, whose ear comprises all sounds an-Nas ,
Sunan, Kitb at -t alq, N
o
3406 and Ibn H anbal , Musnad, Bq musnad
al-Ans r, N
o
23064, is as causal as the frst example or causal at all. But
the cases of al-h amdu li-llhi llad in the Qurn and still more those in
Tradition refer mostly to individual deeds of God and thus express also
an inherent causal relation.
A combination of a non-causal relation and a causal relation is, e.g.,
found in the following passage in Ibn Ab d-Dam a-f s (d. 1244
C.E.) Kitb adab al-qad: al-h amdu li-llhi llad ahidati l-uqlu
bi-qidamihi wa-wah dnyatih / wa-waqafat dna idrki jallihi wa-
az amatih || allad btadaa l-maxlqti bi-badi h ikmatih / wa-s awwara
116). Additional verbal examples occur in Labd s Dwn N
o
6, 1 (h amidtu llha wa-
llhu l-h amdu etc.) and N
o
5, 1 (wa-llhu rabb mjidun mah mdun). For examples of
the related bi-h amdi llhi, see Brockelmann (1922), and an additional example occurs in
the Muallaqah of an-Nbiah , verse 12, likewise in Ahlwardt s edition.
24
Such causal function of relative clauses is, it seems, not dealt with by Lehmann
1989 in his otherwise comprehensive morphological and semantic analysis of the rela-
tive clause in a great number of languages, including Arabic . Nor is his chapter Vom
Relativpronomen zur Konjunktion (1989, 389393) of much help for the problem of
allad as a conjunction , as he considers any relative pronouns that have lost their infec-
tion to be Konjunktionen if there is no resumptive pronoun in the subordinate clause .
How Lehmann thinks the transition from uninfected relative pronouns to true con-
junctions meaning that or because to have been remains unclear, in spite of some
fnal remarks (1989, 391).
88 werner diem
l-kinti bi-lut f s anatih etc.
25
Praise be to God to whose pre-existence
and uniqueness all minds testify and whose magnifcence and power
they are unable to grasp, who made creation with His wonderful and
singular wisdom and designed the existing world with His kind work,
etc. Te frst allad clause is non-causal as it describes attributes of God,
and thus it is not possible to paraphrase it with a causal clause such as
*Praise be to God because all minds testify to His priority and unique-
ness and are unable to grasp His magnifcence and power. In contrast
to the frst allad clause , the second one is inherently causal, which is
evident from the fact that it can easily be re-formulated as a causal
clause : Praise be to God because He made creation with His wonder-
ful and singular wisdom and designed the existing world with His kind
work, etc. Te causal character of this allad clause is due to the fact
that men own their existence to God having created them and thus have
to thank Him for His benevolence. It is also interesting that the content
of the frst allad clause is of a more inherent nature than the second in
that it describes inherent, inalienable attributes of God, while the sec-
ond refers to a deed that God performed of His own will. Tis hierarchy
of the two allad clauses conforms with a rule of Arabic syntax which
says that attributes follow the noun they refer to in order of decreasing
inherence.
Considering that there is an inherent causal relation in many items
of the Qurnic al-h amdu li-llhi llad between the head of the sentence
and the relative clause , it comes as no surprise that the allad that con-
nects the two parts of the sentence and thus holds the position which a
conjunction explicitly expressing the causal relation would have, should
have been re-interpreted as a causal conjunction . Tis re-interpretation
was in all probability triggered by parallel constructions in which the
same causal relation is expressed explicitly. Tese constructions are as
follows:
(a) al-h amdu li-llhi al + noun /pronoun praise be to God for . . . and
h amidtu llha al + noun/pronoun I praised God for . . . In both
constructions, the al phrase indicates the reason why God shall
be praised. Tey are so frequent, particularly the former, that it is
superfuous to give examples.
25
Adab al-qad I, p. 247.
arabic allad as a conjunction 89
(b) al-h amdu li-llhi (al) an(na) praise be to God (for) that . . . and
h amidtu llha (al) an(na) I praised God (for) that . . . Te prepo-
sition al can be elided as can any preposition preceding an(na).
Sentences of this kind seem absolutely normal, but I can adduce
only few examples of the verbal type, one from pre-Islamic times
and the other examples later: h amidtu llha an ams Rubayun *
bi-dri l-hni malh yan muqm (Wfr) I praised God that
Rubay got * in the house of disgrace, placed there in a shameful
way (Maqil b. Xuwaylid ) Dwn al-Hud alyn I, N
o
14, 3; istahalla
yah madu rabbahu an l yakna as bahu d h aqqin f l-fayi and
he began to praise his Lord that none who had a right to the booty
had injured him ad-Drim , Sunan, Kitb al-muqaddimah, N
o
91;
allha ah madu al an jaalan min ulami l-arabyati God I
praise that he made me belong to the scholars of the Arabic lan-
guage az-Zamaxar , Mufas s al, 2, 2. Another example of the verbal
type with God as the subject of an anna-clause is cited by Blau for
Christian Arabic for kr I: akara llha annahu lam yuxallhi yusu
il abdihi He thanked God that He had not let (or made) him
act unjustly to His servant (19661967, 526). With subordinate
clause s whose subject is coreferential with Allh (second and third
examples), this construction is optional instead of (a), whereas it
is the only possible construction for subordinate clauses whose
subject is not coreferential with Allh (frst example). It cannot be
excluded that this construction is more usual in or even restricted to
the verbal type , which is less formulaic than the nominal type . More
examples would be desirable; their being so scarce so far is perhaps
due to my (and my predecessors) not noting them simply because
they are so normal and therefore do not seem to deserve attention.
(c) ah maduhu (al) an(na) I praise Him (for) that . . . Whenever
Allh is replaced by the personal pronoun , only this construction
is possible. Tis construction is especially frequent as a continu-
ation of the relative clause type in the religious arengas (xut bahs)
of Mamluk documents: al-h amdu li-llhi llad . . . nah maduhu al
annahu . . . Tis construction would also be the only one possible
for the nominal head lahu l-h amdu but I have no examples.
(d) h amidtu llha id/fa- I praised God because. In this type, the sub-
ordinate clause is not dependent on the verb of the head as in the
preceding types but forms a subordinate causal clause . Examples
of each of the two conjunctions are the pre-Islamic verse h amidtu
90 werner diem
ilh bada Urwata id naj * Xirun wa-badu -arri ahwanu min
badin
26
(T awl) I praised my God afer (the death of my brother)
Urwah (in battle) because (my son) Xir was (at least) saved, *
and some evil is easier to bear than another (Ab Xir ) Dwn
al-Hud alyn III, N
o
14, 1, and, in Tradition , y-iata h mad llha
fa-qad barraaki llhu O iah, praise God because God has excul-
pated you! al-Buxr , Sah h , Kitb a-ahdt, N
o
2467.
Tese functional and syntactic correspondences are set of in the fol-
lowing table, in which all sentences express the notion of praise being
due to God for the salvation of the speakers, which is directly or indi-
rectly ascribed to Him. Of the types marked with ?, no evidence has
been adduced so far, but they would be normal Arabic from a syntactic
point of view.
(a) Nominal type
Head Expansion Head Expansion
al-h amdu li-llhi llad najjn Praise be to God who saved us
al-h amdu li-llhi al najtin Praise be to God for our salvation
?al-h amdu li-llhi (al) an najjn Praise be to God (for) that He saved
us
?al-h amdu li-llhi (al) an najawn Praise be to God (for) that we were
saved
(b) Verbal type
Head Expansion Head Expansion
h amidtu llha llad najjn I praised God who saved us
h amidtu llha al najtin I praised God for our salvation
h amidtu llha (al) an najjn I praised God (for) that He saved
us
h amidtu llha (al) an najawn I praised God (for) that we were
saved
h amidtuhu (al) an najjn I Praised Him (for) that He saved
us
h amidtu llha id/fa-najjn I praised God because He saved us
h amidtu llha id/fa-najawn I praised God because we were
saved
26
bad in the rhyme.
arabic allad as a conjunction 91
It is from this specifc constellation that the re-interpretation of allad
in al-h amdu li-llhi llad and h amidtu llha llad as a causal conjunction
meaning (for) that started. Since the relative clause type shared with
the other types both the overall construction and the causal meaning of
the subordinate clause , allad could by analogy be interpreted as a par-
ticle expressing this causal relation. It is in sentences such as al-h amdu
li-llhi llad najawn Praise be to God (for) that we were saved and
h amidtu llha llad najawn I praised God (for) that we were saved,
with the subject of the allad no longer being coreferential with Allh of
the head , that the re-interpretation surfaced.
It should be stressed that this re-interpretation of the relative clause
as a causal subordinate clause was only possible because of the three fol-
lowing characteristics of al-h amdu li-llhi llad (and the corresponding
h amidtu llha llad ), of which the frst is syntactic , the second semantic
and the third morphological .
(a) Te relative clause of al-h amdu li-llhi llad is non-restrictive,
which means that it does not serve the purpose of identifying Allh , who
is the object of praise. It is the loose connection typical of non-restric-
tive relative clauses that permitted the syntactic shif of the subordinate
clause from its status as a relative clause depending on the antecedent
Allh to its new status as a causal subordinate clause which no longer
depends on Allh but on the head of the sentence as a whole.
(b) Verbs such as to praise or to thank have, due to their specifc
semantics , an inherent causal complement indicating the grounds for
praise or gratitude. Even if in the case of an isolated H amdalah in a
religious context the reason is not explicitly expressed it is neverthe-
less there, consisting in Gods general bounties, the knowledge of which
exists in the religious subtext. Tus whenever something positive is
mentioned in the syntactic context of a H amdalah it is in principle sub-
ject to being interpreted as being the reason for it. Exactly this happened
in the case of the relative clause of the H amdalah , but it happened only
there because it is only the non-restrictive clause that, due to its struc-
ture, has the syntactic potential of such a subordinate clause , while this
is not the case with, e.g., nominal attributes referring to Allh .
(c) Te re-interpretation of allad may have been helped by the fact
that at the time when it happened the relative pronouns had probably
already been reduced in the spoken language to allad , which etymo-
logically is the pronoun of the 3rd masc. sing. As a result of this develop-
ment, the congruence between the antecedent and the relative pronoun
in terms of gender, number and, in the case of the dual , case had
92 werner diem
disappeared, and thus the syntactic connection between the antecedent
and the relative clause had been weakened.
However, behind this development there is yet another factor, which
is of an ontological nature. Of the constructions mentioned above sub (a)
and (b), those where the subject of the subordinate clause is not coref-
erential with Allh are as a rule not or less possible with other syntactic
object s of to praise, unless the specifc reason of praise is mentioned
before or otherwise known to the person addressed. Tus one could say
in English (and correspondingly in German and Arabic and many other
languages), I would like to thank you that you helped me (coreferen-
tial subject of the subordinate clause ), but usually not I would like to
thank you that we were helped (non-coreferential subject of the sub-
ordinate clause ), whereas the second, non-coreferential construction is
a normal construction with God as the syntactic object of to thank, to
praise: (I) thank God that we were helped. Te diference lies here in
that whenever something (positive) happens it can be attributed by the
believer to God as the one who with His power has caused it. Tis is why
the H amdalah (nominal or verbal ) could be followed from the begin-
ning by non-coreferential causal clauses (al an, an, id) as mentioned
above, and this is also why in al-h amdu li-llhi llad the coreferentiality
of the subject of the relative clause with its antecedent Allh was not an
inherently indispensable syntactic feature , a fact which made it possible
for coreferentiality to be given up in the course of the re-interpretation
of the relative clause .
As for the question of chronological priority of the re-interpreted
nominal and verbal types, Spitaler seems to have assumed the verbal
type to have been secondary (1962, 105). In support of this assumption it
may be argued, as Spitaler has, that there is evidence of the nominal type
from as early as the 9th century C.E. whereas the verbal type is attested
somewhat later, but this does not necessarily mean that the verbal type
had not existed before the earliest written evidence of it. More convinc-
ing is the argument that it is only the nominal type al-h amdu li-llhi
llad that is part of the Qurn and thus a formula the Muslims had been
familiar with from the beginning of Islam . Accordingly, we fnd in Tra-
dition numerous examples of (the correct) al-h amdu li-llhi llad + a
coreferential verb and many examples of al-h amdu li-llhi al + a noun ,
but few examples of the verbal type . We could even say that al-h amdu
li-llhi unexpanded or expanded has been and still is part of everyday
Muslim speech , while this is not the case with the verbal type to the same
arabic allad as a conjunction 93
extent. On the other hand, it is the verbal type who seems to exhibit
greater variability, and thus it may have been this type, or perhaps rather
the existence of this type in its various subtypes, that may have triggered
the re-interpretation of the more formulaic nominal type .
Besides the examples of h md, there are examples of the same basic
structures containing synonymous expressions, which shows a frst
semantic generalization of the re-interpreted allad . Tere is my example
(2) fa-li-llhi l-h amdu wa-l-minnatu llat (!) knati l-aqbatu h amdatan,
and among Spitaler s examples there is ukuri r-Rah ma r-Rah mn /
allad lam tajid al-malik Qays f hd l-makn danke dem Allbarm-
herzigen, dass du den Knig Qais nicht an diesem Ort gefunden hast
from the Antar novel, where additionally Allh is replaced by its syn-
onym ar-Rah mn ar-Rah m, reversed there because of the rhyme. Blau
adduces further examples of synonymous verbs (sbh II, kr I, mjd II),
which, however, are syntactically ambiguous (1967).
We have reason to assume that the synonymous roots had already
been in use, besides h md, as alternatives of h md from the frst, as we have
nominal examples such as wa-li-llhi l-h amdu wa--ukru To God praise
and thanks are due! (Dietrich , Briefe Hamburg, N
o
42a, 3 [c. 916/7 C.E.],
and likewise in Tradition fa-laka l-h amdu wa-laka -ukru Ab Dwd ,
Sunan, Kitb al-adab, N
o
4411); wa--ukru li-llhi Tanks be to God!
(Anawati and Jomier , Papyrus chrtien, line 3 [9th c. C.E.]); wa-li-llhi
-ukru kat ran kam huwa ahluhu wa-mustah iqquhu To God repeated
thanks are due as he is entitled to and worthy of it (Rib , Lettres II,
N
o
17r, 5 [9th c. C.E.]). Furthermore, there are the Christian formulae
as-subh u li-llhi Praise be to God! and al-majdu li-llhi Glory be to
God!, for which the reader is referred to my Briefe Heidelberg, 24 and 25
respectively. An example of ukr and a relative clause is fa-aqimi -ukra
li-llhi llad s naka bi-d lika Terefore extend in the right way your
thanks to God, who preserved you thereby (Rib , Lettres I, N
o
5, 22
[9th c. C.E.]). When al-h amdu li-llhi llad and h amidtu llha llad were
re-interpreted, this development comprised the synonyms of h md also.
A fnal point to be made regards the translation of the re-interpreted
allad as that, da. Asking why allad can be translated with simple
that da, while, according to what I have tried to point out above, its
re-interpretation is based on an implicit causal relation, we should be
aware of a hitherto neglected fact: the syntactic opacity of that, da.
As with Arabic an and anna, prepositions preceding that, da in
English and German can or must be elided, which leads to the basic
94 werner diem
logical relations, as expressed by the prepositions, becoming blurred on
the surface. But the prepositions can also be retained on the surface,
probably more so in German than in English . Tus, instead of the syn-
tactically opaque Lob sei Gott, da wir gerettet worden sind the more
explicit Lob sei Gott dafr, da wir gerettet worden sind is also pos-
sible.
4.3 Type B(a) malh allad (it is) good that
Half a dozen examples of this type were adduced by Spitaler (1962, 109f.)
for modern dialects , intermingled however with examples of the farih tu
llad type (for this latter type, see below 4.4). It was Woidich who distin-
guished between the two groups, of which he characterized the frst one
as satzquivalente Ausdrcke (1980, 226). Examples of his type A for
Cairene Arabic are, apart from il-h amdu li-llhi illi Gott sei Dank, da,
e.g., baraka illi ein Segen, da, ya-xs ra lli schade, da, and kuwayyis
illi gut, da. Corresponding examples were given by Spitaler for other
modern dialects , whereas he had, according to himself, not come across
examples in literary texts (1962, 110). Earlier evidence of this kind is my
example (5) from the 11th century C.E. containing al-wayl lan nah nu
llad woe is us that.
27
Two other examples from that time quoted by
Blau are y-baxtik allad what good fortune for you (fem. sing.) that
and y-h ayf al abk llad what a pity for your (fem. sing.) father
that (1961, 227 sub ). Tese three early examples contain references to
persons in their heads, but this is also the case with two modern Cairene
examples cited by Woidich for this type A, and therefore this should not
lead us to assume historically diferent types.
As Woidich (1980) begins his Cairene evidence of this type with two
examples of the re-interpreted il-h amdu li-llh illi it is clear that he con-
siders this type to be somehow the prototype of the whole group. Tis
conclusion is confrmed by Woidich (1989), where he derives this group
from the al-h amdu li-llhi llad expression, assuming as a frst step
Funktionsschwche and as a second step a re-interpretation accord-
ing to the Satzstruktur Prd.Subjekt as in h ilwa di prima ist die!
or b ikkalm da eine Schande sind solche Worte! As I have remarked
above (2.7), sentences of this kind are not likely to have been connected
by native speakers with al-h amdu li-llhi llad .
27
allad might also be interpreted here as a relative pronoun standing for allad na.
arabic allad as a conjunction 95
In my opinion the solution to this problem should not and need not
be looked for in Cairene examples such as h ilwa di or their early Neo-
Arabic equivalents if any existed. But before going into details, I think it
appropriate to recapitulate some important points:
(a) Tat the type A(a) al-h amdu li-llhi llad is likely to be the starting
point of the type B(a) malh un allad may be assumed for the simple
fact that it is only al-h amdu li-llhi llad where the transition of the
relative particle allad to a conjunction can be explained, whereas
this is not feasible for malh un allad and the other expressions of
this type.
(b) Te sentences of my type B(a) have a comment-topic structure .
(c) When asking for a link between the type A(a) al-h amdu li-llhi llad
and the type B(a) malh un allad , one can rightly assume, as Woid-
ich did, that this link lies in that al-h amdu li-llhi llad had (fnally)
been reduced to being a (positive) evaluation of the fact mentioned
in the allad -clause . Tis reduced meaning of al-h amdu li-llhi llad
would have been equivalent to good that; nice that.
However, there remains the question of the structure of the type B(a)
malh un allad . In this respect, my theory difers from Woidich s. In my
opinion, we should not start from sentences such as Cairene h ilwa di
or b ik-kalm da nor their Neo-Arabic equivalents in order to explain
malh un allad and similar expressions. Asking rather whether there
are other examples of the structure comment + that + topic in Classi-
cal Arabic (and Neo-Arabic ), we can answer in the afrmative: there is
an absolutely normal type consisting of a noun + an(na) clause , as for
example malmun anna it is known that, h asanun anna it is good
that, barakatun anna it is a blessing that and so on.
28
Tis is why
we can transpose all examples of my type B(a) into normal Arabic by
replacing allad by an(na), with potential small changes due to anna
demanding an accusative . It should be noted that the transformation
is always possible from allad to an(na) but not always from an(na)
to allad . Tus the transformation of malmun an(naka) jita into
?
malm allad jita would be very questionable, and for Cairene Arabic
?
malm illi gt would probably not be acceptable to most speakers. Tis
28
In modern literary Arabic , the type mina l-malmi anna would be more likely to
correspond.
96 werner diem
is due to the restriction of this allad type to specifc head s referring to
emotions.
Starting from sentences evaluating events in an emotional manner
of the type barakatun (h asanun) an(na) + perfect, which were struc-
turally and semantically equivalent to al-h amdu li-llhi llad + perfect ,
the next step of the development, as I see it, was the replacement of
an(na) by allad in analogy with al-h amdu li-llhi llad . Tis develop-
ment was made possible by two factors: (a) A semantic -syntactic con-
nection between the type al-h amdu li-llhi llad + perfect and the type
barakatun (h asanun) an(na) + perfect was mentally established, due
to Funktionsschwche of the H amdalah . (b) Te allad of al-h amdu
li-llhi llad was mentally connected with an(na), because it overlapped
with it in function afer it had been re-interpreted as a conjunction .
Te generalization of allad led to an opposition between an(na) and
allad , with allad as the marked element being subject to certain restric-
tions as compared with the unmarked an(na). Tese restrictions, which
the type B(a) had inherited from the primary type al-h amdu li-llhi llad
Praise be to God who via the secondary type A(a), are as follows:
(a) Te structure of the type B(a) is always comment + allad + topic
and cannot be reversed. Tus malh allad jita It is nice that
you came is possible, whereas *
?
allad jita malh is not. Woidich
remarked this for his type B, with which my type B(b) partially
overlaps, but this restriction is valid of his type A and my type B(a)
also (1980, 235).
(b) Te head of the sentence consists of an evaluating expression.
Tus malh allad jita It is nice that you came is possible, whereas
*
?
malm allad jita It is known that you came is usually not.
(c) Te logical subject of the allad -clause is human. Woidich remarked
this for his type B for the syntactic subject of the allad clause , but
the same holds true for this type (1980, 230). Tus a sentence such
as malh llad was ala is possible in the meaning It is nice that he
(sc. a certain person) arrived, whereas the meaning It is nice that it
(sc. a letter or a parcel) arrived is excluded or questionable. What,
however, is attested in both the old and new evidence are allad
clause s with a non-human syntactic subject which however refer
to a logical human subject . Tus, a sentence such as malh allad
was alan It is nice that it (sc. the letter) reached me or, more freely,
arabic allad as a conjunction 97
It is nice that I received it would probably be possible, though
perhaps not accepted by all speakers of individual dialects .
(d) Te subject of the allad clause is almost exclusively pronominal
and as such mostly implied in a fnite verb . Tus again a sentence
such as malh allad was ala It is nice that he arrived is possible,
whereas malh allad was ala Ah mad It is nice that Ah mad arrived
is with all probability either not possible at all or at least doubtful in
most dialects .
(e) Te allad clause refers to actual, mostly past events. Woidich found
out this feature of Faktivitt for his type B, but this is also true of
his type A and my type B(a) (1980, 231). Tus malh allad jita It is
nice that you came is possible, whereas malh allad taj It would
be nice for you to come is not in most dialects .
Tese restrictions of allad in type B(a) as compared with an(na) cor-
roborate the assumption of its being the result of a generalization of
the original al-h amdu li-llhi llad Praise be to God who, whereby the
head was replaced with similar expressions while the other features of
the construction were retained:
(a) al-h amdu li-llhi llad najjn has the structure al-h amdu li-llhi +
allad clause, which is not reversible.
(b) Te head is evaluating.
(c) Te subject of the allad clause , which is Allh , is, in a anthropo-
morphistic perspective, human.
(d) Te subject of the allad clause is pronominal , being implied in the
verb that refers to Allh.
(e) Te relative clause refers to past events only due to the semantics of
to praise in the meaning of to thank.
4.4 Type B(b) farih tu allad I was glad that
Turning now to the type farih a allad he was happy that, that is, allad
afer verbs expressing emotions, the function of allad as a conjunction
is clearly borne out by cases where it follows a preposition as in (6) and
(8) of my examples which have bi-llad by that.
While Spitaler holds that this type goes directly back to the al-h amdu
li-llhi llad type by way of generalization (1962; see above 2.3), Woidich
98 werner diem
declares it to have had a separate origin in relative clause s where the subject
of the head and the subject of the allad clause are coreferential , as in Cairene
ana h mr illi dafat il-h isb (1989). In my opinion, this interpretation
is fawed in several respects, for which the reader is referred to para-
graph 2.7.
Even if I do not fnd Woidich s historical explanation of this type con-
vincing, it is worthwhile to mention that the sentence which, according
to him, was the starting point shares with al-h amdu li-llhi llad , in spite
of their diferent structures, the three characteristics mentioned above:
(a) Te relative clause is non-restrictive , (b) there is an underlying causal
relation between the head and the relative clause (I am an ass! Why?),
(c) illi goes back to the uninfected allad , to which the relative pronouns
had been reduced.
Before continuing, I think it appropriate and useful to draw atten-
tion to the fact that this type B(b) shares with type B(a) the restrictions
(b)(e) mentioned above, that is, all restrictions with the exception of
the one concerning specifc head s. As in the case of type B(a), a histori-
cal theory must take account of these restrictions and explain why they
exist. Te assumption that this type had developed from coreferential
relative clauses as claimed by Woidich would explain the restrictions
(a)(c) but not (d) nor (e), as some emotions can concern both past
and future events. Terefore, it is plausible to assume that this type B(b)
developed from the verbal type A(b) h amidtu llha llad in the same way
as the type B(a) developed from the nominal type A(a) al-h amdu li-llhi
llad . In the frst case, features of an impersonal expression were general-
ized, and in the other case, features of a personal expression.
Tis development, as I see it, was as follows: Te inherited conjunc-
tion in Arabic sentences of the type I was glad that he came was in
Classical Arabic , and still is in most dialects , an(na), e.g. farih tu an (or
annahu) ja in the classical language . In a certain sense, praising God
for an agreeable event that has happened is an expression of an emo-
tion. Saying, for example, in German Ich danke Gott, da ich davon
verschont geblieben bin, is equivalent to Ich bin froh, da etc. Tis
semantic overlap made it possible for the allad of h amidtu llha llad
I praised (thanked) God that to be generalized to verbs expressing
likewise the positive emotions of joy, contentment etc. Subsequently, as
Spitaler assumed, this generalization was extended, by Kontrastanalo-
gie, to verbs expressing negative emotions such as anger, reproach, grief
and so on (1962, 109).
arabic allad as a conjunction 99
As a result of this development, the replacement of the marked allad
in B(b) by the unmarked an(na) is always possible, whereas the replace-
ment of an(na) by allad is only possible where the an(na) clause fulflls
the above-mentioned restrictions.
4.5 Type B(c) alamtuhu llad I informed him that
In Spitaler s, Blau s and my material examples remain where allad that
is preceded by verbs not expressing emotions. As I have mentioned
above (2.3), this was a major problem for Spitaler .
In my opinion, the problem is less serious than Spitaler thought it to
be. It is true that the head s of these sentences do not contain expressions
of emotion, but the contents of the allad -clauses are something calling
for positive or negative comments. Take, for example, the two following
examples cited by Spitaler : wa-axbarhu llad qatalahu Antar und sie
teilten ihm mit, dass Antar ihn gettet hatte (1962, 111) and, quoting
Michel Feghali (1928, 313) for Lebanese Arabic , min hayk illi twafqti
b-jztik cest cause de cela que tu as t heureuse dans ton mariage!,
with Feghali adding: avec ironie. Supporting evidence for this are also
my example (11) with its frst allad and my examples (15)(16). In sen-
tences of this type, allad has the function of expressing a certain emo-
tion or the empathy of the speaker or narrator, the kind of which can be
inferred from the context only. Tus, allad in axbarhu llad qatalahu
Antar indicates a certain emotion on the part of the narrator, which,
depending on the specifc context, can either be positive (joy) or nega-
tive (grief).
It should be stressed that this kind of afective allad is not a result of
a secondary development but something inherent in allad that from
the beginning. From the moment when in al-h amdu li-llhi llad and
h amidtu llha llad the relative pronoun was re-interpreted as a con-
junction it became the marked counterpart of an(na), which through
this development became unmarked. Te diference between them lay
in that an(na) continued to express the simple that, while allad as its
marked counterpart expressed a certain emotional involvement of the
speaker.
Should one try to express the specifc afective content of allad in
sentences of this type B(c), one could do this, for example, by means
of adverbs. Consequently, the frst of the two sentences could be ren-
dered, depending on the context, alternatively with and they informed
100 werner diem
him that Antar had fortunately/regrettably killed him. Strictly speaking,
such explicit translation would be appropriate for any case of the con-
junction allad as long as it is restricted to this and the aforementioned
types, translating, e.g., Cairene ana mutaassif illi taxxart with I am so
sorry that I was late.
4.6 Remaining cases
Tere remain some examples in Spitaler s corpus and elsewhere where
the clause following allad (illi) does not refer to past, but to future or
possible events. It is interesting that two of Spitaler s three examples are
from Tunisian Arabic , among them ammin elli ts il l-martek w-uldek
sei sicher, dass du zu deiner Frau und deinen Kindern kommen wirst.
Here we have the fnal point of the development of allad . As Hans-
Rudolf Singer pointed out for the dialect of Tunis , illi is used there in
every context (1984, 669), apparently in the same sense and distribu-
tion as would be an(na) in Classical Arabic and inn- in many modern
dialects including Cairene Arabic , e.g. andna z-zhar illi lqnhum
29
wir
hatten Glck, da wir sie trafen on the one hand (referring to a past
event) and t h il-ahid illi ma yf h atta . . . er gab ihm das Verspre-
chen, da er sie bis [zum Hochzeitstag] nicht anschauen wrde (future
event).
30
In this dialect , illi that is unmarked, and its marked emo-
tional counterpart for expressing that is kf (< kayfa how), for which
this function is not an isolated phenomenon in modern dialects nor in
the pre-modern substandard language. Tis remark about kf typical of
the dialect of Tunis must sufce here as the hitherto unwritten story
of kayfa as a conjunction is too long to be told here in a few words.
Te dialect of Tunis is not the only one to have generalized the usage
of allad as a conjunction . Te same usage of illi/li that is found in
Maltese Arabic (Schabert 1976, 216 and Aquilina 1987, I 566), which is
no coincidence as La langue maltaise a pour origine un dialecte arabe ,
vraisemblablement proche des vieux dialectes citadins de Tunisie (Van-
hove 1993, 1). Te Jewish dialect of Tripolis in Libya is yet another dia-
lect using lli/li that (Yoda 2005, 278). In addition, similar extreme
examples of illi that are known from other dialects but we lack compre-
29
I have simplifed Singer s complicated transcription.
30
See also Woidich (1980, 224) for the same function of illi in the Judaeo-Arabic
dialect of Tunis.
arabic allad as a conjunction 101
hensive studies of their exact distribution. For Cairene Arabic , one can
state on the basis of Woidich s two studies that examples as cited above
for Tunisian Arabic would most certainly be declared to be ungram-
matical or unusual by native speakers .
As for the early history of Cairene Arabic , one should be cautious
about taking the language of the Jewish-Arabic documents of the Cairo
Genizah from the 11th century C.E. and later as early evidence of Cairene
Arabic , because many of the writers of those documents hailed from
the Maghreb , especially from what is today Tunisia , and there remained
strong bonds between the Jewish traders who had settled in al-Fust t
and elsewhere in the Islamic East and their relatives and partners in the
Maghreb . Tis means that deviations in those documents in the usage of
allad as a conjunction as compared with todays Cairene Arabic might
be ascribed to the Maghrebine background of their writers. Terefore
extreme examples of the conjunction allad in Judaeo-Arabic docu-
ments are to be connected to modern Tunisian Arabic rather than to
modern Cairene Arabic .
4.7 allad vs. an(na)
In describing the development of allad as a conjunction , I have not
dealt in detail with the various functions which it can assume. Some
additional remarks might therefore be helpful.
According to the development of allad as I see it, the relative pronoun
allad in al-h amdu li-llhi llad and h amidtu llha llad was frst re-inter-
preted in the sense of that, which is normally expressed by an(na), and
then, by a gradual generalization, it replaced an(na) in specifc syntactic
and semantic contexts. We can express this by saying that allad took
over part of the functions of an(na) step by step, a process that led to
an(na) and allad becoming the unmarked and marked members of an
opposition in many, if not all, dialects .
Tere is a well-known rule in Arabic that any preposition preceding
an(na) can be deleted. When allad replaced an(na) in specifc con-
texts, which have been described above, allad consequently was felt to
be submitted to the deletion rule as its counterpart allad was. Tis meant
that where, according to the syntactic context, a preposition was to be
expected before allad in analogy with an(na), this preposition could
be considered to exist in the deep structure of the construction. On the
other hand, since a preposition could be retained before an(na) without
being deleted, allad could likewise be preceded by prepositions , again in
102 werner diem
analogy with an(na). To put it another way, the existence of cases such
as (6) wa-an . . . mutabit un bi-llad ttas altu ilayhim I am . . . rejoicing
that I joined them, where allad is preceded by the preposition bi- typi-
cal of the verb bt VIII on which it depends, proves that the process
of the re-interpretation of allad as a conjunction meaning that had
taken place to its full extent. In the modern dialects , the combination
of illi with prepositions does not seem possible; at least Woidich does
not mention it for Cairene Arabic (1980 and 1989) , and I do not have
examples thereof either.
Let us now have a look at some of my examples and see what syn-
tactic status allad has and which prepositions are possibly missing. I
shall insert the prepositions missing in the surface structure but existing
in the depth structure in brackets, and also give a translation of these
prepositions:
(1) li-llhi l-h amdu {al} llad knati l-qibatu li-xayrin
Praise be to God for the result having been good
Remark: Te verb is h md I fulnan al to praise s.o. for s.th.
(15) wa-qad akartu tafaddulahum . . . {bi-}llad qad d akarn f
kitbihim bi-s-salmi wa-bi-fli l-jamli f bb
I am also grateful for their kindness . . . of giving greetings to me
and of performing good deeds to me
Remark: Te verb is fdl V bi- to be so kind as to do s.th.
(13) wa-dqa s adrun kat r {li-}allad lam yakn laka maahu kitbun
yut amminun
and we were very much distressed due to his not having a letter
of yours with him setting our minds at rest
Remark: (li-)llad indicates the cause as (li-)an(na) does.
On the other hand, there do exist many sentences where allad intro-
duces a subject or an object clause so that no preposition can be supple-
mented in the deep structure:
(7) wa-qad sarran llad anfad ta lahu rah lahu
It pleased me that you sent him his merchandise
Remark: the allad -clause is the subject of the sentence.
Tere is also the adnominal usage of allad , forming, as is frequent with
an(na), a syntagma that seems to be an apposition to a noun . It would
arabic allad as a conjunction 103
be difcult to elucidate the true syntactic status of allad here. However,
this is not necessary; it sufces to say that the status of allad is here
analogous to the status of an(na), by which it can be replaced.
(9) inn ajabu minka llad lam tus b man yaktubu laka kitb ill
daf atan
I am astonished at you that you (allegedly) found only once some-
body writing a letter down for you
5. Summary
In the preceding paragraphs, I have endeavored to sketch a picture of
the development of allad as a conjunction valid of all varieties of Neo-
Arabic as far as they are known. Te stages of the development are set
of in the following table:
(a) Original expressions containing relative clause s:
al-h amdu li-llhi llad h amidtu llha llad
Praise be to God who I praised God who
(b) Re-interpretation of allad as a causal conjunction on the pattern of
parallel constructions with semantically explicit syntactic means,
this process being due to (a) the relative clause being non-restric-
tive , (b) verbs such as to praise having an inherent complement
indicating the cause of praise, (c) the relative pronouns having been
reduced to allad , whereby the connection of the relative clause to
the head was weakened:
al-h amdu li-llhi llad h amidtu
31
llha llad
Praise be to God that I praised God that
(c) Generalization of allad for head s expressing positive emotions, by
(optional) replacement of an(na) with allad , while the syntactic
and semantic restrictions typical of (a)(b) were retained:
malh allad (for malh anna) farih tu llad (for farih tu anna)
It is nice that I was glad that
(d) Generalization of allad for head s expressing negative emotions cet-
eris paribus as in (c):
31
And by analogy semantically related verbs such as kr I, sbh II and mjd II.
104 werner diem
wayl laka llad (for wayl laka taassafu llad (for taassafu
anna) anna)
Woe is you that I regretted that
(e) Generalization of allad for head s not expressing emotions, with
allad still having an afective value and retaining its original restric-
tions as in (a)(d):
alamtuhu llad
I informed him (of the pleasant/regrettable fact) that
(f) Generalization of allad as an unmarked conjunction that without
syntactic or semantic restrictions:
t h il-ahid illi ma yf
He made a pledge that he would not see her (Tunisian Arabic )
Stage (f), which is found in the dialect of Tunis and the dialect of the
Jews of Tripolis (Libya) , is the fnal point of the development of allad as
a conjunction . Te development of allad (illi) in Cairene Arabic , which
is the best-known of all Arabic dialects , reached stage (d) only.
6. Appendix: Notes on the origin of al-h amdu li-llhi llad
As demonstrated in paragraph 4.2, there were, besides the relative clause ,
other, more explicit means of expanding the H amdalah . Tis raises the
question, which has not yet been given the attention it deserves, why
in the case of al-h amdu li-llhi a causal relation between a subordinate
clause and the head is expressed by a relative clause if the subject of
the relative clause is coreferential with Allh of the head , and not more
explicitly.
Indeed, in terms of normal Arabic one is in some respect entitled
to say that al-h amdu li-llhi llad expressing causal relation s is a special
case, which is, however, so innate in the religious language of Islam that
Muslims are not likely to be aware of this fact. On the other hand, the
very fact that the relative pronoun was re-interpreted, at some time in the
history of Arabic , as a conjunction demonstrates that this relative clause
was something peculiar not wholly in line with the normal syntax of
Arabic . Tat Western Arabists do not seem to have felt al-h amdu li-llhi
llad to be something special either, must be ascribed to the existence
of corresponding relative clauses in the religious style of the Western
languages . For German , I can say by way of introspection that this kind
of causal relative clauses , apart from the religious language , e.g. Lob
arabic allad as a conjunction 105
sei Gott dem Herrn, der etc., is restricted to a high stylistic level and is
possible for verbs such as danken, preisen, loben only, e.g. Es ist mir
ein Anliegen, an dieser Stelle Herrn N.N. zu danken, der dieses Projekt
hochherzig gefrdert hat. A corresponding formulation in less formal
language is, e.g., Ich mchte Herrn N.N. dafr danken, da er dieses
Projekt untersttzt hat. So the fact that al-h amdu li-llhi llad seems
familiar to us must not prevent us from enquiring as to the reason for
this construction. To this question the following lines will be dedicated.
Considering that al-h amdu li-llhi was in use in pre-Islamic times
already (see above 5.2), we can assume that its expanded form al-h amdu
li-llhi llad is likewise pre-Islamic , although there is no explicit evidence
of it so far, at least none that I am aware of. Te basic formula itself is
probably a calque on the Syriac ub h l-alh, as remarked by Teodor
Nldeke and Friedrich Schwally (1909, 112, footnote 1) for the Ftih ah
of the Qurn : entspricht genau syrischem ub h l-alh
32
bezw.
teboh t l-alh und neutestamentlichem . In this context,
Nldeke and Schwally also mention the so-called Berk in the Old Tes-
tament and the Christian liturgy as a parallel to the Syriac and Greek
formulae, without, however, connecting it to the Qurn . Anton Baum-
stark s article Jdischer und christlicher Gebetstypus im Koran (1922)
is in a certain sense a comment on the short remarks of Nldeke and
Schwally , although he does not refer to them. In the following, I shall
frst sum up some important points of Baumstark s article and then add
some deliberations of my own:
(a) In the Old Testament , there is the so-called Berk (blessing) of the
structure brk Yahw Blessed is (or be) Yahwe, to which a nomi-
nal attribute or a relative clause can be added, e.g. brk Yahw
a
er
his s l et k em miy-yad Mis rayim -miy-yad Par Blessed be Yahwe,
who delivered you from the hand of the Egyptians and the hand of
Pharaoh Exodus 18:10.
33
A corresponding Qurnic expression is
tabraka in the subtypes tabraka llhu rabbu l-lamna Blessed
be God, the Lord of the whole world Q 7:54 and tabraka llad
bi-yadih l-mulku Blessed be the One in whose hand is the king-
dom Q 67:1 , as well as the exceptional brika in brika man f n-nri
32
In the original text in Syriac script.
33
Tis passage has already been cited by Spitaler (1962, 107) but in another con-
text, where he asks whether this expression might be a parallel to the development of
al-h amdu li-llhi llad . Instead of Spitaler s h is s l, read his s l.
106 werner diem
wa-man h awlah Blessed be He Who is in the fre and around it
Q 27:8. Baumstark tends to assume that the Berk was transmit-
ted to Muh ammad via Christian formulae in the New Testament
and elsewhere, in which the Jewish Berk lives on, of the type
, a last echo of this development being Benedictus
dominus (Baumstark 1922, 231f.). A fact not mentioned by Baum-
stark is that the Qurnic formula is syntactically diferent both
from the Biblical Berk and the H amdalah , as it lacks a subtype
tabraka llhu llad Blessed be God who.
(b) Te type which in the Christian liturgy , especially of the East, has
become the prevailing one is the so-called doxology , which has prob-
ably developed from the Hebrew Berk , e.g. o , Gloria tibi
domine, Laus tibi Christe, and so forth, the doxology being [eine]
possessive Form der Prdikation, durch welche Herrlichkeit, Ruhm,
Lob, Ehre oder wie immer man das schillernde griechische
wiedergeben will, als der Gottheit eigen oder gebhrend bezeich-
net wird (Baumstark 1922, 234). Te dative of the Greek formula
can be expanded by eine partizipiale Apposition oder Anrede, and
the reason for Gods praise can also be given in the form of a
clause . Tis Christian doxology is likewise found in the Qurn , in
two forms: as subh na + a pronoun or a genitive (which does not
interest us in this context) and as al-h amdu li- (Baumstark 1922,
234f.). However, as Baumstark remarks:
Der frhchristlichen partizipialen Ergnzung einer Doxologie entspricht
diejenige eines
mil 137
2. Te amal in sentences containing a z arf
2.1 A z arf cannot operate as the mil producing the nominative in the
subject
Sbawayhi believes that in nominal sentences a z arf cannot operate
as the mil producing the nominative in the subject (= al-mubtada ),
irrespective of whether the z arf occurs as an indispensable predicate
(= mustaqarr ) or as a dispensable part of the sentence. Hence, in a sen-
tence like fh abdu llhi qiman (Sb. I:222,15), the mil of the subject
abdu llhi is not the predicate fh, but the abstract mil called al-ibtid .
12
In Sbawayhi s view, the sense of al-ibtid is the act of putting the noun
in a position where it is unafected by any word operating as an mil .
13
Te view that the z arf cannot be the mil of the subject derives from
the notion that in a nominal sentence , a word operating as an mil pro-
ducing the nominative must be logically identical with the noun afected
by it. For example: in the sentence abdu llhi axka Abdallah is your
brother (Sb. I:6, 11), the subject abdu llhi is logically identical with
the predicate axka, since Abdallah is your brother and your brother
is Abdallah.
14
Hence the subject abdu llhi is the mil producing the
nominative in the predicate axka.
15
In contrast, in fh abdu llahi
qiman, the predicate fh is not identical with the subject abdu llhi,
since it is an expression denoting the place where the subject is, and
hence fh cannot be the mil producing the nominative in the subject
abdu llhi.
16
As a result, this sentence does not include any word that
can operate as an mil producing the nominative in the subject abdu
llhi. Hence, abdu llhi takes the nominative because of the efect of
al-ibtid .
17
Sbawayhi contends that examples beginning with inna, such as inna
fha zaydan (Sb. I:222, 20), confrm that fh does not produce the
12
See Sb. I:222, 14223, 18.
13
See Levin , (forthcoming), mubtada , 4.1.
14
This notion is discussed in detail in Levin , 1979, 199202; Levin , 2002, 359,
15360, 11; Levin , 2006, 110111, 5; Levin , forthcoming, Cahiers linguistiques, 3.2.
15
For this notion see Sb. I:239, 59.
16
See Sb. I:222, 14223, 18.
17
See Sb. I:222, 1419.
138 aryeh levin
nominative in the subject , since in the above example inna is the mil
of the subject zaydan, as is shown by its accusative ending .
18
Although Sbawayhi believes that the z arf cannot be the mil of
the mubtada , it is inferred that he holds that the z arf can produce the
accusative in nouns occurring as a h l , as in nominal sentence s of the
type abdu llhi fh qiman and fh abdu llhi qiman Abdallah is
standing in it (Sb. I:222, 15).
19
In his view, in these examples, the mil
producing the accusative in the h l qiman is the z arf fh.
20
Sbawayhi
also explicitly says that the z arf is the mil producing the accusative in
words denoting measures of distance occurring as a tamyz
21
(see below
3).
2.2 Arguments confrming that in Sbawayhi s view the z arf is the mil
of the h l
Sbawayhi does not explicitly say that the z arf is the mil producing the
accusative in the h l in certain constructions of the nominal sentence .
However, his view in this respect is inferred from some places in the
Kitb text.
2.2.1 In his discussion of sentences beginning with m kna, Sbawayhi
refers to two types:
(i) sentences where the z arf fh occurs as a mustaqarr , i.e., as the
indispensable predicate of the sentence, as in m kna fh ah adun
xayrun minka nobody better than you was in it (Sb. I:21, 7).
(ii) sentences where the z arf fh occurs as a dispensable part of the
sentence (mulan or law ), as in m kna ah adun xayran minka
fh (Sb. I:21, 10).
In this discussion Sbawayhi explicitly says that a z arf which is a mus-
taqarr can operate as an mil .
22
He also adds here that since a mustaqarr
18
Sb. I:222, 19223, 1.
19
Sbawayhi refers to the accusative in these examples both as a h l (Sb. I:223, 1) and
a xabar (Sb. I:222, 140). For this special use of xabar see Levin , 1979, 193196, 2.4.
20
See Sb. I:222, 49. See also Sb. I:167, 1116; Sb. I:218, 616, especially lines
1213. See below 2.22.2.3.2.
21
The later grammarians term tamyz does not occur in the Kitb.
22
See Sb. I:21, 719.
sbawayhis view of the zarf as an
mil 139
can operate as an mil , it is preferable to put it in a position where it
precedes the noun afected by it. Te more the speaker makes it precede
this noun the better, he says. In contrast, he says that when the z arf is
not a mustaqarr , it is preferable to put it at the end of the sentence, or at
least in a position close to the end.
23
2.2.2 In his above discussion, dealing with the possibility that a z arf
can operate as an mil , Sbawayhi does not give any example illustrating
this point. However, it is clear that his statements in this respect refer to
examples such as fh abdu llhi qiman and abdu llhi fh qiman,
discussed elsewhere in the Kitb.
24
Since Sbawayhi explicitly says that
in these examples the mil of the mubtada abdu llhi is not the z arf ,
but al-ibtid (see above 2.1), it is inferred that the mil producing
the accusative in qiman, which is a h l , is the z arf fh. It should be
stressed that the z arf fh is the only part of the sentence which can be
the mil producing the accusative in qiman, since the mubtada abdu
llahi is logically identical with qiman, and a noun which is logically
identical with another noun can produce in it only the nominative , as in
fh abdu llhi qimun (see above 2.1).
2.2.3 Te above inference is supported by the following considerations:
2.2.3.1 In his discussion of the example laka -u tan bidirhamin
tan bidirhamin [Te sale of] the sheep became binding on you
25
[in a situation where their price is] one dirham a sheep, one dirham a
sheep (Sb. I:167, 13), Sbawayhi says that the accusative tan occur-
ring twice in this example is a h l . He explains how the accusative is
produced in tan as follows: was ra laka -u id nas abta bimanzilati
wajaba -u kam kna fh zaydun qiman bimanzilati staqarra zay-
dun qiman when you put [tan] in the accusative [the utterance]
laka -u is equivalent to wajaba -u, just as [the utterance] fh
zaydun qiman is equivalent to istaqarra zaydun qiman (Sb. I:167,
13).
26
Since Sbawayhi believes that in all verbal sentences the verb is the
23
See Sb. I:21, 919.
24
See Sb. I:222, 14224,2.
25
The translation here is based on Lane s rendering of the expression wajaba l-bayu
(see Lane VIII, 2922A,1517). According to Sbawayhi laka -u is equivalent to
wajaba -u.
26
See Sb. I:167, 1116. Cf. Sb. I:222, 14223, 1.
140 aryeh levin
mil of the h l ,
27
it is clear that in his view, the mil of tan in wajaba
-u atan bidirhamin tan bidirhamin is the verb wajaba, while the
mil of qiman in istaqarra zaydun qiman is the verb istaqarra. Since
Sbawayhi holds that laka and fh are equivalent to wajaba and istaqarra
which operate as awmil , it is inferred that in his view laka and fh
also operate as awmil in the corresponding examples laka -u atan
bidirhamin tan bidirhamin and fh zaydun qiman respectively. For
similar examples where a combination of li + genitive or li + a relative
clause operates as an mil see Sb. I:223, 18224, 2.
2.2.3.2 In referring to the example fh zaydun qiman fh In it is
Zayd, in it (Sb. I:238, 15) Sbawayhi says: fainnam ntas aba qimun
bistigni zaydin bifh [Te h l qiman] takes the accusative because
of the fact that when [the frst] fha is added to zaydun, it makes the
sentence complete.
28
Te signifcance of this wording is that since the
frst fh in the above utterance is the indispensable predicate making
the sentence complete, it is the mil producing the accusative in the h l
qiman in fh zaydun qiman fh.
3. Te z arf as an mil of a tamyz denoting a measure of distance
Sbawayhi explicitly says that in the example dr xalfa drika farsaxan
the wandering territory of my tribe is situated in a distance of one
parasang from that of yours (Sb. I:176, 67), the mil producing the
accusative in farsax is the z arf xalfa drika, which is the indispensable
predicate of the sentence.
29
He adds that this z arf produces the accusa-
tive in farsaxan and in other measures of distance, since it is not logi-
cally identical with farsaxan, and since farsaxan is not the s ifa of xalfa
drika. He compares the efect of xalfa drika on farsaxan with that of
irna on dirhaman, in the example irna dirhaman.
30
27
See Sb. I:15, 1822.
28
See Sb. I:238, 1418.
29
See Sb. I:176, 612. For xalfa as a z arf see Sb. I:177, 14.
30
See Sb. I:176, 611. For the amal in irna dirhaman see Carter , 1972. See also
Levin , (forthcoming), Cahiers linguistiques, 3.2.
sbawayhis view of the zarf as an
mil 141
4. Te il of the z arf
According to Sbawayhi , one of the syntactic qualities of the z arf is al-
il , lit. the abolishment. It is inferred from the text of the Kitb that
this term, when referring to a z arf occurring in certain syntactic con-
structions , denotes the abolishment of the status of the z arf as an indis-
pensable predicate of the sentence. In this case al-il is opposed to
al-istiqrr the occurrence of the z arf as an indispensable predicate .
31
In other syntactic constructions al-il denotes the abolishment of the
status of the z arf both as an indispensable predicate and as an mil . In
these constructions al-il is opposed both to al-istiqrr and al-iml ,
i.e., the appliance of the amal of the z arf to the case ending of a certain
noun.
32
Te following examples illustrate the contrast between al-il on the
one hand, and al-istiqrr and al-iml on the other hand in certain syn-
tactic constructions :
(1) In the example m kna fh ah adun xayrun minka nobody better
than you was in it (Sb. I:21, 7), the z arf fh is the indispensable
predicate of the sentence
33
(= al-mustaqarr ). In contrast, in m kna
ah adun xayran minka fh (Sb. I:21, 10), the il of the z arf fh
takes place, since the indispensable predicate is xayran minka, and
hence the status of fh as the indispensable predicate is abolished.
34
(2) In the examples abdu llahi fha qiman and fh abdu llhi qiman
(Sb. I:222, 15) the z rf fh is the indispensable predicate ,
35
and it
is also the mil of the h l qiman . In contrast, in fh abdu llhi
qimun (Sb. I:223,2) the il of fh takes place, since qimun is
the predicate , and fh loses its status as an indispensable predicate
and as an mil.
36
31
See Sb. I:21, 1415.
32
The contrast between il and iml is inferred from Sb. I:21, 914, where
Sbawayhi uses the forms milan and yamalni in contrast to al-il and alayta . For
the occurrence of the term iml in the Kitb, also when referring to other awmil , see
Troupeau , 1976, 149, voc. iml.
33
The expression indispensable predicate here is based on Sbawayhi s view of xabar
kna (see Levin , 1979, 203205, 2.6).
34
See Sb. I:21, 714.
35
See Sb. I:222, 14223,1.
36
See Sb. I:223, 12.
142 aryeh levin
Sbawayhi compares the il of the amal of the z arf with that of the
verbs later called afl al-qulb .
37
He says that when the il of the z arf
takes place, it is preferable to pronounce it at the end or close to the end
of the sentence. In contrast, when the zarf is a mustaqarr occurring as an
mil , it is preferable to put it at the beginning of the sentence, like verbs
such as z unnu and ah sibu, which are pronounced at the beginning of
the sentence when they are awmil .
38
Sbawayhi illustrates two types of taqdr construction of sentences
containing a z arf mulan :
(i) In referring to the sentence inna bika zaydan maxd un Zayd is
enchanted by you (Sb. I:242, 2), where the il of the z arf bika takes
place, Sbawayhi says that when the speaker expresses this sentence, it
is as if he were saying inna zaydan maxd un. Similarly, when saying
inna fka zaydan la ribun Zayd covets you (Sib. I:242, 5) it is as if the
speaker were saying inna zaydan ribun.
39
Tese taqdr constructions
illustrate the notion that when the above sentences are pronounced it
is as if the z arf bika and fka are not spoken, and hence they cannot
operate as the mil producing the accusative in maxd un and ribun
respectively.
It appears that Sbawayhi holds this view in order to solve a theo-
retical difculty arising from one of the main principles of the theory
of amal : in his view, the efect of an mil producing the nominative
or the accusative in the noun is always applied, irrespective of whether
this mil is an indispensable part of the sentence or not.
40
Tis principle
seems to be violated if one assumes that when a z arf such as fh is an
indispensable predicate , as in fh zaydun qiman, it is the mil pro-
ducing the accusative in the h l qiman, but when fh is a dispensable
part , as in fh zaydun qimun, its amal is abolished. In order to solve
this difculty Sbawayhi says that when the amal of the z arf is abol-
ished, the z arf does not occur in the taqdr construction. Since accord-
ing to the grammarians the relevant construction, as far as grammatical
analysis is concerned, is that of the taqdr , Sbawayhi assumes that fh
37
The il of the amal of this category of verbs is discussed in chapter 31 of the
Kitb (= Sb. I:49, 452, 15).
38
See Sb. I:21, 1013.
39
See Sb. I:242, 28. Sbawayhi s words expressing this notion are very clear:
. . . kaannaka aradta inna zaydan ribun wainna zaydan maxd un walam tad kur fka
wal bika fauliyat hhun kam uliyat f l-ibtidi (Sb. I:242, 78).
40
See Sb. I:223, 613.
sbawayhis view of the zarf as an
mil 143
does not occur in the taqdr construction,
41
and hence it is clear that its
amal cannot be applied.
(ii) Sbawayhi says that when the speaker expresses the sentence
fh abdu llhi qimun he intends it is as if he were saying abdu llhi
qimun fha. Tis taqdr construction illustrates the view that when
the z arf is not an indispensable predicate and hence is not an mil , the
speaker imagines that it is as if he were pronouncing the z arf at the end
of the sentence, since as regards grammatical theory it is preferable to
pronounce a z arf which is not an mil at the end of the sentence, or at
least in a position close to the end (see above 2.2.1). Sbawayhi says
that when the il of the z arf occurs in an example like fh abdu llhi
qimun the speaker imagines that it is as if he were saying abdu llhi
qimun fh, since in this taqdr construction the zarf which is not an
mil occurs at the end of the sentence.
42
5. An interpretation of a difcult passage from the Kitb
(Sb. I:207, 1721)
Sbawayhi s discussion in Sb. I:207, 1721 is one of the most difcult
passages in his text. It contains some points relevant to the topic of this
paper.
In referring to the example marartu birajulin maahu s aqrun s idan
bihi adan I passed by a man with a hawk who was intending to hunt
with it tomorrow (Sb., Hrn II:52, 67),
43
Sbawayhi says:
walam annaka id nas abta f hd l-babi faqulta marartu birajulin maahu
s aqrun s idan bihi adan fal-nas bu al h lihi lianna hd laysa bibtidin
wal yubihu fh abdu llhi qimun adan lianna z -z urfa tul h att
yakna l-mutakallimu kaannahu lam yad kurh f hd l-mawdii, faid
s ra l-ismu majrran aw milan fhi flun aw mubtadaun lam tulihi
liannahu laysa yarfauhu l-ibtidu, waf z -z urf id qulta fha axawka
qimni yarfauhu l-ibtidu
41
For taqdr constructions which are shorter than their corresponding literal con-
structions see Levin , 1997, 146148, 3.3.
42
See Sb. I:222, 22223,6.
43
Derenbourg s edition has idan instead of s idan (see Sb. I:207, 1718). How-
ever, Hrn s version s idan is supported by Sb. I:206, 8 in Derenbourg . The version
s idan also occurs in all the later grammarians treatises (see, for example, al-Fris , I:
250, 710; as-Srf , VI: 131, 7 (in a quotation from Sbawayhi s text).
144 aryeh levin
know that if you put a noun in the accusative in [syntactic constructions
of] this type,
44
and you say marartu birajulin maahu s aqrun s idan bihi
adan, the accusative ending [occurring in s idan] remains unchanged,
because this [utterance]
45
is not [an independent sentence] occurring at
the beginning of the utterance,
46
and it does not resemble [the nominal
sentence ] fh abdu llhi qimun adan (= Abdallah is standing in it
tomorrow) [where the form qim can take either the nominative or the
accusative ],
47
because [when expressing sentences such as fh abdu llhi
qimun adan ] the [amal ] of the expressions denoting place (= z urf )
is abolished, [and the feeling of] the speaker is that [when expressing his
literal utterance he intends] it is as if he were not pronouncing the z arf
in this place at all.
48
When the noun [rajulin] takes the genitive [in the
example marartu birajulin maahu s aqrun s idan bihi adan], or when
it is afected by a verb [as in the example raaytu rajulan maahu s aqrun
s idan bihi adan]
49
or [when it is afected by the mubtada [in an example
such as hd rajulun maahu s aqrun s idan bihi adan]
50
you do not abol-
ish the amal [of the z arf maa which produces the accusative in s idan],
because the ibtid does not produce the nominative in the noun [rajulin
in the example marartu birajulin maahu s aqrun s idan bihi adan], [and
hence this noun is not a mubtada which can be the mil producing the
nominative in the word s id]. [On the other hand, in sentences beginning
with] a z arf ,
51
when you say fh axawka qimni (= your two brothers
are in it), the ibtid produces the nominative in [the mubtada, which is
axawka]
52
[so this mubtada is the mil producing the nominative in the
xabar , which is qimni (Sb. I:207, 1721).
53
44
I.e., in syntactic constructions including a relative clause beginning with a z arf ,
such as marartu birajulin maahu ksun maxtmun alayhi I passed by a man having
with him a sealed sack (Sb. I:207, 1516). Many examples of this type are discussed in
Chapter 112 of the Kitb (=Sb. I:206,5210,2).
45
I.e., the clause maahu s aqrun s idan bihi adan.
46
The term al-ibtid here denotes a position occurring at the beginning of the utter-
ance. For al-ibtid in this sense, see, for example, Sb. II:295, 16; 296, 1115; 297, 36;
362, 1623.
47
See Sb. I:222, 14223, 2. See above 4.
48
I.e., when the speaker says fh abdu llhi qimun he intends it is as if he were say-
ing abdu llhi qimun (for this notion see as-Srf , VI:135, 18136, 5; see above 4).
49
This example does not occur in the Kitb. It has been introduced here according
to al-Fris s interpretation (See al-Faris , I:251, 35) in order to explain Sbawayhi s
intention.
50
This example does not occur in the Kitb. It is introduced according to al-Fris s
interpretation (see al-Faris , I:251, 37), in order to explain Sbawayhi s intention.
51
In Sbawayhi s manner of expression a combination such as f az -z urf denotes the
sense of in sentences beginning with a z arf . Similarly, the expression f l-fil denotes
in sentences beginning with a verb (see Sb. I:17, 17); f darabain a sentence begin-
ning with daraba (see Sb. I:16, 1820); f knain a sentence beginning with kna
(Sb. I:17, 12).
52
For this notion see Sb. I:222, 14223, 18.
53
Ibid.
sbawayhis view of the zarf as an
mil 145
Te following remarks and conclusions are inferred from the above pas-
sage. Tese conclusions are supported by other texts in the Kitb, dis-
cussed in this paper.
(1) A z arf can operate as an mil only when it is a mustaqarr , i.e.,
only when it is the indispensable predicate of a certain sentence or of a
certain clause.
(2) Te z arf cannot be an mil producing the nominative in the sub-
ject or in the predicate . In a nominal sentence such as fh abdu llhi
qimun the mubtada abdu llhi takes the nominative because of the
amal of al-ibtid , and abdu llahi is the mil producing the nominative
in the predicate qimun.
(3) In a sentence such as marartu birajulin maahu s aqrun s idan
bihi adan, the noun rajulin cannot be the mil producing the nomi-
native in s id, since rajulin is not afected by the ibtid, and hence it
is not a mubtada . In Sbawayhi s view, in a nominal sentence , only a
mubtada can produce the nominative in a noun occurring as a predi-
cate . Te noun afected by the mubtada must be logically identical with
it. Since the sentence marartu birajulin and the clause maahu s aqrun
s idan bihi adan do not contain a mubtada logically identical with
s id, s id cannot take the nominative. Te only word, which can be the
mil of s id in the above utterance is the z arf maa, which can produce
the accusative in s id. Hence it is impossible to abolish the amal of
maa, since if its amal were to be abolished the word s id would remain
without an mil .
(4) But in fh axawka qimni , the mubtada axawka takes the
nominative because of the amal of al-ibtid , and hence it can be the
mil of the predicate qimni. It is also possible to say fh axawka
qimayni in it are your two brothers standing. In this structure
axawka takes the nominative because of the amal of al-ibtid , and
fh, which is a mustaqarr , is the mil producing the accusative in the
h l qimayni.
(5) Te form ibtid contained in the expression lianna hd
laysa bibtidin denotes an expression occurring at the beginning of
the utterance. It does not denote here any of the terms it designates
in Sbawayhi s terminology of the nominal sentence . Te words lianna
hd a laysa bibtid express the notion that the clause maahu s aqrun
s idan bihi is not an independent sentence occurring at the beginning
of the utterance.
(6) Tere is another argument which, according to as-Srf , pre-
vents the il of maahu in the clause maahu s aqrun s idan bihi adan:
146 aryeh levin
maahu, as-Srf says, includes the antecedent -hu , referring to rajulin
in marartu birajulin maahu s aqrun s idan bihi adan. If the il of
maahu were applied, it would have been dropped from the taqdr con-
struction, which in this case would be marartu birajulin s aqrun s idun
bihi adan. Tis taqdr construction cannot exist, since the clause con-
tained in it does not include an antecedent referring to rajulin. Since
in the grammarians view the taqdr construction is the relevant one
as far as grammatical analysis is concerned, it is impossible to apply
here the il of maahu, since this would create an unacceptable taqdr
construction.
6. Conclusions
1 Syntactically, Sbawayhi distinguishes between two kinds of a z arf :
1.1 A z arf which is an indispensable predicate of a nominal sen-
tence , as fh in the example fh abdu llhi qiman In it is
Abdallah standing. Tis kind of z arf is called mustaqarr [A
z arf] denoting the place where the subject is.
1.2 A z arf which is a dispensable part of the sentence, as fh in the
example abdu llhi qimun fh Abdallah is standing in it.
Tis kind of z arf is sometimes called mulan or law or ayr
mustaqarr.
2 A z arf which is an indispensable predicate (= mustaqarr ) is liable to
operate as the amil producing the accusative in a part of a sentence
occurring as a h l or a tamyz denoting a measure of distance. For
example:
(i) In fh abdu llhi qiman, fh is the mil of the h l
qiman.
(ii) In dr xalfa drika farsaxan the wandering territory of
my tribe is behind that of yours, at a distance of one paras-
ang, the z arf xalfa drika is the mil of the tamyz far-
saxan.
3 In contrast, a z arf which is not an indispensable part of the sentence
cannot operate as an mil . In Sbawayhi s view, this z arf undergoes
the process of al-il , i.e., the process of the abolishment of its sta-
tus as an indispensable predicate and as an mil . Hence, this z arf
is called mulan or law [a z arf whose] status as an indispensable
predicate and as an mil is abolished.
sbawayhis view of the zarf as an
mil 147
4 Sbawayhi illustrates two types of taqdr construction of sentences
containing a z arf mulan :
4.1 In referring to examples like inna bika zaydan maxd un Zayd
is enchanted by you (Sb. I:242, 2), where the il of the z arf
bika occurs, Sbawayhi says that when the speaker expresses
this sentence, he intends it is as if he were saying abdu llhi
maxd un. Tis taqdr construction illustrates the notion that
when pronouncing the above sentence the speaker intends that
it is as if the z arf bika is not pronounced, and hence it cannot
operate as the mil producing the accusative in maxd un.
4.2 When the speaker expresses the sentence fh abdu llhi
qimun he intends it is as if he were saying abdu llhi qimun
fha. Tis taqdr construction illustrates the view that when the
z arf is not an indispensable predicate and hence is not an mil ,
the speaker imagines that it is as if he were pronouncing the
z arf at the end of the sentence, since as regards grammatical
theory it is preferable to pronounce a z arf which is not an mil
at the end of the sentence, or at least in a position close to the
end.
7. References
7.1 Primary sources
al-Fris , Ab Al al-H asan b. Ah mad b. Abd al-afr. (d. 377/987). at-Talqa al
Kitb Sbawayhi. Iwad b. H amad al-Qz, ed. 1410 A.H. = 1990. Cairo.
Sbawayhi . (d. 177/793). Le livre de Sbawaihi. Trait de grammaire arabe. Hartwig
Derenbourg, ed. 18811889. Paris. 2 vols.
Ab Bir Amr b. Ut mn b. Qanbar. (d. 177/793). al-Kitb. Kitb Sbawayhi. Abd
as-Salm Hrn, ed. Cairo, 1977. 5 vols.
as-Srf , Ab Sad. (d. 368/979). arh Kitb Sbawayhi. Ramad n Abd at-Tawwb and
others, eds. Cairo, 19882004. 6 vols.
7.2 Secondary sources
Carter, Michael G. 1972. Twenty Dirhams in the Kitb of Sbawayhi, Bulletin of the
School of Oriental and African Studies. London 35, 485496.
Jahn, G. 1895. Sbawaihis Buch ber die Grammatik, bersetzt und erklrt von G. Jahn.
Vol. I, second pagination. Berlin.
Lane, E.W. 18631893. Arabic-English Lexicon. London (8 volumes).
Levin, Aryeh. 1979. Sbawayhis view of the Syntactical Structure of kna wa-axawtuh,
Jerusalem Studies in Arabic and Islam 1, 185211.
. 1997. Te Teory of al-Taqdr and its Terminology. Jerusalem Studies in Arabic
and Islam 21, 142166.
148 aryeh levin
. 2002. An Interpretation of a Difcult Passage from the Kitb. Jerusalem Studies in
Arabic and Islam 27, 356362.
. 2006. An Interpretation of Two Difcult Passages from al-Kitb Referring to the
mil in Elliptical Sentences. Jerusalem Studies in Arabic and Islam 31, 107117.
. (forthcoming). Sbawayhis View of al-mubtada and al-ibtid. Moshe Bar-Ashers
Festschrif.
. (forthcoming). Te mil of the xabar in Old Arabic Grammar. In: Cahiers lin-
guistiques de Inalco.
Troupeau, Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris: Klincksieck.
PROBLEMS IN THE MEDIEVAL ARABIC THEORY OF
SENTENCE TYPES
Yishai Peled
Tel Aviv University
1. Introduction
In medieval Arab grammatical tradition , the two basic sentence types
jumla fliyya and jumla ismiyya , are normally defned by the frst
occurring predicative constituent . A verb followed by its subject sig-
nals a jumla fliyya (verbal sentence ), e.g. daraba abdu-llhi zaydan
(Abdullh hit Zayd). By contrast, a sentence introduced by a nomina-
tival noun is a jumla ismiyya (nominal sentence ), e.g. zaydun rajulun
(Zayd is a man). Tis binary division corresponds with the grammar-
ians theory of amal (regimen), or, to be more specifc, with two basic
types of amal. A jumla fliyya correlates with a verbal mil (opera-
tor), whereas a jumla ismiyya corresponds with ibtid , which is con-
sidered an abstract mil.
1
Te ibtid is normally said to consist of a
zero phonological mil and the predicatival relationship between the
mubtada and the xabar , respectively the subject and predicate in this
type of sentence. Te basic principle of amal, stipulating that the mil
should precede the maml , applies in both sentence types. In a jumla
fliyya, the verbal mil afects the complements following it; in a jumla
ismiyya, the abstract mil, ibtid, occupies in principle a pre-mubtada
position , from where it assigns the raf case to the mubtada; the latter,
in turn, assigns raf to the xabar (according to Sbawayhi s [Kitb I:239]
version). As for cases such as daraba zaydan abdu-llhi and rajulun zay-
dun, these were presented as cases of taqdm wa-taxr (preposing and
postposing), i.e. as the inverted versions of daraba abdu-llhi zaydan
and zaydun rajulun; in other words, as inverted jumla fliyya and jumla
ismiyya respectively.
1
For a detailed discussion of this correspondence, see Levin 1985.
150 yishai peled
However, the linkage between the concept of two sentence types on
the one hand, and the theory of amal on the other, turned out to consti-
tute a major problem with regard to the binary division into jumla fliyya
and jumla ismiyya . Te grammarians realized that such types as daraba
abdu-llhi zaydan and zaydun rajulun, with their inverted versions, leave
various constructions that do not easily ft into any of the two categories.
Te two apparently most problematic cases may be represented by the
two model sentences qimun zaydun (Standing is Zayd), where a par-
ticiple is followed by a defnite noun phrase , and zaydun f d-dr, or f
d-dri (or fh) zaydun (Zayd is in the house/in it), where the predicate
position is occupied either by a defnite prepositional phrase or by an
adverbial phrase such as hun, hunka etc. In the latter type, a defnite
subject noun may either precede or follow the adverbial /prepositional
predicate ; an indefnite subject must obligatorily follow its predicate .
As it were, sentences such as qimun zaydun and f d-dri (or fh)
zaydun may be considered as cases of an inverted jumla ismiyya , prag-
matically motivated. And modern linguists , uncommitted to the the-
ory of amal , would probably regard them as such. Yet for many of the
medieval grammarians , they represent, rather, a sentence type in its own
right. To be more precise, f d-dri zaydun is explicitly presented as such
by some of the grammarians ; qimun zaydun, by contrast, is ofen dealt
with in a way that leads one to believe that it was considered by certain
grammarians as representing a third sentence type.
2
I would argue that
the controversies that arose over these (and other) constructions point
to what may be viewed as gaps in the medieval theory of sentence types .
In other words, an attempt will be made to show that the theory, based
on a binary conception of jumla fliyya and jumla ismiyya , representing
two types of amal, was highly vulnerable and far from stable. Tis paper
concentrates on the theoretical problems presented by the above two
constructions. We start with qimun zaydun.
2
To be sure, there were grammarians who analyzed both constructions as an inverted
jumla ismiyya with a fronted xabar ; others accepted more than one type of analysis. For
a discussion, cf. Ibn Ab r-Rab , Bast I:583ff.
the medieval arabic theory of sentence types 151
2. qimun zaydun
In his bb al-ibtid, Sbawayhi (Kitb I:239) discusses the option of
mubtada-xabar inversion . His starting point is that the standard pre-
ferred (al-h add) structure is for the mubtada to precede the xabar rather
than the reverse, much as the standard word order in the verbal sentence
is for the fil to precede the maf l . When dealing with inversion , he sin-
gles out qimun zaydun as a markedly complex case deserving special
attention. He quotes his teacher al-Xall as saying that qimun zaydun is
an ill-formed (qabh ) sentence unless analyzed as the inverted version of
zaydun qimun. As is well known, some of the later grammarians made
the point that such an inversion is quite problematic, since it places the
maml before the mil , and the Kfans saw a further problem in that it
makes the pronoun implicit in the participial form qimun precede its
antecedent (al-idmr qabla d-d ikrsee, e.g. Ibn al-Anbr , Ins f I:65).
But these problems were easily dismissed by the claim that qimun zay-
dun represents a secondary ( far) or surface (lafz ) structure , whereas in
the basic structure (man , niyya , taqdr ) zaydun, the z hir and mil
precedes qimun, the maml , with the implicit pronoun (the mudmar )
referring backas requiredto zaydun (for a detailed discussion, see,
e.g. Ibn al-Anbr , Ins f I:6566, 68).
However, the real problem with qimun zaydun was associated with a
diferentverbalanalysis of the participle , known to have been advo-
cated by some of Sbawayhi s contemporaries. Citing al-Xall , Sbawayhi
points out that:
fa-id lam yurd hd l-man wa-ard an yajalhu flan ka-qawlihi
yaqmu zaydun wa-qma zaydun qabuh a li-annahu smun
If, however, they do not accept this analysis [= inversion ], and want to
treat [qimun] as a verb , in analogy to such sentences as yaqmu zaydun
and qma zaydun, this should be rejected, because [qimun] is a noun
(Sbawayhi , Kitb I:239).
Yet Sbawayhi immediately makes it clear that under certain conditions
an active participle , while categorized as a noun, may implement a verbal
function (yajr majr l-fl ). Tis could be accepted (h asuna indahum),
he maintains, if the participle functions as part of an asyndetic relative
clause (s ifa ) linked to some antecedent (maws f ), or, otherwise, governed
by a preceding operator such as a mubtada . In other words, qimun
zaydun is disallowed with a verbal analysis , much as dribun zaydan
(hitting Zayd) is unacceptable as a complete sentence. However,
152 yishai peled
qimun zaydun, as part of a larger sentence, is acceptable with a verbal
analysis, in analogy to such cases as an dribun zaydan (I am hit-
ting Zayd).
3
As we shall see shortly, Sbawayhis approach to the verbal
analysis of qimun zaydun was later established as a frm principle in
medieval Arab grammatical thinking .
Ibn as-Sarrj (Us l I:5960), whose position regarding mubtada-
xabar inversion seems to be similar to that of Sbawayhi s, readily accepts
munt aliqun zaydun as an inverted version of zaydun munt aliqun. As
for analyzing qimun zaydun as analogous to yaqmu zaydun where
qimun is not preceded by any supporting element (see below), in
other words, construing zaydun as fil to munt aliqunin principle, Ibn
as-Sarrj , much like Sbawayhi, regards such an analysis as misguided
(qabuh a), yet he admits it as jiz (acceptable). What both grammar-
ians seem to accept without reservation is that the noun following the
participle may be analyzed as a kind of fil provided that it is anchored
(yatamidu al, in Ibn as-Sarrjs words) to some preceding constitu-
ent. As an illustration of how this condition can be met, Ibn as-Sarrj
adduces such sentences as: marartu bi-rajulin qimin abhu (I passed
by a man whose father was standing), zaydun qimun abhu (Zayd
his father is standing), a-qimun abka (is your father standing?).
Tis rule would be developed by later grammarians into a general
principle of itimd , designed to specify the conditions under which a
non-verbal predicate may be analyzed as analogous to a verb preced-
ing its subject .
4
Tis principle stipulates that a non-verbal predicatival
constituent , such as an active participle , or an adverbial /prepositional
phrase (see section 3 below) may exercise amal upon the constituent
following it (the subject ) only if supported by (yatamidu al) some ele-
ment such as an interrogative particle , a relative pronoun , or, otherwise,
when the clause as a whole functions as an asyndetic relative clause or as
xabar to a preceding mubtada . In such cases, the frst predicatival con-
stituent is perceived as behaving analogously to a verb. And as a verb-
like constituent it acts as a mil , assigning the raf case to the following
3
Sbawayhi (Kitb I:239) asserts that while the active participle and the verb may be
similar in some respects, one must appreciate the difference between them. Other gram-
marians (e.g. Ibn al-Anbr , Asrr, 70) pointed out that the active participle is weaker
than the verb, and cannot, therefore, exercise verbal amal , unless supported by some
preceding element (see below).
4
For the concept itimd as it is used in al-Xall s Kitb al-Ayn with reference to other
grammatical structures , see Talmon 1997, 210.
the medieval arabic theory of sentence types 153
nominal constituent .
5
For discussion, see, e.g. Ibn al-Anbr , Asrr: 70;
cf. Goldenberg 2002, 199201.
Ibn as-Sarrj is aware, however, of the implications of a verbal analy-
sis of qimun zaydun for the theory of amal . He argues (Us l I:60) that
in qimun abka (Your father is standing), qimun is assigned the
raf case by the ibtid , and abka is assigned the same case by the verb
preceding it. He indicates further that abka flls a xabar position.
In any event, both Sbawayhi and Ibn as-Sarrj reject the use of dribun
bakran amrun ( Amr hits Bakr) as an independent sentence, on the
ground that the active participle , while being analogous to the verb , is
by defnition a nominal , and as such cannot be made to function identi-
cally to a verb in terms of case assignment. Te above examples, where
the participle is linked to a preceding antecedent (maws f ), a mubtada ,
or an interrogative particle , are viewed as analogous to the construc-
tion dribun bakran when anchored, under the principle of itimd , to
some external constituent (mah ml al ayrihi), such as a mubtada,
thus presenting a well-formed independent sentence (e.g. hd dribun
bakranthis [person] is hitting Bakr) (Ibn as-Sarrj , Us l I:60; and cf.
Sbawayhis position above; Levin 1985, 125126).
Like Sbawayhi and Ibn as-Sarrj , as-Zajjj (Jumal, 3738) was aware
of the theoretical problems raised by sentences consisting of an active
participle followed by a noun phrase . In particular he demonstrated the
implications for the grammatical agreement between the two constitu-
ents. To the extent that qimun in qimun zaydun is conceived of as
xabar muqaddam (a fronted xabar ), it must be replaced by qimni or
qimna, once zaydun is substituted by a dual or a plural form respec-
tively. But under the alternative analysis cited by as-Zajjj , in which
qimun is assigned a verbal function , the active participle preceding
its subject should invariably take the singular form . In other words, the
proponents of this analysis would have qimun az-zaydni/az-zaydna
rather than qimni z-zaydni and qimna z-zaydna.
6
5
Ibn Ab r-Rab (Bast I:585) remarks that some grammarians rejected the idea of
an adverbial /prepositional phrase assigning case. They argued that such phrases were
different in status (manzila ) from the adjective . The latter, they argued, is capable of
inflection, and as such is more powerful than the adverbial /prepositional phrase . There-
fore, they concluded, the adverbial /prepositional phrase may not be analyzed as a case
assigner even where the principle of itimd is met. I return to this issue later.
6
See, e.g. Ibn Ab r-Rab (Bast I:584), who also indicates that the proponents of
akaln l-bart must, by extension, say qimni z-zaydni and qimna z-zaydna,
154 yishai peled
Tis rule, as stated by as-Zajjj , highlights the verbal function of the
participle under this particular analysis, relating qimun az-zaydni/az-
zaydna to qma z-zaydni/z-zaydna. It should be noted that the medi-
eval grammarians treated participles , as well as other types of adjective ,
as complex forms incorporating a personal pronoun (for discussion, see
Goldenberg 2002, 195). Ibn Ya (arh I:8788) states clearly that parti-
ciples and other adjectives are derived from the verb , and that in virtue
of having a verbal meaning (man fl ) they must have a fil .
7
How-
ever, construing munt aliqun zaydun as modeled on yant aliqu zaydun
implies neutralizing the pronominal element in munt aliqun exactly as
it is done in yant aliqu.
8
Te argument is that the participle , much like a
regular fl , cannot assign the raf case twice (l yarfau filayni).
9
Zajjj (Jumal, 38) indicates that in such cases the active participle
introducing the sentence is assigned raf by the ibtid , whereas the con-
stituent following it is assigned the same case by its verb (bi-flihi
apparently referring to the active participle ; cf. Ibn as-Sarrj s analysis
above). Zaydun in qimun zaydun, it is argued, replaces the xabar (ya-
suddu masadd al-xabar); and the participle preceding it, he points out,
is invariably singular li-annahu qad jar majr l-fli l-muqaddami (for
it behaves analogously to a verb preceding [its subject ]).
10
Observe that, unlike the vast majority of grammarians , Zajjj did not
make the point of linking the verbal analysis of qimun zaydun to an
obligatory application of the principle of itimd . It should, indeed, be
noted that some of the later grammarians held a narrower version of
this principle, restricting the use of a participle in sentence-initial posi-
tion to cases where the participle is attached to a negative or interroga-
since in this version of the language the verb preceding the subject agrees with it in
number and gender (for a detailed discussion of akaln l-bart , see Levin 1989).
7
The grammarians , however, recognized that the personal pronoun incorporated
in an active participle cannot qualify as fil in the way an implicit personal pronoun
in a verb can. Thus, while allad daraba zaydun (The one who hit is Zayd) is a per-
fectly grammatical sentence, allad dribun zaydun is not, since, unlike allad daraba,
allad dribun cannot implement the function of a subject clause (see Jurjn , Muqtas id
I:463464).
8
This is, perhaps, why the Kfans , who rejected the analysis of qimun zaydun as an
inverted nominal sentence , could accept it as modeled on a verbal sentence : under the
verbal analysis the pronoun in qimun is disabled so there is no problem of cataphora
(cf. above).
9
For Ibn Ya (arh I:8788), then, a sentence such as zaydun qimun abhu
(Zayd, his father is standing) consists of a mubtada (zayd) and a xabar , the latter ana-
lyzed as a complex construction consisting of a fil (qimun) and a fil (abhu).
10
This type of analysis is attributed to Axfa ; see, e.g. Ibn Us fr , arh I:341.
the medieval arabic theory of sentence types 155
tive particle : m qimun az-zaydni (the two Zayds are not standing),
a-qimun az-zaydni (Are the two Zayds standing?). According to Ibn
Aql (d. 1367) (arh I:189), qimun, in each of the last two sentences,
functions as a mubtada, whereas az-zaydni is a fil sadda masadd al-
xabar (a fil substituting for the xabar). Indeed, this is the common
formula employed by those later grammarians who adopted the verbal
analysis of the construction in question (cf. Carter 1981, 189).
At this point one might ask how frequent in classical Arabic are such
sentences as qimun az-zaydni? I looked for this construction in the
Qurnic text , but no example of it was attested. In all the recorded cases,
a singular participle is followed by a singular noun phrase , or, otherwise,
a singular feminine participle by a plural (non-human) noun phrase .
It is interesting to note, however, that all cases display some kind of a
supporting element. In the vast majority of examples, the construc-
tion in question functions as predicate to a preceding subject realized
as either a referential nominal (typically, but not necessarily, in a sen-
tence introduced by inna or one of its sisters), or, otherwise, as a non-
referential damr a-an: wa-z ann annahum mniatuhum h us nuhum
(Tey believed that their fortresses would protect themQ. 59:2),
wa-huwa muh arramun alaykum ixrjuhum (You are not allowed to
expel themQ. 2:85). Huwa in the latter example functions as damr
a-an. One example was noted where the supporting element is the
interrogative particle a-: a-ribun anta an lihat (do you loathe my
gods?Q. 19:46).
What is then the efect of the supporting element that makes [zay-
dun/rajulun] qimun abhu or a-/m qimun abhu an acceptable
verbal construction , as opposed to qimun abhu? Al-Xall (see above)
does not provide an elaborate answer. He argues, however, that a partic-
iple cannot easily replace a verb in pre-subject position , because it is an
ism . A verb and a noun , he maintains, may in certain positions imple-
ment similar functions, but they must still be diferentiated. Nor did later
grammarians elaborate on the function of the supporting element. But
their discussion of the relevant cases might give us a clue. To phrase the
question diferently, how does the supporting element impart further
verbal force to the adjectival predicate that enables it to act analogously
to a verb in such cases? If we compare the two constructions qma zay-
dun and qimun zaydun, we can see that the diference between the
two is that the fnite verb , while devoid of a pronominal element , is still
infected for person, whereas the participle is not. Lacking either a pro-
nominal element or infection for person, the participle is excluded as a
156 yishai peled
pre-subject verbal predicate in an independent sentence. Te function
of the supporting element zaydun in zaydun qimun abhu, is to make
up for the lack of person infection in qimun, and thus empower the
latter to implement a verbal function .
11
Further, as has been indicated, the supportive function may be
implemented not only by a noun but also by an interrogative or a nega-
tive particle . Tese are, in other words, further sources from which the
adjective could derive a verbal force . Already Sbawayhi attributed a
verbal efect to certain interrogative particles (Sbawayhi, Kitb I:39f.,
41f.the latter dealing specifcally with a-). Ibn Ab r-Rab (Bast
II:712) points out that the interrogative particle a- requires a verb (al-
hamza t libatun bi-l-fl ), and that underlying (taqdr ) a-zaydun daraba
amran is a-daraba zaydun amran (Did Zayd hit Amr?) (and cf. his
similar attitude to halII:679). Regarding the construction at issue he
maintains that underlying a-qimun zaydun is a-yaqmu qimun
zaydun, and that the former is derived from the latter by suppressing the
redundant yaqmu, for which qimun serves as an exponent (tafsr) .
Tis explains why the interrogative a- qualifed in the grammarians
view as a supporting element in sentences consisting of an adjectival
predicate followed by a subject .
However, the verbal analysis of such constructions as (a-)qimun
zaydun raises a difcult problem for the medieval theory of sentence
types . For if a sentence consists of a mubtada followed by a fil , how
should it be categorized in terms of sentence types ? Te fact that the
fil is presented as replacing the xabar does not make the issue any
simpler. For under the suggested analysis, a verb-acting constituent is
considered, in terms of irb , as a mubtada assigned the raf case by
the ibtid . Te conception of the participle in such cases as mubtada is
quite understandable, given that it is a nominatival constituent in sen-
tence-initial position . If one declines the inversion analysis of sentences
such as qimun zaydun, how else can one account for the raf case of
qimun? Te main problem with this analysis lies in its stipulating that
qimun as a mubtada assigns raf to a fil occupying a xabar position.
11
That, I believe, is what is intended by al-Xall (Sbawayhi , Kitb I:239) when he
refers to the participle in such cases as [kna] s ifatan jar al maws fin aw jar al
smin qad amila fhi [the participle] is an adjective agreeing with a head or [other-
wise] with a noun acting upon it, and further when he says: l yaknu maf lan f
dribin h att yaknu mah mlan al ayrihi fa-taqlu hd dribun zaydan . . . drib
cannot take an object unless it is linked to some other constituent, as for example in
hd dribun zaydanthis [person] is hitting Zayd.
the medieval arabic theory of sentence types 157
Once we correlate qimun zaydun with qma zaydun, rather than with
zaydun qimun, with the consequence that qimun is to remain in
the singular irrespective of the number of the following noun (qimun az-
zaydni/az-zaydna), one can hardly see how sentences such as qimun
zaydun under a verbal analysis , let alone qimun az-zaydni, may be
viewed other than as cases of jumla fliyya . Indeed, when presenting the
fil , some of the grammarians , like Ibn Ya (arh I:74), indicate that
the position preceding the fil is available for a verb ( fl ) or a nominal that
is analogous to a verb (abahuhu, m huwa f man l-fl min al-asm).
In this latter category they normally include the active and passive par-
ticiples , as well as such adjectives as h asan (s ifa muabbaha bi-smi l-
filan adjective analogous to the active participle , e.g. Ibn Ya , arh
I:87). It is argued that in a sentence such as zaydun dribun ulmuhu
12
(Zayd, his slave is hitting), dribun, much like yadribu, assigns raf to
ulmuhu. One may infer, then, that Ibn Ya would regard a sentence
such as qimun zaydun as a verbal sentence . Yet I have not recorded
any explicit reference to this type of sentence as a jumla fliyya . In any
case, the prevalent analysis of the construction under discussion was, as
already indicated, mubtada +fil sadda masadd al-xabar . It is not sur-
prising, however, that the grammarians adhering to this analysis did not
commit themselves to explicitly categorizing such sentences as either
jumla fliyya or jumla ismiyya .
A remarkable exception is Ibn Him al-Ans r (d. 1360) who, in his
famous book Mun l-Labb, provides an elaborate discussion of Arabic
sentence types . Ibn Hims classifcation will be discussed in detail in
section 5 below. As we shall see, he defned three sentence types (rather
than two!) by the kind of constituent introducing the sentence. Tus,
a sentence introduced by a nominal element is a jumla ismiyya . And
among his examples of jumla ismiyya we fnd the sentence qimun az-
zaydni. Ibn Him was, indeed, aware of the controversy surrounding
this sentence, indicating that it was accepted as a well-formed sentence
by Axfa and the Kfans . As we saw above, qimun az-zaydni was
adduced as an acceptable sentence in Arabic also by Zajjj , but the latter
did not classify it as jumla ismiyya.
12
Note, however, that in Zamaxar s and Ibn Yas examples the construction at
issue is itself a xabar following a mubtada . As we have seen, this is consistent with the
principle of itimd .
158 yishai peled
In sum, then, the analysis of qimun zaydun into a mubtada +fil
replacing a xabar appears to refect a twofold attempt:
(1) to make this kind of sentence conform to the principle of ibtid ;
this is motivated by the fact that the frst constituent is a nominal
exhibiting a raf case ending , and;
(2) to apply the principle of verbal tadiya , given the verbal properties
of the active participle .
But, to the extent that this analysis holds, does not it follow (at
least from the grammarians viewpoint) that qimun zaydun rep-
resents a sentence type in its own right? To my knowledge, no such
proposition has ever been advanced in medieval Arab grammatical
literature .
13
3. fh / f d-dri zaydun
Only a small minority of the grammarians suggested that sentences
such as fh zaydun (In it there is Zayd) should be regarded as repre-
senting a sentence type in its own right. Tey designated this type jumla
z arfyya , but difered on whether this term should or should not cover
also sentences such as zaydun fh/f d-dr (Zayd is in it / in the house).
I return to this later. At this point, let us examine the grammarians con-
ception of this construction, starting with Sbawayhi .
3.1 Sbawayhi
In his bb al-ibtid (chapter 132), Sbawayhi does not develop any dis-
cussion of this sentence sub-type. But elsewhere in the Kitb (I:170171;
cf. Levin 1987, 362, and Owens 1989, 224) he argues that in cases such as
huwa xalfaka (He is behind you) it is the subject huwa that assigns the
nas b case to xalfaka. Indeed, this is consistent with his argument (Kitb
13
Badawi (2000, 8f.) claims that the grammarians recognized three types of sentences
namely: fliyya , ismiyya and was fyya , introduced, respectively, by a verb , a noun and an
adjective (a participle or otherwise). He emphasizes the use of diferent terms for the
subject and predicate in each sentence type , indicating that in the jumlat was f these are
referred to as mubtada and fil sadda masadd al-xabar . However, the term jumlat was f
has not been attested in the grammarians writings studied for the present work.
the medieval arabic theory of sentence types 159
I:239) that in zaydun munt aliqun, it is the subject that assigns raf to the
predicate .
14
However, what about sentences such as fh zaydun, con-
sisting of an adverbial /prepositional phrase followed by a nominatival
noun phrase ? Sbawayhi deals with these cases within the framework of
his discussion of sentences such as fh abdu-llhi qiman and abdu-
llhi fh qiman (Abdullh is in it, standingcf. Talmon 1993, 281).
He starts his discussion analyzing qiman as an accusatival xabar to
abdu-llhi. Ten he goes on to indicate that abdu-llhi in these cases:
irtafaa bi-l-ibtidi li-anna llad d ukira qablahu wa-badahu laysa bihi
wa-innam huwa mawdiun lahu wa-lkinnahu yajr majr l-ismi l-mab-
niyyi al m qablahu
is assigned the raf case by the ibtid since the constituent occurring
either before or afer it [= the adverbial ] is not it [= is not identical in refer-
ence with it], but rather signals its location. Yet [this adverbial] functions
analogously to a noun built upon the [subject ] preceding it(Sbawayhi ,
Kitb I:222).
Sbawayhi , as can be seen, points out that fh is non-coreferential with
abdu-llhi, but rather refers to Abdullhs location (mawdi). But it is
precisely this observation that underlies his endeavor to establish, frst
of all, the acceptability of fh abdu-llhi/abdu-llhi fh as a complete
independent sentence. To this end he draws an analogy between fh
abdu-llhi and hd abdu-llhi (Tis is Abdullh), claiming that
in terms of completeness, the former, much like the latter, is a kalm
mustaqm (a correct sentence) that h asuna [badahu] s-suktu (that
may appropriately be followed by silenceSbawayhi, Kitb I:222 and
239240). Similarly, abdu-llhi fh is presented by him as analogous
to abdu-llhi axka (Abdullh is your brother): in both cases the
second constituent is built (mabn) upon the frst. As for abdu-llhi,
Sbawayhi states clearly that, whether preceding or following the prepo-
sitional phrase , it is assigned the raf case by the ibtid .
But once a predicatival relationship is established between fh and
abdu-llhi, Sbawayhi reanalyzes the sentence assigning fh the func-
tion of xabar and abdu-llhi the function of mubtada . Te position of
qiman is then demoted to that of h l (see Figure 1 below). He points
out, however, that since fh represents the persons location, fh abdu-
llhi is paraphrasable by istaqarra abdu-llhi. In other words, fh
14
Note, however, that the amal in huwa xalfaka is presented by Sbawayhi as analo-
gous to that in his model construction irna dirhaman (Twenty dirhams).
160 yishai peled
behaves analogously to the verb istaqarra. As we shall see in 3.2, the verb
istaqarra, or otherwise the participle mustaqirrun, have since become
the grammarians most common device for explaining the grammatical
structure of sentences such as zaydun fh / fh zaydun.
A further indication of Sbawayhi s consideration of fh as a verb-like
element is his statement (Kitb I:223) that qiman in the above sen-
tence may, alternatively, be replaced by qimun in the nominative . Tis,
he explains, is the result of abrogating (alayta ) fh. In the medieval
grammarians writings, ilg is normally used as a technical term denot-
ing the annulling of amal . It is typically used with reference to potential
awmil , that is, elements that normally exercise amal upon other ele-
ments in the sentence (for discussion see, e.g. Peled 1992a, 150152).
One may infer, then, that Sbawayhi considered fh, in virtue of its act-
ing analogously to istaqarra , as an mil assigning nas b to qiman in
fh abdu-llhi qiman. As we shall see later, such adverbials as fh
were considered by some early grammarians as an mil assigning raf
to the following subject in such cases as fh zaydun. Tis view is typi-
cally attributed to the Kfans . Yet, Sbawayhi (Kitb I:223224) then
enters into an extensive discussion designed to exclude the possibility
that fh in sentences such as fh abdu-llhi qimun is the mil assign-
ing raf to abdu-llhi. He draws an analogy between this sentence and
bika abdu-llhi maxd un (Abdullh is fascinated by you). He argues
that an operator assigning case to an optional constituent (qimun in
the former sentence) has the same status (manzila ) as an operator act-
ing upon an obligatory constituent (maxd un in the latter).
15
Sbawayhi
emphasizes that in both cases (as well as in similar ones adduced by
him) the adjective is built upon the noun , thus establishing a predicati-
val relationship between the two. Te prepositional phrase , by contrast,
is a law , i.e. a constituent that neither assigns nor receives amal . In fh
15
Indeed, in later grammatical writings , the model sentence bika zaydun maxd un
features regularly in the Bas ran arguments against the Kfan claim that in fh zaydun
it is fh that functions as the mil assigning raf to zaydun (cf. Ibn al-Anbr , Ins f
I:5253).
mubtada xabar
fh abdu-llhi qiman
xabar mubtada h l
Figure 1
the medieval arabic theory of sentence types 161
abdu-llhi qimun, fh is only designed to specify the location where
the standing is taking place.
Let us now return to the original construction fh abdu-llhi to
which Sbawayhi devotes a separate bb, following his bb al-ibtid. In
chapter 133 of the Kitb, he deals with such cases as fh abdu-llhi,
t amma zaydun and ayna zaydun. Compared to his discussion above,
Sbawayhi here seems to be less specifc about the rf of zaydun:
wa-llad amila fm badahu h att rafaahu huwa llad amila fhi h na
16
kna qablahu
Te operator assigning raf to the following constituent is the same oper-
ator that assigned it the raf case when that constituent was before it [i.e.
before fh](Sbawayhi , Kitb I:239; cf. editors notes as well as the edi-
tions of Blq and Hrn).
17
Sbawayhi thus asserts that the operator assigning raf to zaydun in
fh zaydun is the same one that assigns the raf case to zayd in zaydun
fh, where/when zayd occurs before it (i.e. before fh). Apparently,
he leaves it to the reader to conclude that it is the ibtid that functions
as mil in both cases. Tis could lead one to believe that for Sbawayhi,
fh zaydun represents an inverted version of zaydun fh. Yet Sbawayhi
does not make any explicit claim for taqdm wa-taxr in this particular
case. He presents fh zaydun as a case in which fh:
yaqau mawqia l-ismi l-mubtadai wa-yasuddu masaddahu
occupies the position of the mubtada and replaces it(Sbawayhi , Kitb
I:239; quoted also in Kouloughli 2002, 9).
As we saw in section 2, there were among later grammarians those who
analyzed qimun in qimun zaydun as a mubtada followed by a fil
(replacing a xabar ). It seems more than likely that they were infuenced
by the above Sbawayhian passage.
Regarding ayna zaydun, Sbawayhi points out that ayna is paraphras-
able by f ayyi makn (in which place?), and emphasizes that ayna,
as an interrogative , must obligatorily occur sentence-initially. In other
words, fh zaydun is distinct from ayna zaydun only in that in the latter
case the xabar occupies sentence-initial position obligatorily.
16
The words h ayt u and h na alternate in this position in two different versions of
the text.
17
Talmon (1993, 283284) confronts the long version cited here with a shorter one
that to me looks rather obscure.
162 yishai peled
It is interesting to note that Sbawayhi does not include in his discus-
sion such cases as fh rajulun (In it there is a man), where the subject ,
being indefnite , obligatorily follows the predicate (much like in ayna
zaydun). Indeed, as we will see later, this construction received little
attention from the grammarians , compared to fh zaydun. As we shall
see, while the grammarians never failed to point out that the predicate-
subject order in fh rajulun is obligatory, only a small minority of them
regarded this construction as representing a sentence type in its own
right. For the vast majority, fh rajulun, much like fh zaydun, repre-
sented an inverted jumla ismiyya . A detailed discussion of constructions
with an obligatorily fronted xabar is provided in section 4 below.
3.2 Te istaqarra /mustaqirrun hypothesis
Following Sbawayhi , the medieval grammarians continued to address
themselves to the twofold problem posed by such sentences as zaydun
fh and fh zaydun. Tis problem, as we have seen, consisted in estab-
lishing a predicatival relationship between the nominal and the adverb-
ial /prepositional phrase , and accounting for the mil assigning case to
each. Te suggested solution of positing an underlying linking element
such as istaqarra /mustaqirrun gave rise to extensive discussions that
developed into what may be referred to as the istaqarra/mustaqirrun
hypothesis . As we saw in 3.1, the origins of this theory can be easily
traced back to Sbawayhis Kitb.
18
Troughout the centuries, the istaqarra /mustaqirrun hypothesis
has become the common strategy used by the medieval grammarians
in order to account for such sentences as zaydun f d-dr (or, for that
matter, zaydun xalfaka, where xalfaka alternates with min xalfkafor
discussion of the status of the z arf , see Levin 1987, 351357).
19
As for
the grammatical status of the adverbial following the mubtada , here the
grammarians difered. For Sbawayhi (Kitb I:222), munt aliqun and fh
were equally admissible as xabar for zaydun. In Ibn as-Sarrj s (Us l
I:6263) view, the xabar in zaydun f d-dr is the underlying mustaqir-
18
As we have seen, however, Sbawayhi used this device in dealing with sentences
such as fh abdu-llhi qiman, to account for the nas b of qiman; he did not employ
it in cases such as fh zaydun or zaydun fh, pointing rather to the ibtid , in both, as
the mil assigning raf to zaydun.
19
For the Kfan theory of xilf (or muxlafa), see Astarbd , arh I:214; Ibn al-
Anbr , Ins f I:245247; Muji , arh , 87, n. 216; and cf. Carter 1973).
the medieval arabic theory of sentence types 163
run which, when coreferring with the mubtada, is optionally omitted
(h ad f ). Te adverbial f d-dr flls the xabars slot, acting as a comple-
ment to the omitted genuine xabar (for a similar view, see Ibn Us fr ,
arh I:347). Zaydun xalfaka/f d-dr is thus paraphrasable by zaydun
mustaqirrun xalfaka/f d-dr. Ibn as-Sarrj points out that the omitted
element normally conveys some meaning of existence, and is redundant
because it is retrievable from the adverbial.
20
He emphasizes that postu-
lating an underlying constituent in such cases is obligatory, because, in
itself, f d-dr (or xalfaka) does not qualify as a predicate ; that is, it does
not predicate any quality of zayd (laysa bi-h adt ),
21
but only specifes the
location (mawdi) of Zayd.
Jurjn (Muqtas id I:274275) argues that an adverbial , with either
an explicit or implicit preposition , in principle presupposes a verb with
which it is linked to form a syntactical unit (cf. Ibn al-Anbr , Ins f
I:246). Consequently, underlying f d-dr, in zaydun f d-dr, is the verbal
clause istaqarra f d-dr. Tis indeed represents the view of the major-
ity of grammarians for whom positing an underlying (muqaddar) fnite
verb (istaqarra) is consistent with the clausal status of f d-dr in sen-
tences such as zaydun f d-dr. Obviously, once established as the (head
of the) xabar , istaqarra/yastaqirru is referred to as the mil , assigning
case to the following adverbial /prepositional phrase .
However, the assumption of an underlying fnite verb in cases such
as zaydun f d-dr was not universally accepted. Te controversy here is
linked to the fact that not all the grammarians regarded f d-dr in zaydun
f d-dr as having a clausal status. Some grammarians posited a parti-
ciple rather than a verb as the underlying element linking the adverbial /
prepositional phrase to the preceding mubtada . We have just seen that
Ibn as-Sarrj was one of the proponents of this hypothesis; indeed, he
20
Ibn Us fr (arh I:347348) emphasizes that using an adverbial /prepositional
phrase as a xabar substitute is only admissible when the deleted element is recoverable
from the surface constructionotherwise, the xabar should appear in full. Thus, for
example, zaydun f d-dr is only allowed if it is intended to convey the meaning mus-
taqirrun f d-dr, because f, signalling a receptacle (wi ), is compatible in meaning
with istiqrr (staying). If, however, zaydun f d-dr is intended to convey the mean-
ing of dh ikun f d-dr ([Zayd] is laughing in the house), then the word dh ik must
occur; for, unlike the meaning of staying, that of laughing cannot be recovered from
the preposition f. Cf. Astarbd , arh I:215, for linking elements like h s il and kin
(be); Levin 1987, 360.
21
Ibn as-Sarrj s use of the term h adt in this case is signifcant, for it signalizes predi-
cate realized whether as fl or as xabar (cf. Goldenberg 1988, 4649).
164 yishai peled
viewed sentences such as zaydun xalfaka as displaying a single-phrased-
rather than a clausal xabar (Ibn as-Sarrj, Us l I:63). Astarbd (arh
I:215) cites Ibn as-Sarrj and Ibn Jinn as two grammarians who advo-
cated the participle rather than the verb hypothesis, on the ground that
the participle as a single phrase (mufrad ) is compatible with the basic
structure of the xabar. Another proponent of the participle hypothesis
is Muji (arh , 87) who derives such sentences as zaydun ammaka
(Zayd is in front of you) and amrun min al-kirm ( Amr is one of the
honorable) from the underlying (taqdr ) structures zaydun mustaqir-
run ammaka and amrun kinun min al-kirm respectively.
22
Muji
makes it clear that for him an adverbial /prepositional phrase in xabar
position has the status of, and is therefore a substitute for, a participle
(not a clause ). Postulating a personal pronoun implicit in the participle,
Muji argues further that this pronoun moves to, and resides in, the
adverbial /prepositional xabar occupying the position of the deleted par-
ticiple (in Mujis words: wa-afd d-damru llad kna f smi l-fil il
n-nib anhu fa-statara fhi).
23
For further discussion of the istaqarra /
mustaqirrun hypothesis , see Jurjn , Muqtas id I:275f; Ibn al-Anbr ,
Ins f I:245247; Ibn Us fr , arh I:344, 349351.
We can see that the istaqarra /mustaqirrun hypothesis is used by the
grammarians to ft the construction zaydun fh/fh zaydun into their
theory of amal , and, by implication, to their binary system of sentence
types . Once zaydun in both zaydun fh and fh zaydun was recog-
nized as a mubtada , both constructions could be said to represent a
jumla ismiyya , the latter being an inverted version of the former. Te
xabar , when following the mubtada, is presented as either clausal or
phrasal, depending on whether one assumes yastaqirru or mustaqirrun
to be the underlying linking element. In both cases, this element is made
accountable for the case of f d-dr (see Kouloughli 2002, 1316, for
further discussion). However, the istaqarra/mustaqirrun hypothesis was
not universally accepted. And as we shall see, the alternative hypotheses
had substantial implications for the theory of sentence types.
22
Astarbd (arh I:215) claims that the underlying element is obligatorily deleted,
rejecting such sentences as zaydun kinun f d-dr. He indicates that Ibn Jinn did allow
such constructions, but points out that there is no evidence to support this position.
23
This is evidently Mujiis way of claiming a xabar status for the adverbial /prepo-
sitional phrase . Astarbd (arh I:216217) points to Fris and his followers as advo-
cating the same hypothesis. But Srf is cited by him as claiming that the pronoun is
deleted as part of the linking constituent.
the medieval arabic theory of sentence types 165
3.3 Ab Al l-Fris
Ab Al l-Fris (d. 987) is considered to be one of the frst grammar-
ians who advanced in an explicit way and developed the idea of sentence
types in Arabic (cf. Owens 1988, 3637). He defned each type, and
spelled out the problematic nature of the dichotomy verb+noun ( jumla
fliyya ) versus noun+noun (jumla ismiyya ).
24
Indeed, he was the frst to
present a detailed argument with the conclusion that zaydun f d-dr is
neither a jumla fliyya nor a jumla ismiyya .
Having defned the two basic sentence types in Arabic , Fris turns
to concentrate upon the construction represented by zaydun f d-dr.
Indeed it looks as though Friss defnition of the two basic sentence
types is meant as an introduction to his discussion of this particular
construction (Fris, Askariyya, 105109). He starts by indicating that,
although such sentences are composed of a nominal element (i.e. the
two nouns ) and a particle (the preposition ), they do not have the same
status as inna sentences , where the particle enters into a sentence made
up of two nouns. Tis is because f d-dr is non-coreferential with zay-
dun. And since zayd and f d-dr are not identical in reference, they
cannot be analyzed simply as subject and predicate . However, given that
zaydun f d-dr is defnitely a well-formed sentence in Arabic, one must
assume some underlying (muqaddar) linking element to account for the
predicatival relationship between its two constituents. As we saw in 3.2,
this linking element must inevitably be either a noun or a verb (a par-
ticle does not bear any reference). To the extent that either of these can
be posited, a sentence such as zaydun f d-dr must eventually belong
either to the verb+noun or to the noun+noun type .
25
24
Anxious to provide accurate and valid defnitions, Fris (Askariyya, 104105)
points further to the option of a particle (h arf ) entering into either of the two defned
jumlas, to form a kalm. What the reader is invited to infer is that the resulting con-
struction is an independent grammatical sentence whose basic type (i.e. fliyya or
ismiyya ) is unafected. He exemplifes this by sentences introduced by hal, inna, m,
qad and lam. (As a matter of fact, the same principle had already been stated by Ibn
as-Sarrj , Us l I:43.)
25
Fris (Askariyya, 109) draws a comparison between the case in question and
address (nid) expressions. He argues that y zaydu (O, Zayd!), much like f d-dri
zaydun, consists of nominal elements and a particle, and constitutes an independent
sentence. The difference between the two, he maintains, is that in the case of y zaydu a
verbal element should be assumed, which renders the address expression a sub-type of
a jumla filiyya , whereas in the case of zaydun f d-dr/f d-dri zaydun no such element
can be posited (see below).
166 yishai peled
However, from this point onward, Fris s argument continues in a
direction designed to prove that this is not the case. In other words, a
sentence such as zaydun f d-dr, while conforming to the general prin-
ciple governing the production of well-formed sentences, constitutes an
exception in that it does not fall into any of the two basic categories,
jumla fliyya and jumla ismiyya . In Friss words:
a-l tar anna l-kalma wa-in kna l yaxl mimm d akarn f l-as li fa-
qad s ra lahu l-na h ukmun yaxruju bihi an d lika l-as li
Notice that although a sentence must obey the basic principles indicated
by us, in this case there is an [overriding] rule leading the sentence away
from the basic principles (Fris , Askariyya, 105).
What Fris is now trying to prove is that neither a verb nor a noun can
be posited as a linking element establishing a predicatival relationship
between zaydun and f d-dr, a relationship modeled on that obtain-
ing between the subject and predicate of a regular verbal or nominal
sentence . And if it can be proved that neither a verb nor a noun can be
posited as an underlying linking element between zaydun and f d-dr,
the conclusion must be that sentences such as zaydun f d-dr represent
a sentence type in its own right.
He starts (Fris , Askariyya, 105) by adducing the sentence inna f
d-dri zaydan, were the particle inna enters into the sentence f d-dri
zaydun (his choice of this construction rather than zaydun f d-dr is
signifcant, as will be seen below.) Ten he makes the following two
points:
(1) An underlying verbal link cannot be assumed, because a verb would
exclude the use of inna. In other words, while f d-dri zaydun may
be preceded by yastaqirru, inna and yastaqirru are mutually exclu-
sive: inna f d-dri zaydan is a perfectly grammatical sentence in
Arabic , but *inna yastaqirru f d-dri zaydan is disallowed.
(2) A linking noun cannot be posited either, because that would
amount to assumingfalselythat inna exercises its efect (amal )
upon zaydan across the underlying linking noun (Fris , Askariyya,
108).
Having disqualifed both noun and verb as possible underlying linking
elements in cases such as zaydun f d-dr and f d-dri zaydun, Fris
(Askariyya, 108) argues further that in such cases the adverbial con-
stituent as such cannot be claimed to implement a verbal function . Tis,
the medieval arabic theory of sentence types 167
he maintains, is borne out by the fact that the adverbial may not be pre-
ceded by a circumstantial phrase; a sentence such as *qiman f d-dri
zaydun is inadmissible, but it would be allowed if f d-dr had a verbal
value (a sentence such as qiman dah ika zaydun is considered as per-
fectly acceptable by the grammarians .)
All the above boils down to a rejection of the istaqarra /mustaqirrun
hypothesis , and that, in turn, leads Fris to the conclusion that sen-
tences such as zaydun f d-dr/f d-dri zaydun should be considered
neither as jumla fliyya nor as jumla ismiyya ; they must be thought of,
rather, as representing a sentence type in its own right. Note, however,
that Fris did not assign the type of sentence under discussion any spe-
cial designation. Te term jumla z arfyya to which we will be introduced
below was coined in a later period.
But if one is supposed to assume no underlying element linking the
two predicatival constituents in zaydun f d-dr/f d-dri zaydun, what
is then the assigner of raf to zaydun in such cases? As we saw in 3.1,
Sbawayhi , who was not committed to any theory of sentence types , had
no problem presenting the ibtid as raf assigner to zayd in both zaydun
f d-dr and fh zaydun. But for Fris , making a similar claim would
imply classifying fh zaydun as a jumla ismiyya . Regarding the raf
assigner in zaydun f d-dr Fris does not develop any elaborate discus-
sion, apparently because in such cases one would automatically refer to
the ibtid as the raf assigner. However, when it comes to fh zaydun,
the construction on which he focuses his discussion, Fris presents a
clear position as to the rf of zaydun. Having shown that neither a verb
nor a noun can be posited as a linking element, and having proved, fur-
ther, that the adverbial itself cannot be claimed to function as a verb,
Fris (Askariyya, 108109) refers the reader to Ab l-H asan [al-Axfa ]
(d. 733), explaining that these are the reasons why Ab l-H asan regarded
the adverbial per se as the rf when preceding a noun functioning as
muh addat anhu (of whom the message is predicated, subject ). Notice
that it is not the term fil that is used with reference to that noun, but
rather muh addat anhu , a term that cuts across all sentence types. As
we shall see, however, later grammarians did not refrain from using the
term fil in this particular context.
Obviously, attributing the assigning of raf to an adverbial /preposi-
tional phrase constitutes a serious problem for the theory of amal . Since
the formulation of this theory, the grammarians always insisted that the
function of case assignment is implemented by either a verb or a par-
ticle . Various elements, notably active participles and other adjectives
168 yishai peled
were claimed to have a verbal force . But in our case, as we have seen,
Fris argued that f d-dr was not verbal enough to allow a circum-
stantial phrase to precede it. So one might ask what it is that qualifes
fh as raf assigner. To my knowledge, this point has never been clari-
fed by the grammarians . And it is no wonder that the concept of jumla
z arfyya , where a predicatival prepositional phrase assigns raf to the
following subject , remained marginal and never became part of main-
stream medieval Arab grammatical thinking . Anyhow, for Fris , Axfa s
position regarding the raf assigner in fh zaydun constituted further
support for his thesis that this construction represents a sentence type
in its own right.
4. Obligatory fronting of the xabar
4.1 Formal aspects
A remarkable feature of the grammarians (including Fris s) discussion
of the adverbial /prepositional xabar is that they base their argument on
such sentences as zaydun f d-dr/f d-dri zaydun, where the defnite
subject may either precede or follow the predicate (cf. Kouloughli 2002,
10, n. 7). But for the third sentence type advocated by Fris, sentences
such as ind mlun (I have money) would surely be a better example.
For in this case the subject follows the predicate obligatorily; reversing
the order is disallowed.
Most of the grammarians adduce such sentences by way of illustrating
an obligatorily fronted (taqdm ) xabar . However, within the framework
of their grammatical discourse, obligatory fronting of the xabar presents
a major conceptual problem. A xabar, by defnition, must follow, not
precede, the mubtada . Te very concept of an obligatorily fronted xabar
appears to confict with two fundamental principles in medieval Arab
grammatical theory : 1. Te formal principle stipulating that the mil
precede the maml see, e.g. Ibn Ab r-Rab , Bast I:587); 2. Te func-
tional principle placing the constituent representing the given informa-
tion before the one representing what is new for the addressee. To the
extent that the mubtada is identifed with the given, and the xabar with
the new, such sentences pose a serious problem.
Obviously, in cases such as f d-dri rajulun, the grammarians could
not present rajulun f d-dr as the as l of f d-dri rajulun, since the for-
mer is disallowed as an independent sentence in Arabic . One may argue,
the medieval arabic theory of sentence types 169
therefore, that in f d-dri rajulun, as well as in other cases of obliga-
tory fronting to be discussed below, the very concept of taqdm is not
normally intended in the sense that the above structure is the output of
reversing the order of some basic structure in which the indefnite raju-
lun precedes f d-dr. Rather, one must assume that taqdm in such cases
is used in the sense of placing in initial position, with no transformation
involved. Ibn Ya (arh I:86) argues that rajulun f d-dr is excluded
(1) because it could be wrongly interpreted as a noun phrase (with f d-
dr functioning as the attributes ifa of rajulun) rather than as a com-
plete sentence (see also Ibn Us fr , arh I:343), and (2) in order to avoid
introducing a declarative (wjib ) sentence by an indefnite noun .
26
Te idea of transformation in cases such as f d-dri rajulun was not,
however, universally excluded. Not surprisingly, it was suggested, albeit
in a rather idiosyncratic manner, by Ibn Jinn , a grammarian noted
for his originality and for frequently advancing dissenting arguments
incompatible with mainstream medieval Arab grammatical thinking .
In his Sirr s inat al-irb, Ibn Jinn (Sirr I:276) uses the concept as l
marfd (a rejected basic construction) in dealing with cases which
he regards as transformed constructions, but whose underlying struc-
ture (as l) is inadmissible. Te principle of as l marfd is not explicitly
applied by him to rajulun f d-dr. But in his Xas is (I:300) he maintains
that, while mubtada-xabar is the basic word order of a jumla ismiyya ,
a certain intervening factor (rid ) might impose the reversal of that
order. Te occurrence of an indefnite mubtada at the beginning of an
afrmative sentence constitutes in Ibn Jinns view such an rid , a kind
of contingency imposing the movement of the mubtada into the second
position in the sentence (cf. Peled 1992b, 105106).
27
Tis is regarded
26
Astarbd (arh I:232) maintains that the problem of ambiguity between xabar
and s ifa is acute, owing to the common occurrence of an adverbial in xabar position
in Arabic . He cites, however, one case where an adverbial xabar follows an indefnite
mubtada , pointing out that it is perfectly acceptable when the sentence is used as an
exclamation (du). Astarbd also remarks that fronting a non-adverbial xabar to an
indefnite mubtada does not eliminate the ambiguity. Tus, if you transform rajulun
qimun into qimun rajulun, rajulun could be analyzed as xabar of qimun or as an
apposition (badal ) to it, whereas a fronted adverbial in similar cases is bound to be inter-
preted as xabar, due to its nas b case , whether explicitly marked (lafz an ), or understood
by position (mah allan ).
27
Note that an indefnite mubtada introducing a negative or interrogative sentence
is readily accepted by Ibn Jinn . Tus he admits (Xas is I:300) sentences such as hal
ulmun indaka (Is there a boy with you?) and m bist un tah taka (Tere is no
carpet under you), claiming that they are communicatively useful, as opposed to sen-
tences such as rajulun indaka (A man is with you). Te argument is that one can
170 yishai peled
by him as a corrective procedure (is lh al-lafz ):
28
the second position,
he reminds us, is in principle the xabars position; and since the xabar is
essentially indefnite, the indefnite mubtada now flls the appropriate
slot as far as (in)defniteness is concerned. He emphasizes, however, that
in the underlying theoretical (muqaddar) level the mubtada precedes
the xabar (Ibn Jinn , Xas is I:318).
29
Tere are other cases adduced by the grammarians as examples of
obligatory fronting of the xabar . Te frst example presented by Ibn
Ab r-Rab (Bast I:587) is the interrogative construction ayna zaydun
(Where is Zayd?). Tis grammarian strived to demonstrate that such
cases of obligatory xabar fronting do not violate the basic principle that
the mil must precede the maml . Regarding ayna zaydun he indi-
cated that the basic underlying (as l) structure in this case is a-zaydun
f d-dri am f s-sqi am f l-h nti (Is Zayd at home, in the market, or
in the shop?). Te word ayna is an economy device designed to replace
both the adverbials and the interrogative particle a-, as well as the par-
ticle am. It implements a xabar function in virtue of its being a replace-
ment for the adverbials . At the same time, it is obligatorily fronted as a
substitute for the interrogative particle a-. Another semantic compo-
nent in ayna is that of specifcation (tayn ), formally represented in the
basic underlying structure by the particle am. Te transformation from
the basic to the fnal surface structure thus proceeds in the following
stages: First, the xabar ( f d-dri am f s-sqi etc.) is moved to a frontal
position immediately following a- (a-f d-dri am f s-sqi . . . zaydun
the interrogative a- is always positioned sentence-initially). Ten ayna
is introduced to replace all the constituents preceding zaydun. It is thus
the adverbial semantic component in ayna that warrants its occurrence
in pre-mubtada xabar position , whereas the interrogative component
accounts for the obligatoriness of the movement. Ibn Ab r-Rab indi-
cates that the same applies to other interrogatives such as mat, kayfa,
negate the existence of, or pose a question with regard to, an unknown entity (mankr
l yuraf ), while there is no communicative value in predicating of an unknown entity
afrmatively.
28
Similarly, Astarbd (arh I:232) regards the obligatory fronting of the xabar in
cases such as f d-dri rajulun as a corrective (mus ah h ih ) procedure designed to handle
the indefiniteness of the mubtada .
29
Ibn Jinn (Xas is I:319320) then refers to sentences displaying an indefinite
mubtada in sentence-initial position . He argues, however, that these are not predicatival
sentences , in the sense that they are meant to express a wish or imprecation rather than
convey information. Another case is explained by him as paraphrasable by a negative
sentence (cf. n. 27 above).
the medieval arabic theory of sentence types 171
man and m: kayfa axka (How is your brother?), man axka (Who
is your brother?), and the like.
Ibn Ab r-Rab s second case of obligatory xabar fronting is the con-
struction f d-dri rajulun. Tis, however, is dealt by him in pragmatic
rather than in purely formal terms, and will, therefore, be reviewed in the
following sub-section. Te third case is exemplifed by Ibn Ab r-Rab
(Bast I:588) by the sentence al t-tamrati mit luh zubdan (on the date
there is butter of an equal amount). He indicates that the reverse order
(mit luh zubdan al t-tamrati) is disallowed since -h in mit luh is an
anticipatory pronoun in both lafz (surface) and martaba (underlying)
structures , thus violating the rule of the anticipatory pronoun (al-idmr
qabla d-d ikrcf. section 2 above). For further discussion of the relation-
ship between anaphora and the position of the xabar, see Astarbd ,
arh I:232233.
Introducing his fourth and fnal case of obligatory xabar fronting , Ibn
Ab r-Rab (Bast I:588) cites the exceptive sentences m frisun ill
zaydun (No one is a horseman but Zayd) and m f d-dri ill amrun
(No one is in the house but Amr). In these two sentences the subject
nominal occurs sentence-fnally and is dominated by the exceptive par-
ticle ill . Reversing the order in such cases, Ibn Ab r-Rab points out,
would not violate the rules of Arabic grammar , but result in a sentence
diferent in meaning from the original one. Te sentence m frisun ill
zaydun assigns to Zayd, and only to him, the attribute of horsemanship.
Tis sentence, however, is neutral as to whether or not Zayd possesses
other qualities as well. But if the order of constituents is reversed so
as to make the subject zaydun precede the predicate , the resulting sen-
tence m zaydun ill frisun unmistakably excludes the possibility that
Zayd possesses any quality beside horsemanship. Similarly, the sentence
innam frisun zaydun is equivalent in meaning to m frisun ill zay-
dun, whereas innam zaydun frisun is synonymous with m zaydun
ill frisun, which explains why a mubtada-xabar order is inadmissible
in this related case as well.
Ibn Us fr (arh I:353) adds two more cases where the xabar is oblig-
atorily placed sentence-initially: 1. When the mubtada is a nominalized
clause introduced by anna: f ilm annaka qimun (It is known to me
that you are standing); 2. When the xabar is a kam al-xabariyya phrase :
kam dirhamin mluka (How many dirhams you have!).
Te frst of these two cases is dealt with also by Astarbd (arh
I:233234) who cites Fris as claiming that the adverbial /prepositional
phrase in such cases exercises amal (raf ) upon the following anna
172 yishai peled
clause with no supporting element (cf. 3.3 above; for itimd , see sec-
tion 2 above). Astarbd (arh I:233) explains that the reason for the
obligatory fronting of the xabar (whether adverbial or not) in such cases
is that if the anna clause were placed sentence-initially, the word nna
could be misread as inna rather than anna. For, between the two par-
ticles , it is the former rather than the latter that is associated with the
initial position in the sentence. Astarbd points out further that if the
xabar precedes the anna clause it is bound to be correctly analyzed as
xabar to the following clause as a whole rather than as a fronted constit-
uent governed by anna, because a constituent within the scope of inna/
anna cannot be preposed to either of these particles. Furthermore, once
the adverbial /prepositional phrase is established as the xabar of the fol-
lowing clause, then the particle heading that clause will be easily read as
anna, because a mubtada clause , being a noun clause , cannot be intro-
duced by inna. For further discussion of this issue, see Ibn Ya , arh
VIII:5960.
As we have just seen, f d-dri rajulun was only one item, and not nec-
essarily the frst, on the list of constructions presented by the grammari-
ans as examples of obligatory fronting of the xabar . But it was apparently
the most difcult to deal with in purely syntactic terms. For one thing,
like the related f d-dri zaydun, it presented a challenge to the theory
of amal . For another, the indefniteness of the mubtada could not, in
itself, constitute sufcient grounds for ruling out its occurrence in sen-
tence-initial position (cf. Astarbd , arh I:202207 for a detailed dis-
cussion of cases of an indefnite mubtada in sentence-initial position).
Te grammarians main formal explanation for the obligatoriness of
predicate-subject order in this case was that an adverbial/prepositional
phrase following an indefnite nominal could be wrongly interpreted
as an attribute rather than a predicate . But as we have just indicated,
sentences with an indefnite mubtada do occur in Arabic . Indeed, as
we will see shortly, the strongest argument against *rajulun f d-dr was
pragmatic rather than syntactic.
4.2 f d-dri rajulunpragmatic aspects
When examining the construction f d-dri rajulun in terms of informa-
tion structure, most of the grammarians appreciated that sentences of
this kind represent a special case. Indeed they recognized that in these
cases it is the defnite adverbial/prepositional phrase , occupying sen-
tence-initial position that represents the given information, whereas the
the medieval arabic theory of sentence types 173
following nominatival indefnite phrase signals the new information,
and not the other way around (see, e.g. Ibn Ya , arh I:8687).
30
Ibn
Ya (arh I:86) argues that, judging by the meaning of sentences such
as laka mlun (You have money), it is the defnite complement of the
preposition (in this case the pronoun -ka) that represents the muh addat
anhu (what the sentence is about), even though, formally ( f l-lafz ), it
is the nominatival noun that implements that function.
31
In support of
his claim he indicates (1) that laka mlun is paraphrasable by anta d
mlin, and (2) that an indefnite nominal is inadmissible in the position
of the complement of the preposition: *li-rajulin mlun is disallowed
(lam yakun kalmanis not an [acceptable] sentence).
Another attempt to relate the construction in question to a mubtada-
xabar order was made by Ibn Ab r-Rab . Tis grammarian presents
a number of cases where the xabar is obligatorily fronted (cf. 4.1
above), taking great pains to demonstrate that fronting the xabar in
these cases is wholly justifed, if not on purely formal, then on func-
tional/semantic, grounds. Regarding f d-dri rajulun, Ibn Ab r-Rab
states the following:
fa-hd yulzimu t-taqdma wa-l yajzu taxruhu fa-taqla rajulun f
d-dri li-annahu l yubtadau bi-n-nakirati wa-innam jza l-i-btidu
hun bi-n-nakirati li-anna l-maqs da l-ixbru an-i d-dri bi-annah
masknatun wa-laysati n-nakiratu l-maqs data bi-l-ixbri wa-kna l-as lu
an taqla ad-dru mamratun bi-rajulin t umma ard l-i-xtis ra fa-ql
f d-dri rajulun wa-alzam d-dra t-taqdma li-annah l-muxbaru anh
bi-l-h aqqati
in such cases [the xabar] is preposed obligatorily . It may not be post-
posed to yield rajulun f d-dr, because an indefnite noun may not fll a
mubtada position. In our case, the mubtada
32
may be indefnite, because
the intention is to predicate of the house that it is inhabited, rather than
to predicate of the indefnite noun . Underlying [our sentence] is the sen-
tence ad-dru mamratun bi-rajulin. But for the sake of brevity, they say
30
According to Talmon (1993, 285287), the idea is already attested in ninth century
writings where the locative is typically referred to as s ifa and the nominatival noun fol-
lowing it as xabar as -s ifa .
31
This obviously rests on the assumption that in a sentence containing only one
nominatival noun , it is this noun that should be construed as the muh addat anhu . This
term, while referring literally to a pragmatic function, signals in the grammarians usage,
the subject , irrespective of sentence type ; its counterpart h adt signals the predicate (see
Goldenberg 1988, 4649, for discussion).
32
Notice that yubtadau and ibtid are both construed in this case as used as
mubtada or implementing a mubtada function.
174 yishai peled
f d-dri rajulun. Tey obligatorily prepose the [phrase f] d-dr, because
it is really the house that is predicated of [i.e. the topic] (Ibn Ab r-Rab ,
Bast I:587588).
In other words, the sentence f d-dri rajulun is paraphrasable by ad-dru
mamratun bi-rajulin (Te house is inhabited by a man), where
ad-dru obviously functions as mubtada . Ibn Ab r-Rab explains that
the meaning intended by f d-dri rajulun is that the house is inhabited;
it is ad-dr of which something is predicated, not the indefnite rajulun.
Te version f d-dri rajulun, Ibn Ab r-Rab maintains, is preferred to
ad-dru mamratun bi-rajulin for reasons of economy. Te constituent
ad-dr is obligatorily preposed, for it is the one that implements the
function of muxbar anhu , which obviously entails that rajulun func-
tions as xabar . (For the terms muxbar/muxbar anhu, see Goldenberg
1988, 4651).
A diferent, indeed exceptional, view on this issue is held by Jurjn
(Muqtas id I:308309). Sentences such as ind mlun (I have money)
are dealt with by him within the framework of his discussion of sen-
tences with an indefnite mubtada . For Jurjn , the mubtada in this
particular case, albeit indefnite, implements the same pragmatic func-
tion as a defnite sentence-initial mubtada . Mlun in the above sentence
implements the function of mubtada
li-ajli h us li l-ixtis s i f l-xabari id kullu wh idin l yalamu anna indaka
mlan
because the xabar signals some specifcation [regarding the mubtada ],
for it is not common knowledge that you have money (Jurjn Muqtas id
I:308).
For Jurjn , then, ind in its sentence-initial position makes the same
contribution to the communicative value (ifda) of the sentence as
would a xabar occupying a post-mubtada position . Much like other
grammarians , however, Jurjn argues that in such cases the xabar is
obligatorily fronted because mlun ind would be wrongly interpreted
as a noun phrase , with ind analyzed as a complement (s ifa ) to mlun
(see 4.1 above).
Most of the grammarians , however, did not follow this line of thought.
Rather, they discerned a discrepancy between the syntactic analysis of
sentences such as f d-dri rajulun into xabar and mubtada , and the
pragmatic functions of muxbar anhu and xabar implemented by f d-
dr and rajulun respectively. Tey did, however, emphasize that in terms
the medieval arabic theory of sentence types 175
of defniteness sentences like laka mlun conform to the principle that
the defnite constituent, representing the given information, should pre-
cede the indefnite noun signalling the new information. In any case,
the grammarians could not accept an analysis of f d-dri rajulun into
a mubtada followed by a xabar, on the ground that if a sentence of this
kind is introduced by inna, the noun phrase rajul automatically takes
the nas b case , which marks it unmistakably as subject . For discussion,
see Kouloughli (2002, 1617), and his references.
5. Ibn Him al-Ans r s tripartite division
5.1 Background
Returning now to the question of the raf assignment in cases such as
fh zaydun, most of the grammarians from Sbawayhi onwards regarded
ibtid as the operator assigning raf to zaydun. Tis, however, was by no
means universal, as we have seen. Indeed, Ibn al-Anbr (Ins f I:5155)
attributes this position to the Bas rans while presenting the other (much
less common) position as Kfan . We learn that the Kfans , as well as
the Bas ran grammarians Mubarrad and Axfa (cf. 3.3 above), regarded
the adverbial/prepositional phrase in the above case as the assigner of
raf to the following noun zaydun.
33
Ibn al-Anbr maintains that both
the Bas rans and the Kfans resort to Sbawayhi for support for their
respective claims. Te Bas rans obviously point to chapter 133 in the
Kitb, where, as we have seen, Sbawayhi refers to the ibtid (though
without using the actual term) as the mil assigning raf to zaydun. Te
Kfans, for their part, draw upon a number of cases where, according to
Sbawayhi, an adverbial assigns raf to a following noun (Ibn al-Anbr ,
Ins f I:52). It is upon such cases, Ibn al-Anbr argues, that the Kfans
base their claim that in fh zaydun it is fh that should be regarded as
the rf of zaydun.
Astarbd (arh I:218) indicates that the analysis of zaydun in f
d-dri zaydun as fil of the adverbial/prepositional phrase was advanced
by the Kfans , as well as by Axfa in one out of two statements he made
33
This position is clearly evidenced in Farrs Man l-Qurn (e.g. I:195196;
III:133). See Talmon (1993, 279) for further details and references.
176 yishai peled
on the subject.
34
Te argument in both cases is that the adverbial/prepo-
sitional phrase has a verbal force (man l-fl ), analogously to qim in
qimun zaydun.
In Astarbd s view, the Kfans position emanates from their cat-
egorical objection to xabar fronting, irrespective of whether the xabar
is a phrase or a clause. Tis objection, in its turn, is designed to fore-
stall the occurrence of an anticipatory pronoun . To this, however,
Astarbd ofers an outright rejection, claiming, as could be expected,
that the anticipatory pronoun occurs only in the surface structure; in
the basic structure the mubtada precedes the xabar with no anticipa-
tory pronoun involved (cf. section 2 above).
As for Axfa , according to Astarbd he did not object to xabar
fronting, and (in his other statement) indeed regarded the ibtid as the
assigner of raf to zaydun in f d-dri zaydun. Axfas position here is
divergent from the one cited above. It rests upon two assumptions: 1.
Te verbal force of the adverbial/prepositional phrase is weaker than
that of the adjective . 2. Te acceptability of the construction f drihi
zaydun (in his house Zayd [is located]). In this case, the option of ana-
lyzing zaydun as fil is excluded on the ground that it would lead to an
unacceptable anticipatory pronoun (since the pronoun , under this anal-
ysis, would be cataphoric both in the lafz and man confgurationscf.
Astarbd, arh I:202).
Te debate, as could only be expected, concludes with the Bas rans
having the upper hand. Tere is evidence to suggest, however, that, as in
many other cases, the position dismissed as Kfan represented a view
that was much more widely accepted than the mainstream medieval
grammarians would have us believe. As we shall see in the next sub-sec-
tion, it remained viable centuries later, in the writings of one of the most
prominent medieval grammarians , Ibn Him al-Ans r . Te relevance
of this issue to our discussion is clear: it is closely related to the question
of whether or not sentences such as fh zaydun represent a sentence
type in its own right.
34
Ibn al-Anbr (Ins f I:51) adds Mubarrad to the proponents of this kind of analy-
sis (and cf. Ibn Us fr , arh I:158159).
the medieval arabic theory of sentence types 177
5.2 Ibn Hims categorization and defnitions
Coming now to the 14th century grammarian Ibn Him al-Ans r
(d. 1360), we fnd both a critical discussion of the concept jumla as com-
pared with kalm, and an unambiguous division into three sentence
types . Recall that Fris s starting point was that there were two basic
sentence types in Arabic . Tis was followed by an elaborate argument
designed to prove that zaydun f d-dr represented a sentence type in its
own right. For Ibn Him the tripartite division is an established linguis-
tic fact, and he makes it the starting point of his discussion. He does not,
however, ignore the problems raised by this division, as we shall see.
Ibn Him (Mun, 492) starts by defning the concepts kalm and
jumla . Te latter is defned by him as a construction made up of either
fl + fil or, otherwise, mubtada + xabar . Ten, however, he distin-
guishes three types of jumla, ismiyya , fliyya and z arfyya , introduced
respectively by a noun , a verb and an adverbial (z arf ) or a prepositional
phrase ([ jrr wa-] majrr ).
35
Being aware of the problem arising from
his defning each type by the initially occurring constituent, he remarks
(Mun, 492) that the defnitions refer only to predicative constituents
(musnad and musnad ilayhi ). Tus, the sentence a-zaydun axka (Is
Zayd your brother?) and the conditional clause in qma zaydun (If
Zayd stands up) are, respectively, ismiyya and fliyya , even though the
frst noun in the former and the verb in the latter are each preceded by
a particle .
Once the problem of the particles is settled, Ibn Him appears to
be remarkably uncompromising in applying his principle, that Arabic
sentence types must be defned by the initial predicative constituent .
And this, indeed, leads to some conspicuous peculiarities. Te jumla
ismiyya , for instance, is exemplifed by him by the following three sen-
tences: zaydun qimun (Zayd is standing), hayhti l-aqq (How far
is the ravine!) and qimun az-zaydni (Standing are the two Zayds).
While the frst of the three sentences is straightforward, the other two
are not. Te word hayht in the second example is regarded by the
medieval grammarians as ism fl representing, as the term suggests, a
special word category whose members are considered as neither nouns
nor verbs (see, e.g. Astarbd , arh III:165f. for details). Such asm
35
A fourth type, jumla art iyya , which, as he indicates, was proposed by Zamaxar ,
is rejected by Ibn Him on the ground that the conditional clause should be categorized
as jumla filiyya .
178 yishai peled
al-af l , in constructions like hayhti l-aqq, are normally described as
occupying a verb position, with the implication that the following noun
implements the function of fil . However, since hayht as an ism fl is
viewed as a special kind of noun (rather than as a kind of verb), a sen-
tence introduced by it must, according to Ibn Hims rigid principle of
classifcation, be regarded as a jumla ismiyya rather than as jumla fliyya .
Te special problems relating to the third case (qimun az-zaydni)
have already been discussed in section 2 above, and will not be repeated
here. It is, however, noteworthy that for Ibn Him the fact that qimun
az-zaydni is introduced by a participle (viewed by the grammarians as
a nominal element ) was sufcient for classifying this sentence as jumla
ismiyya; the fact that qimun is followed by a noun in the dual form did
not require, in his eyes, any further argument or elaboration.
Te jumla fliyya is illustrated by Ibn Him by six sentences intro-
duced by a verb . Out of these, fve are straightforward: he uses qma,
yaqmu, qum, duriba and z anna to demonstrate that a jumla fliyya
may be introduced by any fnite verb form. Specifcally, by adducing
the sentence z anantuhu qiman (I believed him to be standing), Ibn
Him makes the point that a cognitive verb may, like any other verb,
introduce a jumla fliyya. As is well known, in medieval Arab gram-
matical theory , from Sbawayhi onwards, cognitive verbs such as z anna
(z anna wa-axawtuhz anna and sisters) are presented as analogous
to inna and kna (and their respective sisters), in that they enter into
(yadxulna al) sentences composed of a mubtada and xabar , nullifying
in the process the abstract operator ibtid , and assuming in its stead the
function of a formal mil assigning case to both nominal constituents in
the sentence (see, e.g. Sbawayhi, Kitb I:6; Ibn Ya , arh VII:7778).
In any case, the status of z anna as a verb was never disputed. By
contrast, the verbal status of kna was a matter of controversy among
the grammarians . Te vast majority of grammarians considered kna
as a semantically defcient (nqis ) verb,
36
in the sense that it lacks the
semantic component of action.
37
As such, its only function is to signal
the time of the nominal sentence into which it enters. Signifcantly,
36
Unless signalling existence, in which case it is labelled kna at-tmma complete
kna, and treated as an ordinary verb .
37
This had been recognized already by Sibawayhi, although in his account (Si-
bawayhi, Kitb I:16) he depicted kna as analogous to daraba in terms of transitivity
(tadiya ): like daraba it takes two nominal complements, one in the nominative , the
other in the accusative.
the medieval arabic theory of sentence types 179
while acknowledging the predicative relationship between zaydan and
qiman in z anantu zaydan qiman, the grammarians analyze the two
nominal constituents as direct objects (maf l ) of z anna. By contrast,
in kna zaydun qiman, zaydun and qiman are normally analyzed in
terms of a jumla ismiyya , namely as mubtada and xabar , or otherwise,
as ism kna and xabar kna respectively.
True to his rigid defnitions of the three sentence types , Ibn Him
includes in his examples of jumla fliyya the sentence kna zaydun
qiman (Zayd was standing). And this already seems to run counter
to the mainstream conception of jumla fliyya. Indeed, the grammar-
ians do not normally bring kna into their discussions of sentence types.
But given their special treatment of this verb , it is highly unlikely that
they would have classifed kna zaydun qiman as jumla fliyya just by
the verbal morphology of kna.
But the most interesting for our present discussion is Ibn Hims
jumla z arfyya . Tis he illustrates by the two sentences a-indaka zaydun
(Is Zayd with you?) and a-f d-dri zaydun (Is Zayd in the house?),
where the frst predicatival constituent is an adverbial and a preposi-
tional phrase , respectively. Ibn Him points out that sentences such as
these can qualify as jumla z arfyya only
id qaddarta zaydan filan bi-z -z arf wa-l-jrri wa-l-majrri l bi-l-
istiqrri l-mah d f wa-l mubtadaan muxbaran anhu bihim
if you assume zayd to be a fil [acted upon] by the adverbial/preposi-
tional phrase , not by a deleted [verb /participle conveying the meaning of]
istiqrr , and [only if] you do not analyze zayd as a mubtada for which
the adverbial/prepositional phrase serves as xabar (Ibn Him al-Ans r ,
Mun, 492).
Te signifcance of this passage lies in that it seems to suggest that Ibn
Hims sentence-type defnitions were not as rigid as they appeared
to be when we looked at his defnitions and illustrations of the jumla
ismiyya and the jumla fliyya . It now turns out that for him, the predica-
tive constituent that comes frst in the sentence is not in itself the only
criterion for determining the type of sentence. Rather, for a predicati-
val constituent to qualify as sentence-type identifer it must act as mil
upon the second predicatival constituent.
Another important point to note is that in both of Ibn Hims exam-
ples the adverbial/prepositional phrase is preceded by the interrogative
particle a-. Tis may be taken to suggest that by the time of Ibn Him
the principle of itimd , which, as we have seen (section 2 above), can be
180 yishai peled
traced back to Sbawayhi , had already been frmly established in medi-
eval Arab grammatical thought . When Ibn Him states that a sentence
can only qualify as a jumla z arfyya if the second constituent functions
as fil to the frst, illustrating this with examples displaying the inter-
rogative particle a- preceding the frst predicative constituent , one is
bound to conclude that for him a jumla z arfyya is a sentence whose frst
predicatival constituent is an adverbial/prepositional phrase , where that
phrase acts as a verb , thus assigning raf to the following constituent on
the strength of the principle of itimd .
Ibn Him then goes on to make a critical remark directed at
Zamaxar . He indicates that Zamaxar exemplifed jumla z arfyya by
the phrase f d-dr in zaydun f d-dr (cf. Ibn Ya , arh I:88). Tis posi-
tion, he argues, is based:
al anna l-istiqrra l-muqaddara flun l ismun wa-al annahu h ud ifa
wah dahu wa-ntaqala d-damru il z -z arf bada an amila fhi
on [the assumption] that [the underlying word conveying] istiqrr is a
verb , not a noun , and that that verb was deleted alone while the pronoun
implicit in it moved to the adverbial phrase , afer [the verb] had exercised
amal upon the adverbial (Ibn Him al-Ans r , Mun, 492).
Recall that zaydun f d-dr was described by Fris as representing a
sentence type in its own right on the ground that, in his view, neither a
verb nor a noun could be posited as a linking element between zayd and
f d-dr. Here Zamaxar is quoted as elevating f d-dr to the status of
a clausal xabar (jumla z arfyya ). Tis is done, so the argument goes, by
positing an underlying verb that is deleted while the pronoun implicit
in it is transferred to the adverbial , the verb having exercised amal upon
that adverbial. In other words, the clausal status of f d-dr stems from
the pronoun it receives from the deleted verb istaqarra . What we see
here is, indeed, another attempt to account for the predicatival relation-
ship between zaydun and f d-dr, two non-coreferential elements, as
well as for the irb of the xabar . However, this attempt is based on the
istaqarra hypothesis (3.2 above), and that is precisely the reason why it
is rejected by Ibn Him. Zamaxars analysis is incompatible with Ibn
Hims conception of jumla z arfyya . For Ibn Him, once an underly-
ing verb is assumed, the clause should be regarded as a jumla fliyya ;
exactly as, when one posits a xabar-mubtada relationship between f
d-dr and zayd, the sentence must be considered as jumla ismiyya . Tis
will be further clarifed below.
In 3.3 we pointed to Fris s reference to Axfa , who had attributed
to the adverbial/prepositional phrase the function of operator assigning
the medieval arabic theory of sentence types 181
raf to the nominal following it. He did not, however, refer to that nomi-
nal explicitly as fil . Te term muh addat anhu which he used, signals
in medieval Arabic grammatical literature the subject of the sentence,
whether a fil or a mubtada . Indeed, the specifc grammatical status of
the nominatival constituent , as determined by the mil assigning it the
raf case , never ceased to be a topic of debate among the grammarians .
Yet one thing emerges quite clearly. Te analysis of zaydun as fil in
both f d-dri zaydun and qimun zaydun, normally attributed to the
Kfans and Axfa (see section 2 above), always comes up when these
two constructions are discussed. It was never abandoned. However, of
these two constructions, it is only fh/f d-dri zaydun that was con-
sidered, albeit by a small number of grammarians , as a sentence type in
its own right. Te reason for this should by now be clear. Te opening
predicative constituent in each of the three sentence types was regarded
as an operator (mil ) assigning case to the following constituent(s): sen-
tence types were unmistakably correlated with amal types. And since
the participle (qimun) could not be viewed as other than a verbal or
a nominal element , it could not be regarded as introducing a sentence
type in its own right. By the same token, a sentence such as zaydun f d-
dr, introduced as it is by a nominatival noun , could only be defned as
jumla ismiyya . Te concept jumla z arfyya was by and large associated
with cases where a z arf could be claimed to be a mil assigning raf to
the nominal following it. As we will see in the next sub-section, it was
Ibn Him , an eminent proponent of the tripartite division, who also
appreciated and spelled out the problems arising from the actual notion
of sentence types in Arabic , whether two or three.
5.3 Problems
As we have seen throughout, the problems the grammarians encoun-
tered in categorizing Arabic sentences stemmed from the fact that their
conception of sentence types was deeply embedded in the theory of
amal . Tis is manifested also in the way these problems are illustrated
by Ibn Him (Mun, 493497). He ofers an illuminating discussion
of ten cases where a sentence can be construed as either a jumla fliyya
or a jumla ismiyya , or, otherwise, raise a controversy among grammar-
ians as to the right categorization. Signifcantly, no case is cited as an
unambiguous jumla z arfyya . Since the basic arguments recur through-
out his discussion, I will review only four of his examples that, I believe,
well illustrate the problematic aspects of the traditional categorization
of sentence types.
182 yishai peled
Let us start with Ibn Hims fourh example md s anata (What have
you done?). He points out that this sentence may be paraphrased as
either m llad s anatahu, or as ayya ayin s anata. Since llad s anatahu
is a nominalized constituent, the sentence, according to the frst para-
phrase, must be categorized as a jumla ismiyya . Ibn Him indicates that
the frst constituent m is analyzed as a fronted xabar by Axfa , and
as mubtada by Sbawayhi . By contrast, the proponents of the second
paraphrase, ayya ayin s anata, would categorize the same sentence as a
jumla fliyya , analyzing ayya ayin as a fronted direct object . (And see,
further, Ibn Hims discussion of the sentence md s anatahu.)
Ibn Hims sixth example reads qm axawka (Your two brothers
stood up). Tis sentence is presented by him as acceptable, subject to
specifc types of analysis. (To what extent this construction was in actual
use in medieval Arabic is immaterial for the present discussion). First,
the sentence could be categorized as jumla fliyya if (1) the ending - in
qm is interpreted as a dual-marking particle (h arf tat niya ), much as
the -t in qmat Hindun is analyzed as a feminine marker (and not as a
pronoun ); or alternatively if (2) the ending - is interpreted nominally
and the following axawka is analyzed as apposition (badal ) to it. Sec-
ond, qm axawka may be categorized as jumla ismiyya with a fronted
xabar (with the ending - interpreted nominally and axawka analyzed
as a postposed mubtada ). Note that Ibn Him does not mention the
possibility of analyzing qm axawka as a jumla fliyya with axawka
functioning as fil to qm (luat akaln l-bart see above, section
2, n. 6).
Te seventh example presented by Ibn Him is nima r-rajulu zaydun
(What a nice man is Zayd). Tis sentence, he explains, may be viewed
as an inverted jumla ismiyya , with nima r-rajulu functioning as a pre-
posed xabar to zayd. Under an alternative analysis, however, zaydun
could function as xabar to a deleted mubtada . Ibn Him argues that
under the latter analysis, nima r-rajulu zaydun consists of two asyndeti-
cally coordinated clauses, the frst one (nima r-rajulu) verbal, and the
second nominal.
But perhaps the most interesting is Ibn Hims second example,
where he makes the following statement regarding a-f d-dri zaydun
and a-indaka amrun:
fa-inn in qaddarn l-marfa mubtadaan aw marfan bi-mubtadain
mah d fn taqdruhu kinun aw mustaqirrun fa-l-jumlatu ismiyyatun d tu
xabarin f l-l wa-d tu filin munin an-i l-xabari f t -t niyati wa-in
qaddarnhu filan bi-staqarra fa-fliyyatun aw bi-z -z arf fa-z arfyyatun
the medieval arabic theory of sentence types 183
if we analyze the nominatival constituent as mubtada , or otherwise, as
a nominal assigned the raf case by a deleted mubtada such as kinun or
mustaqirrun, then the sentence should be considered as nominal, with a
xabar under the frst analysis, or with a fil replacing the xabar under the
second. If, however, we analyze it [i.e. zaydun or amrun] as fil of [an
underlying] istaqarra , then the sentence is verbal; if [the operator assign-
ing raf to the fil ] is the adverbial , then the sentence should be consid-
ered as a jumla z arfyya (Ibn Him al-Ans r , Mun, 494).
Here Ibn Him ofers four ways for analyzing sentences such as a-f
d-dri zaydun, correlating each analysis with a diferent sentence type .
Tese are the four options as presented in the above passage, in Ibn
Hims order:
1. zaydun could be analyzed as mubtada . Tis would imply that the
sentence is a jumla ismiyya , with the adverbial/prepositional phrase
implementing the function of (a preposed) xabar .
2. We could posit an underlying mubtada , such as kinun or musta-
qirrun, assigning the raf case to zaydun. In this case zaydun would
implement the function of fil replacing (munin an) the xabar . Te
sentence under such an analysis would be regarded, according to Ibn
Him , as jumla ismiyya .
3. zaydun could be analyzed as fil assigned the raf case by the under-
lying verb istaqarra . In this case the sentence would be considered as
jumla fliyya .
4. If, however, we analyze zaydun as a fil receiving its raf case from the
preceding adverbial/prepositional phrase , then the sentence should
be regarded as jumla z arfyya .
In Ibn Hims view, then, a fil is not necessarily preceded by a fnite
verb . But it is only when the operator assigning raf to the fil is a fnite
verb (whether explicit or underlying) that the sentence may be con-
ceived of as jumla fliyya . When the raf assigner is a participle (whether
explicit or underlying) or an adverbial/prepositional phrase , the sen-
tence must be conceived of as a jumla ismiyya in the frst case, and as a
jumla z arfyya in the second, even though the actual use of the term fil
suggests that the participle and the adverbial / prepositional phrase in
such cases behave analogously to a verb .
As can be seen, Ibn Him presents the four options without any
attempt to defend his categorization. His analyses are consistent with
his defnitions of the three sentence types (see 5.2 above), and mani-
festly refect the controversies relating to the constructions in question.
184 yishai peled
Te proponents of the frst option would presumably regard a-f d-dri
zaydun as the inverted version of a-zaydun f d-dr. Te occurrence of
the interrogative a- in this case is irrelevant, as is the case also under
the third analysis, where the sentence is presented as an unmistakable
jumla fliyya . Indeed, positing an underlying verb like istaqarra in order
to account for the raf case of zaydun in sentences of this kind was com-
mon practice among the grammarians , as we saw in 3.2. What is really
remarkable in Ibn Hims third analysis is that it leads to the important
conclusion that under a certain analysis a sentence such as a-f d-dri
zaydun could be conceived of as jumla fliyya .
Under the second analysis, zaydun implements the function of fil
following a deleted mubtada , thus occupying a xabar position. Te
adverbial/prepositional phrase , under this as well as under the third
analysis (see above), would be analyzed as an adjunct. Obviously, the
second analysis is reminiscent of the analysis of (a-)qimun zaydun
into a mubtada followed by a fil replacing the xabar, as we saw in sec-
tion 2. Note that, unlike Sbawayhi (Kitb I:239; and cf. 3.1 above), Ibn
Him does not view the prepositional phrase as occupying a mubtada
position. Rather, the mubtada in this case is an underlying participle .
Here, at any rate, a sentence whose subject is labeled fil is categorized
as jumla ismiyya .
As we have already indicated, the fourth analysis is consistent with
Ibn Hims theory of three sentence types , each defned by the pre-
dicative constituent introducing the sentence (and acting as mil upon
the second constituent). But we have already seen (3.3 above) that it
remains unclear how an adverbial/prepositional phrase can function
as a verb assigning raf to a following nominal constituent . Following
Sbawayhi , it was ofen argued that such a phrase may act analogously to
a verb when preceded by a supporting element such as the interroga-
tive particle a- (itimd cf. sections 2, 4.1 above). But does that in itself
warrant categorizing the construction a-f d-dri zaydun as represent-
ing a sentence type in its own right? Te concept of jumla z arfyya does
not seem to have been seriously discussed in the writings of the medi-
eval Arab grammarians . Evidently, the vast majority found it difcult to
ft the concept of jumla z arfyya into their theory of amal . Indeed, this is
manifested even in Ibn Hims position, which does not present a-f d-
dri zaydun as a straightforward jumla z arfyya . Rather, it makes it clear that
the actual identifcation of a sentence as jumla z arfyya is essen-
tially dependent upon conceiving the adverbial/prepositional phrase as
a mil assigning raf to the following nominal. Te other two types,
the medieval arabic theory of sentence types 185
in contrast, could be determined straightforwardly, since both the verb
and the ibtid were established awmil in the medieval theory of amal.
Hence the grammarians adherence to the binary system of jumla fliyya
and jumla ismiyya .
38
6. Summary
Since the grammarians theory of sentence types grew out of, and has
always been closely related to, their theory of amal , it is not surprising
that in elaborate discussions of Arabic sentence types , particularly those
of Fris and Ibn Him , problems relating to the categorization of cer-
tain constructions were couched in terms of case assignment (amal ).
Te basic types of jumla fliyya and jumla ismiyya are shown through-
out to represent two types of amal: verbal tadiya and ibtid . Sentences
such as qimun zaydun and f d-dri zaydun/rajulun are shown to be
problematical in terms of amal. With regard to qimun zaydun, we have
seen that many grammarians advocated the rather awkward analysis of
mubtada +fil sadda masadd al-xabar . Tis was designed to deal with
the essentially nominal nature of the participle occurring sentence-ini-
tially, as well as with its verb -like behavior in this particular case. Apart
from Ibn Him who regarded this construction as an example of jumla
ismiyya , the proponents of the above analysis did not commit themselves
to any clear-cut categorization of this particular structure. Regarding f
d-dri zaydun/rajulun, the very fact that this construction displays an
adverbial/prepositional predicative constituent in sentence-initial posi-
tion , gave rise to the hypothesis that it represents a sentence type in its
own right, a jumla z arfyya . And it comes as no surprise that this was
associated with the hypothesis that in such cases it is the adverbial/prep-
ositional phrase that assigns raf to the nominal constituent following it.
Obviously, this hypothesis and the long established istaqarra /mustaqir-
run hypothesis were mutually exclusive. In 3.3 we saw Friss attempt to
refute the istaqarra/mustaqirrun hypothesis, arguing from the theory of
amal. Tis was his line of defending a tripartite sentence-type system.
38
For a modern study advocating a three-type division, see Kouloughli (2002, 2124)
who argues that sentences such as f d-dri rajulun/zaydun (referred to by him as loca-
tive sentences) should be viewd as representing a sentence type in its on right, since
they exhibit a number of syntactic and semantic properties not shared by regular topic-
comment sentences.
186 yishai peled
For him, indeed, claiming that fh zaydun represented a sentence type
in its own right was tantamount to presenting fh as the assigner of raf
to zaydun. But such an argument could never be accepted by the major-
ity of grammarians , because it was considered a major violation of a cen-
tral principle of the theory of amal . Te vast majority of grammarians
did fnd the istaqarra/mustaqirrun hypothesis a convenient tool for ft-
ting such constructions as zaydun fh and fh zaydun into their theory
of amal. Te question of sentence type was apparently secondary. Once
they established the status of mubtada to zaydun in both cases, they
could argue that both represent a jumla ismiyya , irrespective of whether
the underlying element linking the adverbial/prepositional phrase to
the mubtada is a verb or a participle . Fh zaydun was thus conceived
of as the inverted version of zaydun fh. To Ibn Him, as we have seen,
this was unacceptable. For, if underlying fh zaydun is the structure
yastaqirru fh zaydun, it follows that zaydun is assigned the raf case
by the underlying verb occupying sentence-initial position. And Ibn
Him, indeed, drew the conclusion following from that assumption,
namely that under the above analysis fh zaydun must be categorized
as jumla fliyya . But this position of Ibn Hims is clearly exceptional
in the medieval grammatical literature . Te majority of grammarians,
conforming to the theory of amal, did posit istaqarra as the underlying
case assigner to zaydun, but they never went as far as categorizing fh
zaydun as jumla fliyya . Rather, fh zaydun has always been regarded
an ibtid construction, clearly associated with jumla ismiyya.
7. References
7.1 Primary sources
Astarbd , arh = Rad ad-Dn Muh ammad b. al-H asan al-Astarbd, arh Kfyat
Ibn al-H jib. Emil Bad Yaqb, ed . Beirut: Dr al-Kutub al-Ilmiyya, 1998.
Fris , Askariyya = Ab Al l-Fris, al-Masil al-askariyyah. Muh ammad a-t ir
Ah mad Muh ammad Ah mad, ed . Cairo: Mat baat al-Madan, 1982.
Farr, Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ah mad
Ysuf Najt and Muh ammad Al an-Najjr, eds . Cairo: ad-Dr al-Mis riyya li-t-Talf
wa-t-Tarjama, 19551972.
Ibn Ab r-Rab , Bast = Ubaydallh b. Ah mad b. Ubaydallh Ibn Ab r-Rab, al-Bast f
arh jumal az-Zajjj . Ayyd b. d at -T abt, ed . Beirut: Dr al-arb al-Islm, 1986.
Ibn al-Anbr , Asrr = Ab l-Barakt Abd ar-Rah mn b. Muh ammd b. Ab Sad al-
Anbr, Kitb Asrr al-Arabiyyah. Muh ammad Bahjat al-Bayt r, ed. Damascus:
Mat bt al-Majma al-Ilm al-Arab bi-Dimaq, 1957.
Ibn al-Anbr , Ins f = Ab l-Barakt Abd ar-Rah mn b. Muh ammd b. Ab Sad al-
Anbr, Kitb al-Ins f f masil al-xilf bayn an-nah wiyyn al-bas riyyn wa-l-kfyyn.
Muh ammad Muh ad-Dn Abd al-H amd, ed . Beirut: Dr al-Fikr, n.d.
the medieval arabic theory of sentence types 187
Ibn Aql , arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al alfyyat Ibn Mlik.
Muh ammad Muhy ad-Dn Abd al-H amd, ed . n.p: Dr S ab, n.d.
Ibn Him , Mun = Jaml ad-Dn Ibn Him al-Ans r, Mun l-labb an kutub al-
arb. Mzin al-Mubrak and Muh ammad Al H amdallh, eds. Beirut: Dr al-Fikr,
1985.
Ibn Jinn , Sirr = Ab l-Fath Ut mn Ibn Jinn, Sirr s inat al-irb. H asan Hindw, ed .
Damascus: Dr al-Qalam, 1985.
Ibn Jinn , Xas is = Ab l-Fath Ut mn Ibn Jinn, al-Xas is . Muh ammad Al an-Najjr,
ed . Cairo: al-Haya l-Mis riyya l-mma li-l-Kitb, 19861988.
Ibn as-Sarrj , Us l = Ab Bakr Muh ammad b. Sahl Ibn as-Sarrj, Al-Us l f n-nah w.
Abd al-H usayn al-Fatl, ed . Beirut: Muassasat ar-Risla, 1987.
Ibn Us fr , arh = Al b. Mumin b. Muh ammad b. Al Ibn Us fr, arh jumal az-
Zajjj. S h ib Ab Janh , ed . Mosul: Ihy at-Turt al-Islm, 19801982.
Ibn Ya , arh = Muwafaq ad-Dn Ya b. Al Ibn Ya, arh al-Mufas s al. Cairo:
Maktabat al-Mutanabb, n.d.
Jurjn , Muqtas id = Abd al-Qhir al-Jurjn, Kitb al-muqtas id f arh al-dh . Kz im
Bah r al-Marjn, ed . Baghdad: Dr ar-Rad li-n-Nar, 1982.
Muji , arh = Ab l-H asan Al b. Fadd l al-Muji, arh uyn al-irb. Abd al-
Fatth Salm, ed . Cairo: Dr al-Marif, 1988.
Sbawayhi , Kitb = Ab Bir Amr b. Ut mn Sbawayhi, Al-Kitb. Hartwig Derenbourg,
ed . Hildesheim and New York: G. Olms, 1970.
Zajjj , Jumal = Ab l-Qsim Abd ar-Rah mn b. Ish q az-Zajjj, Kitb al-jumal f n-
nah w. Al Tawfq al-H amad, ed . Beirut: Muassasat ar-Risla, Dr al-Amal, 1988.
7.2 Secondary sources
Badawi, El-Said M. 2000. Ray f man l-irb f fus h t-turt : H lat al-jumla l-
ismiyya. Diversity in Language: Contrastive Studies in Arabic and English Teoretical
and Applied Linguistics, Z. M. Ibrahim, S. T. Aydelott and N. Kassabgy, eds., 120.
Cairo: Te American University in Cairo Press.
Carter, Michael G. 1973. Sarf et H ilf: Contribution lhistoire de la grammaire arabe.
Arabica 203, 292304.
. (ed.). 1981. Arab Linguistics: An introductory classical text with translation and
notes (irbns Nr al-sajiyyah). Amsterdam: John Benjamins.
Goldenberg, Gideon. 1988. Subject and Predicate in Arab Grammatical Tradition.
Zeitschrif der Deutschen Morgenlndischen Gesellschaf 1381, 3973.
. 2002. Two Types of Phrase Adjectivization. Sprich doch mit deinen Knechten
aramisch, wir verstehen es!: 60 Beitrge zur Semitistik, Festschrif fr Otto Jastrow
zum 60. Geburtstag, W. Arnold and H. Bobzin, eds. 193208. Wiesbaden: Harras-
sowitz.
Kouloughli, D. E. 2002. On locative sentences in Arabic. Zeitschrif fr arabische Lin-
guistik 41, 726.
Levin, Aryeh . 1985. Te distinction between nominal and verbal sentences according
to the Arab grammarians. Zeitschrif fr arabische Linguistik 15, 118127.
. 1987. Te views of the Arab grammarians on the classifcation and syntactic func-
tion of prepositions. Jerusalem Studies in Arabic and Islam 10, 342367.
. 1989. What is meant by akaln l-bart u? Jerusalem Studies in Arabic and
Islam 12, 4065.
Owens, Jonathan. 1988. Te Foundations of Grammar: An Introduction to Medieval Ara-
bic Grammatical Teory. Amsterdam: John Benjamins.
. 1989. Te Syntactic Basis of Arabic Word Classifcation. Arabica 36, 211234.
Peled, Yishai . 1992a. Amal and Ibtid in Medieval Arabic Grammatical Tradition.
Abr-Nahrain 30, 146171.
. 1992b. Cataphora and Taqdr in medieval Arabic grammatical theory. Jerusalem
Studies in Arabic and Islam 15, 94112.
188 yishai peled
Talmon, Rafael . 1993. Two Early non-Sbawaihian Views of amal in Kernel-Sen-
tences. Zeitschrif fr arabische Linguistik 25, 278288.
. 1997. Arabic Grammar in its Formative Age: Kitb al-Ayn and its attribution to
Xall b. Ah mad. Leiden: E. J. Brill.
ARABIC AVANT LA LETTRE. DIVINE, PROPHETIC, AND
HEROIC ARABIC
Stefan Wild
Bonn
1. Introduction: the Qurn
Te history of the Arabic language is indelibly marked by the fact that
the Qurn has made Arabic a prophetic language with its own holy
book and a worldwide appeal. Arabic had been a mantic language in
pre-Islamic times . When the poet recited verses or when the soothsayer
uttered his sayings, reciter and listener were sure that behind his voice
there was another voice. Tis voice really speaking was that of a higher
power.
With the Qurn , the mantic voice behind the voice of the Prophet
Muhammad became in the believers ear the voice of the one and only
God . Muslim dogma and the consensus of the unlearned considered
the Qurn to be the direct, undiluted Arabic word of God. Te status of
Classical Arabic , the standardization of Arabic including the develop-
ment of Modern Standard Arabic , the diglossia Standard Arabic versus
Arabic dialects , the nature of the Arabic vocabulary , Arabic orthography ,
Arabic style and vocabularyall are unthinkable without the Qurnic
fact. Te history of the Arabic language down to our times cannot be
written without constant reference to the Qurn.
Conversely, the Qurn is deeply marked by its Arabness. Te Qurn
is the frst literary document in Arabic . It is also the frst Arabic docu-
ment to mention the Arabic language. In contradistinction to the atti-
tude of the Jewish Bible and the Christian New Testament toward their
own linguistic forms , the language of the Qurn is an important topic
of Qurnic self-refection. While neither the Bible nor the New Testa-
ment refer to their Hebrew (or Aramaic ) and Greek literary forms, the
Arabic character of the Qurn is in its self-view a cornerstone of its
divine quality.
Te adjective arab in the Qurn always refers to the language, never
to a tribe or a social class. Te term as an ethnic designation is also
190 stefan wild
very rare in Jhiliyya poetry .
1
In the Qurn, arab refers always to the
language of the holy text. It occurs eleven times in the Qurn, and only
in Sura s traditionally dated to the Middle and late Meccan period . In
six passages, the adjective arab is a qualifcation of the word Qurn,
a word meaning primarily recitation, reading aloud . . . (Sura 12:12;
20:113; 39: 2728; 41:14; 42:7; 43:14). Te conjunction lisn arab
Arabic tongue occurs three times in the Qurn (Sura 16:103; 26:195,
46:12), and is used to describe the language of the Qurn . Te conjunc-
tion h ukm arab (Sura 13:37) an Arabic judgment also refers to the
holy text, and in 41:44, in which the possibility of a Qurn ajam a
non-Arabic Qurn is dismissed, the adjective arab again refers to the
holy text. Te Arabic quality of Qurnic revelation could scarcely be
more solidly established.
On the other hand, the Qurn seems indiferent to the linguistic
shape of preceding revelations . In the Qurn, the only language used is
Arabic . Arabic is also the only language mentioned by name. Te Qurn
does not specify in which languages Nh , Ibrhm , Isml , Ms , s or
other prophets and messengers spoke to their peoples or in which lan-
guages their holy books might have been.
God speaks Arabic to Adam and his wife, Satan whispers in Arabic
(Sura 20:120), the angels and the jinn speak Arabic (Sura 72:115),
Moses addresses the Pharaoh , Joseph addresses the Egyptian ministers
wife in Arabic, Jesus speaks Arabic from his cradle, D l-Qarnayn and
the People of the Cave they all use Arabic. Every single soul is made to
speak Arabic at the Day of Judgment , animals like the ant (Sura 27:18) or
the hoopoe (Sura 27:22), even inanimate entities like Hell (Sura 50:30)
speak Arabic. Everybody and everything that speaks in the Qurn must
necessarily speak Arabic, because Arabic is the only language used
throughout the Qurn.
But the intention of the text is in no way to convey that all mankind
throughout history shared and will share the same language. I do not
know of any exegete who concluded from the Qurnic accounts that
the language used between Ms and the Egyptian Pharaoh was Arabic
or that the language used between the Egyptian notables wife and her
lady-friends must have been Arabic, or that the Messiah spoke in Ara-
bicjust because the Qurn reproduces their words in Arabic. It is a
diferent matter for Adam and Isml (see below).
1
Cf. Agha and Khalidi : Poetry and Identity 70.
divine, prophetic, and heroic arabic 191
On the other hand, the Qurn does mention the existence of lan-
guages other than Arabic and emphasizes that the plurality and variety
of human languages is a sign of divine grace. Te divine creation of dif-
ferent languages is as important as the creation of heaven and earth:
wa-min ytih xalqu s-samawti wa-l-ardi wa-xtilfu alsinatikum
wa-alwnikum nna f d lika la-ytin li-l-limn
And of His signs is the creation of heavens and earth and the variety of
your tongues and hues. Surely in this are signs for people who know (Sura
30:22).
Even as God created man and woman, heaven and earth, he created
diferent colors (alwn, or: kinds of human skin) and diferent human
languages (cf. the enumeration in Sura 30:2025). Te existence of dif-
ferent languages is one of Gods signs for those who know. Unless the
word ya is here taken to mean a portent of warning, such a view seems
difcult to reconcile with the myth of the Babylonian tower (Gen 11:
19), according to which the origin of a multitude of human languages
is divine punishment. Sura 30:22 may even be an inter-textual stab at the
narrative of the Babylonian Tower . Tere were, however, extra-Qurnic
traditions that preserved the motif that the diference between human
languages was due to an act of divine wrath (p. 196).
A second important element in the Qurnic linguistic outlook fows
forth from this esteem of other languages. In the course of history, God
sent each messenger (rasl ) to a specifc people (qawm), and this mes-
senger brought the divine message to that people in its language.
And We have sent no messenger save with the tongue of his people that he
might make all clear to them
wa-m arsaln min raslin ill bi-lisni qawmih li-yubayyina lahum
(Sura 14:4).
Te primary raison dtre of the Qurn is that the Prophet Muhammad s
message was in Arabic . Whereas the other prophets and messengers had
been sent earlier with a message in the languages of their peoples, who
did not speak Arabic, Muhammad is sent to the Arabs . Te Arabic lan-
guage vouchsafes the understanding of the Arabic-speaking audience.
For a great part of the Qurn, the fact that this revelation was in Arabic
was the most important diference between Muslim revelation and all
previous revelation.
Tese Qurnic elements infuenced a theological-linguistic scenario
that gave rise to a particular image of linguistic history . Te claim that
192 stefan wild
Arabic as the language of the Qurn had become a divine language,
led to the construction of a religious pre-history of Arabic . Tere were
three competing strands of knowledge that were interwoven to form a
comprehensive imaginaire of the history of Arabic and by implication of
human languages in general:
1. the philosophical question of the origin of language in general;
2. exegesis of the Qurn , including the Qis as al-anbiy literature, i.e.,
reports on earlier prophets , which depended heavily on Biblical and
Aggadic material ; and
3. historical traditions about pre-Islamic history , including the genealo-
gies of the Arab tribes .
One of the main issues was the origin of Arabic , an issue ofen framed as
the question Who was the frst human being to speak Arabic? Muslim -
Arab scholarship tried sometimes to disentangle, sometimes to com-
bine and ofen to harmonize the heterogeneous strands of philosophical
speculation, exegetical H adt , and genealogical traditions . In many cases,
however, contradictory reports were just lef standing side by side. An
extensive and useful overview of much of the Arabic material was given
by as-Suyt in his Muzhir.
2
Te following remarks will outline the main
lines of this colorful and ofen contradictory linguistic yarn.
2. Adam s Arabic and Nh s Syriac
Te basic divide between Arab philosophers and theologians concerned
with the origin of language was whether it rested on human convention
(is t ilh , tawdu ) or on a divine act of revelation (ilhm , wah y , tawqf ).
3
Te Platonic controversy whether human language was what it was phy-
sei or thesei, by nature or by imposition, here took a new form. A further
problem was whether, theologically speaking, Arabic had a special lin-
guistic status that set it apart from all other languages.
2
Andrzej Czapkiewicz , The Views of the Medieval Arab Philologists on Language and
its Origin in the Light of Al-Suyutis Al-Muzhir, Krakow 1989. Czapkieviczs translations
are sometimes hard to understand.
3
For the difference between tawqf and ilhm , cf. J. van Ess , Theologie und Gesell-
schaft IV 325.
divine, prophetic, and heroic arabic 193
An important part of tradition seems to be in general agreement that
Adam , who is in Islam a prophet, was the frst human being to speak Ara-
bic . Te revelationists based their view mainly on the early exegesis of
Qurn 2:31 wa-allama dama l-asma kullah and He (God ) taught
Adam all names. Weiss explains: According to at -T abar , the majority
of early exegetes , including the noted Companion and tafsr -authority
Ibn Abbs , interpreted this verse as meaning that God taught Adam the
Arabic names of all existing things. Tis meant that Adams language
was revealed by divine teaching. It also clearly implied that God taught
Adam language in its entirety.
4
According to Ibn Jinn (d. 392/1002),
most Mutazilites , on the other hand, taught that all languages including
Arabic rested on convention, not on revelation .
5
Even those who accepted the revelationist theory could argue that
Sura 2:31 meant that Adam was taught the names of all things created in
all languages, in Arabic, of course, but also in Persian , Syriac , Hebrew ,
Greek , and all other languages as Ibn Jinn explains. Adam and his of-
spring spoke all these languages. But in the course of time, Adams of-
spring spread over the world and each of them stuck to one of these
languages until this language became his main language and other
languages eluded him because of his lack of contact with them (qad
fussira hd bi-an qla inna llha subh nah allama dama asma
jami l-maxlqti bi-jami l-lugti l-arabiyyati wa-l-frisiyyati wa-l-
suryniyyati wa-l-ibrniyyati wa-r-rmiyyati wa-gayri d lika min siri
l-lugti fa-kna damu wa-waladuh yatakallamna bih t umma inna
waladah tafarraq f d-duny wa-aliqa kullun minhum bi-lugatin min
tilka l-lugti fa-galabat alayhi wa-dmahalla anhu m siwh li-budi
ahdihim bih, Ibn Jinn , Xas is I 41). Tis model safeguarded the pri-
ority of Arabic ; Arabic as the language of the divine word must be prior
to all other languages, whether their origin was revealed or due to con-
vention. But according to some, all other languages were also revealed
to Adam by God . Others with a diferent timetable thought that difer-
ent linguistic communities emerged only afer the Flood. Te idea that
Adam had known all human languages found a parallel in the belief of
some Shiites in the 9th
century that the Prophet Muhammad had known
4
Weiss , Muslim Discussions 37.
5
akt aru ahli n-naz ari al anna as la l-lugti innama huwa tawdu wa-s t ilh l
wah y wa-tawqf (Al-Xas is I 4041).
194 stefan wild
all languages.
6
Te plurality of human languages is sometimes seen as
due to a benevolent divine intervention: He (God) shaped every nation
conforming to a language in which He made them speak and which
He made easy for them.
7
God had made His Arabic revelation easy
for the Arabs (e.g. Sura 54:17); in a similar way, earlier revelations had
been made in other languages in order to make them easy for their
listeners.
Te most widely accepted historical model that explained the exis-
tence of languages other than Arabic was built on the supposition that
Arabic had been at a certain time the universal language. Te idea of
a universal common language can also be found in Gen 11:1, where,
however, this language does not have a name. For most Arabic scholars ,
Arabic was the earliest existing language.
8
But this universality had at
some point in history come to an end, and decadence, corruption, and
confusion ( fasd , tah rf, tabalbul) had set in. Tis statement was hard
to reconcile with the Qurnic assertion that God himself had created
the variety of human languages (Sura 30:22). Te reason for the devel-
opment of other languages was nevertheless frequently seen either in
a divine act of punishing Adam or mankind, or in a general confusion
(balbala ), which mixed up what had been one common human lan-
guage in a gradual process of corruption (tah rf ). Tah rf and tabalbul
were sometimes expressly linked to a punitive act of God, sometimes
they seem to be seen more as a general tendency of history. According
to Ibn Askir s chronicle, Arabic was Adams language in Paradise until
he disobeyed God. Ten God deprived him of Arabic and he started
speaking Syriac , evidently considered a lesser language or a corrupted
form of Arabic. When Adam repented, however, God gave the Arabic
language back to him.
9
A second less revelationist tradition was the following:
Te frst language with which Adam came down from Paradise was Ara-
bic . When the contact with Arabic became far and distant, Arabic was
corrupted (h urrifa) and became Syriac (suryniyya ) which is a Nisba
to ard sr or Suryna . Tis is the land of al-Jazra , where Nh and his
6
Van Ess , Theologie und Gesellschaft IV 324.
7
az-Zubayd , Tabaqt an-nah wiyyn (ed. Muh ammad Ab l-Fad l Ibrhm ), Cairo
1954/1373, 1 jabala kulla ummatin mina l-umami al lugatin ant aqahum bih wa-
yassara lahum.
8
lugatu l-arabi asbaqu l-lugti wujdan, as-Suyt , Muzhir I 28, 4.
9
as-Suyt , Muzhir I 30, 10.
divine, prophetic, and heroic arabic 195
people lived before the food. Syriac resembled Arabic, but it was cor-
rupted (muh arraf ), and Syriac was the language of all people in Nh s ark
except for one man, whose name was Jurhum , and his language was still
the original Arabic. When they lef the ark, ram b. Sm married one of
his (Jurhums) daughters. From them the Arabic language came down on
his ofspring Aws Ab d , Ubayl , Jir Ab T amd , and Jdis . d was
given the name of Jurhum, because he was their maternal forefather. And
Syriac stayed with the ofspring of Arfaxad b. Sm until it reached Yajub
b. Qaht n . He was in Yemen . Tere the Ban Isml settled and the Ban
Qaht n learned from them the Arabic language.
10
It is interesting to see that the term tah rf , which is the usual word
denoting the falsifcation of the revealed scriptures by Jews and Chris-
tians , is here used to explain a fact of imagined linguistic history : the
corruption and subsequent loss of Arabic . According to this account,
Nh s language was not Arabic, but Syriac . Arabic had survived only
with Jurhum and his tribe .
Abdalmalik b. H abb (d. 238/852) developed a stemma of prophetic lan-
guages . According to him, these prophetic languages are Arabic , Syriac ,
and Hebrew . All the sons of Israel spoke Hebrew, the frst to speak it
was Ish q . Syriac was the language of fve prophets : Idrs , Nh , Ibrhm ,
Lt and Ynus . And twelve of the Prophets spoke Arabic: Adam , Sht ,
Hd , S lih , Isml , uayb , al-Xidr , the three in surat Ysn i.e. the three
nameless messengers who were sent as the Companions of the City
(as h b al-qarya, sura 36:13f.), Xlid b. Sinn al-Abs , the legendary
forerunner of the Prophet Muhammad , and the Prophet Muhammad
himself.
11
Others added more and more details. According to al-Azraq , Nh
had eighty men with their families on his ark. When the ark came to a
halt on Mount Jd , Nh founded a village called Tamnn (eighty).
Te next morning, they found that their tongues/languages had been
confused and that there were now eighty languages, one of which was
Arabic . Tey did not understand each other any more.
12
Tis account
ofers no explanation for the emergence of these eighty languages, but
10
as-Suyt , Muzhir I 30,-6 according to Abdalmalik b. H abb .
11
Claude Gilliot and Pierre Larcher Language and Style of the Qurn. EQ 3, 109
135, especially The mythical narratives on the superiority of Arabic 118ff.
12
fa-btan qaryatan wa-sammh Tamnin fa-as bah d ta yawmin wa-qad tabal-
balat alsinatuhum al t amnn lughatan ih dh l-arabiyyatu (al-Azraq , Axbr Makka,
/ed. F. Wstenfeld ) vol. 1, 20.
196 stefan wild
is based on the idea that Nh and his followers originally did not speak
Arabic.
In other accounts, Ibrhm is said to have spoken two languages,
Syriac and Hebrew . When he fed from the persecution by Namrd in
Kt a in the land of Babylon , his language was Syriac (lisnuh suryn).
Namrd wanted to catch him and gave order that anybody speaking
Syriac should be arrested and brought before him. But when Ibrhm
crossed the Euphrates from H arrn , God changed his language and
Ibrhm miraculously started speaking Hebrew (ibrn ). He escaped
because his persecutors did not know this language.
13
A further version closer to the narrative of the Babylonian Tower is
reported by at -T abar : Namrd has a high building (s arh ) made, until it
reaches the sky. God destroys the building and on that day the languages
of people become confused from fright. Tey then speak in seventy-
three languages. Tis is why the place was called Bbil . Te language of
mankind before that was Syriac.
14
A diferent kind of etiological expla-
nation is linked to Yarub (see below section 4). Te etymological con-
nection between balbala and the name Bbil is ofen invoked and fnds
its way into Arabic lexicography .
15
Tis connection between the Arabic
word for confusion (balbala) and the name of Bbil is prefgured in the
Bible , in which the name of the city of Babel is linked to the Hebrew
verb balal, also meaning to confuse (Gen 11:7). Te number of exist-
ing languages is given in diferent ways. According to al-Masd , there
were seventy-two languages divided under Nh s sons: the descendants
of Sm spoke nineteen languages, the descendants of H m seventeen,
and those of Yft thirty six.
16
As far as I can see, nobody in pre-modern times ever claimed that the
prophet Ibrhm spoke Arabic although his unequalled importance
for the link between earlier monotheistic religions and Islam , especially
through his role in building the Kaba , would have made such a claim
attractive. Abraham is called a Muslim in the Qurn (h anf muslim
13
Ibn Sad , Tabaqt ed. Eugen Mittwoch, Leiden 1905, I:1 21, 14; a similar tradition
in Ibn Mut arrif at -T araf , Qis as al-anbiy no. 124.
14
fa-tabalbalat alsunu n-nsi min yawmaid in mina l-fazai fa-takallam bi-t alt ati
wa-sabna lisnan fa-li-d lika summiyat Bbil wa-innam kna lisnu n-nsi qabla
d lika s-suryniyyata., At -T abar , Trx I 322.
15
Lisn al-Arab s.v. bll: summiyat ardu Bbil li-anna llha h na arda an yuxlifa
bayna alsinati ban dama baatha r-rh an fa-h aarahum min kulli ufuqin il Bbil fa-
balbala llhu bih alsinatahum t umma farraqahum tilka r-rh u fi l-bild.
16
al-Masd , Murj ad-d ahab, ed. Barbier de Meynard I 78.
divine, prophetic, and heroic arabic 197
sura 3:67)but never an Arab. Tis claim was put forward only by zeal-
ous Arab nationalists . When in 1999 Pope John Paul II announced his
plan to visit Ur in Iraq , members of the ruling Bat -party criticized that
the Pope in his announcement had not spoken of Ibrhm as an Arab.
Tey argued: Ibrhm was born in the land of the Arabs and he lived
in Iraq, in the Sumerian city of Ur . When he was chased out of his place
of birth because of his monotheist creed, Ibrhm started his combative
journey through the lands of the Arabs, H arrn , Palestine , Egypt , and
Mecca . . . Combative was a favorite attribute of praise in the Bat -party .
Te Bat s do not expressly identify Arabic as Ibrhms languageeven
though the reader is forced to conclude that the language that the com-
bative Arab Ibrhm spoke must have been Arabic.
17
Such a linguistic
myth could be elaborated. Tere is frequently a tendency in modern
popular Arab discourse to call Arabic what is elsewhere called Semitic,
in order to extol the importance of Arabic. Te Canaanite tribes preced-
ing the Israelite settlers are then called Arab Canaanite tribes , and the
variant of Akkadian to be found in the Ebla texts is called Arabic.
18
For
the claim that the Aramaic of the Nabatean inscriptions is really Ara-
bic, see below.
3. Isml s Arabic
Next to Adam , the most important fgure with whom the introduc-
tion of Arabic is frmly connected is Ibrhm s son Isml . In one of the
most important foundational Muslim narratives , God orders Ibrhm
to migrate to Mecca with young Isml and the latters mother Hagar .
Ibrhm builds the Kaba together with Isml . Tere they meet mem-
bers of the (Arabic-speaking) Jurhum tribe . Isml grows up with their
children, learns how to shoot (the bow) and to speak in their language,
and he takes a Jurhum-wife.
19
Another report says, without reference to
where and how, Ismail learned Arabic: Sarah gave Hajar to Ibrhm, he
slept with her and she bore him Isml, who was Ibrhms eldest son.
His name used to be Iml which was later arabicized (wa-kna smuh
17
Amatzia Baram , Der moderne Irak, die Baath-Partei und der Antisemitismus in
Jahrbuch fr Antisemitismusforschung 12 (2003) 99119 p. 114.
18
Welt des Islams 21 /1982, 240f.
19
Ibn Qutayba , al-Marif, ed. T arwat Uka, Cairo 1960, 34.
198 stefan wild
mwyl fa-uriba).
20
Tis Arabic tradition closely follows the Biblical
account (Gen 21:20)even the bow is already thereand preserves
in Iml a dialectical Hebrew form of the classical Hebrew Yiml .
Abraham had given his son a Hebrew name, because he spoke Hebrew.
Te link between Isml and the Arabic language became part of the
awil-literature . Te Prophet had said: all Arabs descend from the of-
spring of Isml b. Ibrhm . Te question who was the frst to speak
Arabic could, therefore, be answered and the answer is contained in a
short didactic dialogue in which an Ab Jafar Quray b. Uqba b. Bar
asks and a Muh ammad b. Al answers:
Q: Who was the frst to speak Arabic ?
A: Te frst to speak Arabic was Isml b. Ibrhm , when he was thirteen
years old.
Q: And what did people speak before that?
A: Hebrew .
Q: And what was God s language that was sent down to his messengers
and servants?
A: Hebrew .
21
Here, Adam is not even mentioned as the frst recipient of Arabic . And
God s language of pre-Arabic revelation is simply Hebrew . In awil-
collections , the particular that Ishmael was the frst to write in Arabic is
not too frequent.
22
See below p. 200.
Te report that Isml picked up Arabic from Arabic-speaking tribes
was widely disseminated, but there was a competing tradition saying
that Isml received the Arabic language by revelation on the day he was
born. Te three other sons of Ibrhm stuck to their fathers (Hebrew )
language.
23
According to a H adt , the Prophet Muhammad recited (the
verse ) an Arabic Qurn for people who know (Sura 41:3); then he said:
Isml received this Arabic language by a great revelation (ulhima
Ismlu hd l-lisna l-arabiyya ilhman). Tis tradition contradicted
the view that Isml had learnt Arabic from the Jurhum tribe . In the
latter case, the tribe must have logically preceded Isml in speaking
Arabic. Another H adt mentions Ismls age: when he learned Arabic,
he was 14 years old.
24
20
Ibn Sad , Tabaqt I/1 23, 9.
21
Ibn Sad , Tabaqt I/1 24, 16. Tottoli p. 82. Cf. Sibt Ibn al-Jawz, Mirt 310; Ibn
Askir , Trkh II 331.
22
See the comment of ibl , Mah sin al-wasil f marifat al-awil 143.
23
Ibn Sad , Tabaqt I/1 24, 21.
24
as-Suyt , Muzhir I 22, 13.
divine, prophetic, and heroic arabic 199
Ibn Sad transmits a report that contradicts the idea that Isml was
born with Arabic and traces Arabic only to Ismls ofspring. Accord-
ing to this tradition, Isml never spoke Arabic, because in his flial
piety he did not deem it permissible to act diferently from his father.
According to this report, the frst of his ofspring to speak Arabic were
the Ban Rala bint Yajub b. Yarub b. Ldan b. Jurhum b. mir b. Saba
b. Yaqt n b. bir b. likh b. Arafxaad b. Sm b. Nh .
25
Nevertheless,
the fact that Isml forgot his fathers Hebrew is expressly stated in a
tradition traced to Muh ammad b. Salm:
26
Te frst one to speak Arabic
and to forget his fathers language was Isml. Muslim Arab tradition ,
therefore, agrees that Ibrhm s language was not Arabic, whereas there
is disagreement on whether it was Isml who was the frst to speak
Arabic and, if so, when and how Isml learned it. For the tribal aspects
of such traditions, cf. below under 4. Yarubs Arabic .
Te most frequently quoted account, successful as a canonical H adt ,
was, however, that Isml learned Arabic from the Jurhum tribe in his
youth (wa-shabba l-ghulmu wa-taallama l-arabiyyata minhum, i.e.,
Jurhum, s. Buxr , Sah h , Anbiy 21). He marries twice, in both cases
a wife from the Jurhum tribe . Tis again meant, of course, that Isml
was, strictly speaking, not the frst one to use Arabic, because he had
to learn it from somebody else. In some reports, the gif of the Arabic
language is mentioned next to other privileges of Isml: the tradition-
ists report that Isml was the frst one to speak Arabic, the frst one to
build the h aram afer his father Ibrhm , and the frst one to install the
rites of pilgrimage . He was also the frst to ride full-grown horses, which
before were wild and could not be ridden. Some say: Isml was the frst
whose tongue God opened to speak Arabic. And when he grew up, God
gave him the Arabic bow. Tis report implies that God revealed Arabic
to Isml and that Isml did not have to learn it from the Jurhum.
27
Tis report became part of Adab literature . At -T alib mentions it in his
Lat if al-Marif. Te frst person to speak Arabic was Isml , peace be
upon him; all the Arabs came subsequently from his progeny, except for
three tribes , those of Auz , H adramawt and T aqf . He was the frst to
25
Ibn Sad , Tabaqt I/1 24, 22.
26
as-Suyt , Muzhir I 32, -2.
27
al-Yaqb , Trx I 22.
200 stefan wild
ride horses, these mounts being originally wild horses which had never
been ridden before.
28
Tere was also some concern about the fact that the long time that had
elapsed between the period in which Isml had started speaking Arabic
and the time of the Prophet Muh ammad made it difcult to accept that
the Arabic language should not have changed during this interval. Te
Prophet gave the answer: Ismls language had been obfuscated (daru-
sat). But Jibrl came and made me retain the language and I retained it,
29
i.e. the Qurnic message saved Arabic.
In other versions, it is not only the Arabic language whose origin is
attributed to Adam or Isml, but also the Arabic script , which is in
this case imagined as written on clay tablets: Te frst who installed
the Arabic script, the Syriac script , and all other scripts was Adam, 300
years before his death. He wrote them in clay and baked them. When the
earth was hit by the food, each people received its script and used it to
write. Isml b. Ibrhm received the Arabic writing .
30
Some of these conficting mythological reports are woven together
and transformed into scholarship by modern Wahhabi scholars . Taking
such accounts as factually historical, Muh ammad Mus t af al-Az am is
led in a recently published book
31
to a re-writing of linguistic history
on a grand scale. He asks himself the question What language did the
Nabateans speak? And the answer is:
Growing up in Makkah from his earliest childhood Isml , eldest son of
Ibrhm , was raised among the Jurhum tribe and married within them
twice. Tis tribe spoke Arabic , and so, undoubtedly, must have Isml.
Te Jurhum Arabic probably lacked the sophistication and polish of the
Quraishi Arabic , preceding it as it did almost by two thousand years.
Ibn Ushta records a statement from Ibn Abbs , that the frst person to
initiate set rules for the Arabic grammar and alphabet was none other
than Isml. Eventually, Allah commissioned Isml as a messenger and
prophet, to call his people for the worship of the one true God Allah, to
28
In: C.E. Bosworth (ed.), The Book of Curious and Entertaining Information. The
Lat if al-marif of al-Thalib, Edinburgh 1968, p. 40, cf. Lammens , La cit arabe de
Taif la veille de lHgire, Beirut 1922, 5768.
29
as-Suyt , Muzhir I 35, 3.
30
as-Suyuti, Itqn IV 143 quoting Ab Bakr M. b. Abdallh b. Muh ammad b. Uta
al-Is bahn (d. 360) and his Kitb al-mas h if after Kab al-Ahbr: cf. as-Suyt , Bugya
59; as -S afad , Wf III 347; and Nldeke GdQ II 53 and GdQ III 1 fn. 2.
31
Muh ammad Mus t af al-Az am , The History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments, UK Islamic
Academy, Leicester 2003.
divine, prophetic, and heroic arabic 201
establish prayers and pay alms to the poor. Since Allah sends every mes-
senger in the language of his own people, Isml must have preached in
Arabic. Genesis credits Isml with twelve sons, among them Nebajoth/
Nabat ; born and nurtured in these Arabian surroundings they must have
adopted Arabic as their mother tongue. Tese sons may have preserved
their fathers message by using the prevailing Arabic script ; certainly, they
would not have resorted to whatever script was then current in Palestine
(Ibrhms homeland), since two generations had already lived in Arabia .
When Nabat subsequently migrated northwards he must have taken the
Arabic language and alphabet with him. It was his descendants who estab-
lished the Nabatean Kingdom (600 B.C.E.105 C.E.).
Al-Az am dates Isml and his early Arabic at around 1400 B.C.E. and
concludes: Te Nabatean language and script were . . . a form of Arabic
(121).
Such a sentence disregards that the Nabateans spoken language was
indeed Arabic , while the texts they wrote were in Aramaic . Te thesis
that Nabatean is a form of Arabic comes close to a linguistic panorama ,
which sees in all languages usually called Semitic a form of Arabic.
Te Arabic language and script , in their primitive forms, gave birth to
the Nabatean and most probably predated the Syriac (121) is but one
example of such a view.
4. Yarub s Arabic
A further source in tracing the frst Arabic speaker was tribal history .
Te Arabs , in their own self-view, were not only a linguistic community ,
but also marked by common ancestors. Descent was a primary symbol
of a tribal community , and the self-esteem and prestige of individuals
and communities were linked to the purity of descent. Te eforts of the
Arab genealogists to establish a link between living Arab tribal groups
and past forefathers were, of course, highly tenuous. Te skeptical obser-
vation of Ibn H azm (d. 456/1064) that on the face of the earth there is
no one whose descent from them is verifable did not prevent the emer-
gence of the most speculative lineages.
32
By and large, the Arab tribes
claimed descent from one of two ancestors, either a North Arabic origin
32
Ibn H azm , Jamharat ansb al-arab, ed. Lvy-Provencal , Cairo 1948, 8, quoted by
EI 2nd ed. I 546.
202 stefan wild
connected to the North Arab Adnn (also to Maadd and Nizr , later
Qays ) or a South Arabic origin connected to Qaht n (also called Yama-
nis , later also Kalb ). Some tribes such as d , T amd , ram , Jurhum ,
T asm and Jads were believed to have disappeared before Islam . Hostil-
ity between the ofspring of Qaht n and those of Adnn emerged afer
the advent of Islam in the form of the cleavage between the Ans r in
Medina and the Quray . Te fact that the Prophet Muh ammad belonged
to the Quray brought enormous prestige to the Adnn, i.e., the North
Arabs . Te narrative that linked Isml to Arabic was counterbalanced
by the argument that Isml had learned Arabic from the Jurhum, a
South Arabic tribe . While al-Masd accepted that Isml had been
given Arabic by God , he did not deny that Yarub b. Qaht n , the ances-
tor of the Yamanis , was the frst to speak Arabic. Which of the three
Arab group, the extinct Arabs, the Nizris , or the Yamanis was the frst
to speak Arabic? Te Nizri and Yamani groups vigorously upheld their
own claims to this honour. Te Nizris held that Isml was given the
language by God, while the Yamanis contended that Isml had learned
the language from a Yamani tribe living in Mecca .
33
Tis confict of
claims masked a deeper social and political confict between the two
groups. Al-Masd , while granting that Yarub Ibn Qaht n , the ancestor
of the Yamanis, was the frst to speak Arabic (Murj sec. 71), believes
that Ishmael too was granted this honour by God, independently of his
association with the Yamanis.
34
Te tribal family trees set up by Arab
genealogists always have an agenda. When names are inserted or omit-
ted, insertion and omission usually serve a purpose. When in the list
of Qaht ns ancestors there are two new names, those of Yajub and
Yarub, inserted between Saba and Qaht n (Yoqtan) in a family tree
that is otherwise based on Gen 10, 132, this happens because these
two names fulfll two important functions. Yarub symbolizes and per-
sonalizes the change from Syriac to pure Arabic, while Yajub/Yaman
gives his name to the land Yaman . Te appearance and etiological func-
tion of persons with such telling names is a common feature of Arab
genealogy .
35
A tradition quoted by Yqt and traced to ad-Dnawar s Mujlasa
links Bbil with Yarub and Arabic as the heavenly language: When God
33
Tarif Khalidi , Islamic Historiography. The histories of Masudi, Albany 1975, 116;
conflicting claims are set forth in Tanbh 7983 and Murj 99699.
34
Tarif Khalidi , Islamic Historiography ibid.
35
Manfred Kropp , Geschichte der reinen Araber II 379f.
divine, prophetic, and heroic arabic 203
assembled mankind in Babylon , He sent them an Eastern, a Western, a
Southern, and a Northern wind. And He assembled them in Babylon
and they assembled to see why they had been called. Suddenly, a voice
called out: He who has the West on his right, the East to his lef, and
faces Mecca his is the language of the people of heaven. Ten Yarub b.
Qaht n stood up, and the voice said: Yarub b. Qaht n b. Hd , you are
the one. And he was the frst to speak Arabic.
36
According to Ibn Durayd , the pure Arabs were seven tribes d ,
T amd , Imlq , T asm , Jads , Umayyim und Jsim . Most of them have
vanished except for some remnants dispersed in the sub-tribes . And
Yarub , whose name was Muhazzim b. Qaht n , received his byname
because he was the frst whose tongue switched from Syriac to Arabic .
Tis is the meaning of al-Jawhar s statement in as -S ih h : the frst to
speak Arabic was Yarub b. Qaht n .
37
Here, the eponymon Yarub is
linked to the root rb, another example for the favorite method of the
genealogists of connecting proper names with imagined history, ety-
mology with linguistic history .
In this mixture of traditions, Yarub b. Qaht n is fnally also located in
Babylon and celebrates his virtues in the style of a poetic self-praise.
Te majority of historians among them, the author of Tawrx al-
umam (al-Is fahn ), and the author of al-Marif (Ibn Qutayba ) tell us
that Yarub b. Qaht n was the frst to speak in clear Arabic , the frst to
bear the crown in Yaman , the frst who was greeted by his sons with the
formula May you avoid the curse and your day be happy!
Ibn Hishm , the author of the Book of Crowns, which he wrote on
the kings of H imyar reports: It was Yarub who went with the Arabs
to Yaman and settled there. Terefore Yaman was called afer him,
because Yarubs name was Yaman. Al-Bayhaq says: Yarub was the
frst to speak in clear Arabic and lef the confused way of talking used
by the Arabs. He and his ofspring were successful and God bestowed
on them the land of Yaman as heritage. To them belong the kings of
Tubba, who conquered the regions of the earth. On the strength of this
privilege they are ranked before Adnn , while the Adnn can boast the
prophethood of Muhammad . . . . At the time, in which the languages
had already been confused, the Ban H m had come to the highlands of
36
Yqt , Mujam al-buldn ed. F. Wstenfeld , I 447, 19.
37
as-Suyt , Muzhir I 31, -3.
204 stefan wild
Yaman. . . . And Yarub was the most high-minded youth in Babylon and
spoke clear Arabic, while the others spoke in confused language.
38
And
Yarub recited the following verse:
I am the youth favored by the richest gif, the happy one, well known for
his virtue. I am the son of Qaht n , the infuential and rich, I spoke in
Arabic , while the people were in (linguistic) confusion. (I spoke) in the
clearest unambiguous language and in the perfect language of the king-
doms afer me.
an l-gulmu d n-nas bi l-ajzali / al-aymanu l-marfu bi-t-tajammuli /
an bnu Qah t na l-hammmi l-aqyali / arabtu wa-l-ummatu f tabalbuli /
bi-l-mant iqi l-abyani gayri mukili / wa-mant iqi l-amlki badi l-kamali.
Tis is pure poetic self-praise in the rajaz meter as we know it from the
earliest Arabic poetry . According to this report, Yarub b. Qaht n was
the frst to speak clear Arabic and turned away from the confused ver-
sion of Arabic (namat u l-arabiyyati l-mubalbalati). He let his sons greet
him with the greeting abayta l-lana (may you avoid the curse!) and
with good morning! Tis Arabic is not so much seen as a prophetic
language but as a heroic language spoken by a tribal noble forefather.
Te hero of this poem , Yarub, praises himself for speaking pure Arabic,
a royal language of noble kings, while lesser Arab mortals used mixed
and confused languages.
In a further verse, Yarub predicts the coming of the Prophet
Muhammad :
Muh ammadu l-hd n-nabiyyu l-mursalu / li-llhi darru l-mjidi l-mus-
taqbili
Muhammad , the guide, the God -sent Prophet / how praiseworthy is the
blessed one who is coming.
Such a tradition combined tribal history with revelation . And to round
things out, the report claims that the same Yarub was the frst to recite
Arabic poetry and to put it into meters. He invented the poetic genres
and composed praise-poetry , self-praise, and love poetry .
39
Earlier tra-
ditions see the Ban Hshim and the Ban Yarub in ferce competi-
tion. When the rival of the Prophet Muh ammad , Musaylima the Liar,
who traced his descent to the Ban Yarub , heard of Muh ammads death,
he hoped to outstrip the Prophet and is said to have recited the verse :
38
Kropp , I Arab. text 9ff.
39
Kropp 11, transl. 149.
divine, prophetic, and heroic arabic 205
Passed away the prophet of the Ban Hshim / and rose up the prophet
of the Ban Yarub!
40
5. Te Prophet Muh ammads Arabic
As we have seen, afer the advent of Islam , the unique character of Ara-
bic was extrapolated by theologians , philosophers , and genealogists back
to times immemorial. In their view, afer the revelation of the Arabic
Qurn , Arabic became even more of a special language. Arabic itself,
not only its Qurnic form, was something sacred, superhuman. Yet, in a
famous description, a-f (d. 204/820) explained the special charac-
ter of Arabic mainly by its most important hermeneutical challenge, i.e.,
its vast and ramifed vocabulary. According to him, this was the primary
reason for the privileged status of Arabic:
We do not know that any man except for a prophet can claim to know the
Arabic language completely. However, nothing Arabic escapes the Arabs
collectively, so that there would be nobody among the Arabs who knows
it. Te knowledge of Arabic among the Arabs resembles the knowledge
of the traditions (sunan) among the scholars of religious Law . We do not
know anybody who could claim to know all traditions, so that not a single
one would escape him. But when the knowledge of traditions of all schol-
ars is collected, the whole of tradition is found. When the knowledge of
each of these scholars is divided, something of the traditions will escape
each of them. But what escapes him will be found with someone else.
41
Te Prophet , who as an individual can claim to know Arabic completely,
is, of course, Muh ammad . His knowledge of Arabic is superhuman. But
the Arabs collectively also know the whole language.
Te Prophet s own mastering of Arabic was related to the Qurnic
revelation , as well. When a man admires the Prophets rhetorical tal-
ent and says, What a good speaker you are: We have not seen anybody
speaking better Arabic than you! the Prophet answers: Tis is my right.
For the Qurn was revealed to me in clear Arabic.
42
And the Prophet
could boast: I am the best Arabic speaker (an afs ah u l-arab
43
or an
40
Ibn Kat r , Bidya vi, 341, quoted according to M. J. Kister , Musaylima in EQ 3
(2003) 462.
41
A-fi , Risla, ed. Ah mad Muh ammad kir , Cairo 1940, 42, 8ff.
42
as-Suyt , Muzhir I 35, 6.
43
Muzhir I 209, 2.
206 stefan wild
arabukum).
44
It is not always clear whether such statements refer to
Qurnic revelation or to the Prophets non-Qurnic words . In the same
vein, the Prophet was immune against linguistic mistakes (lah n ) in his
speech. He was reported to have said: I am from the Quray and I grew
up among the Ban Sad how should I commit a linguistic error? (an
min Quray wa-naatu f ban Sad wa-ann fyya l-lah n).
45
Te Ban
Sad b. Bakr b. Hawzin were the clan of Muh ammad s wet-nurse. In this
case, the Prophets claim to linguistic excellence was defnitely rooted in
his tribal background and not in divine grace. Among the further signs
of the Prophets unprecedented eloquence was that he was said to have
coined expressions that allegedly had never been used in Arabic before
him and that later became part of an elevated Arabic style, such as mta
h atfa anfh, he died a natural death.
46
In some traditions, the Prophet became a second Adam . In a H adt ,
he said: My community appeared to me in water and clay. And I was
taught all names even as Adam was taught all names.
47
Arab grammarians connect the linguistic quality of the Arabic of the
Quray with pre-Islamic times and with the role the Quray played at
the Kaba in Mecca . A privileged position vis--vis the Arabic language
was also conferred on the companions of the Prophet they also spoke
nothing but the purest Arabic. Te proof was that they never coined
a new word (m alimnhumu s t alah al xtiri lugatin aw ih dthi
lafz atin lam tataqaddamhum).
48
Tis was most probably connected to
the fact that the companions were the guardians of the Prophets word
and deeds. If their Arabic had been doubtful, it could have infuenced
the correctness of tradition. In a comparable manner, the ethical, philo-
sophical, and religious letters and pronouncements, ascribed under the
title Nahj al-balga (Te Method of Eloquence) to Al b. Ab T lib ,
were considered by many Shiites in essence and style as second only to
the Qurn .
49
44
Ibn Sad , Tabaqt I 113.
45
Ab t -T ayyib al-Lugaw , Martib an-nah wiyyn, ed. Muh ammad Ab l-Fad l
Ibrhm , Cairo n.d. <1955>, Maktabat Nahd at Mis r 6,1.
46
Muzhir I 209, 5.
47
Muzhir I 35, 4.
48
Muzhir I 10, 9.
49
Nahj al-Balgawa-huwa majm m xtrahu -arf Ab l-H asan Muh ammad
ar-Rd b. al-H asan al-Msaw min kalm Amr al-muminn Ab l-H asan Al b. Ab
T lib, ed. S ubh S lih , Beirut 1982; Moojan Momen, An Introduction to Shii Islam, The
History and Doctrine of Twelver Shiism, New Haven/London 1983, 25.
divine, prophetic, and heroic arabic 207
6. Te fate of Arabic afer the Islamic conquest
Te Arabic language celebrated its greatest triumph with the vast expan-
sion of the Islamic conquest and the concomitant expansion of the Ara-
bic language. But most Arab grammarians and lexicographers did not
see it that way. In an almost incredible volte-face, they saw the fate of the
Arabic language sealed and its purity doomed to corruption precisely
at the moment when the Muslim faith and the Qurn won over the
world:
wa-lam tazal il-arabu tant iqu al sajiyyatih f s adri islmih wa-md
jhiliyyatih h att az hara llhu l-islma al siri l-adyni fa-daxala
n-nsu fhi afwjan wa-aqbal ilayhi arslan wa-jtamaat fhi l-alsinatu
l-mutafarriqatu wa-l-lugtu l-muxtalifatu fa-fa l-fasdu f l-lugati l-ara-
biyyati.
50
Te Arabs did not cease to speak Arabic according to its disposition in
the beginning of the era of Islam and in the pre-Islamic past until God let
Islam prevail over all other religions. Ten people entered Islam in masses
and turned to it in focks. In Islam all separate languages and diferent
tongues came together and therefore corruption ( fasd ) spread in the
Arabic language.
Tis rather anticlimactic statement marks the beginning of Arabic as
the language of an Islamic world civilization . For many an Arab gram-
marian and lexicographer , this period was at the same time the starting
point of its decadence and corruption.
7. References
Agha, Saleh Said and Tarif Khalidi . 2002/03. Poetry and Identity in the Umayyad Age.
Al-Abhth 5051, 55119.
al-Az am, Muh ammad Mus t af . 2003. Te History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments. Leicester: UK
Islamic Academy.
al-Azraq . 1858. Die Chroniken der Stadt Mekka (F. Wstenfeld, ed. ), vol. 1 Axbr Makka.
Leipzig.
Czapkiewicz, Andrzej . 1988. Te Views of the Medieval Arab Philologists on Language
and its Origin in the Light of Al-Suyutis Al-Muzhir. Krakow: Universitas Jagellonica
Acta Scientiarum Litterarumque CMIX.
50
az-Zubayd , Tabaqt an-nah wiyyn wa-l-lugawiyyn (ed. Muh ammad Ab l-Fad l
Ibrhm ), Cairo 1954, 1.
208 stefan wild
EQ 20012006 = Jane McAulife, ed. Encyclopaedia of the Qurn, vol. 15. Leiden:
E. J. Brill
Fischer, A. and A.K. Irvine . 1978. K aht n. EI 2nd ed., 4: 447449.
Gilliot, Claude and Pierre Larcher . 2003. Language and Style of the Qurn. EQ 3,
109135, especially Te mythical narratives on the superiority of Arabic, 118f.
Goldziher, Ignaz . 1873. Beitrge zur Geschichte der Sprachgelehrsamkeit bei den
Arabern. Sitzungsber. d. Kaiserl. Ak. d.W. Wien, Phil-Hist.Kl. 73, 51152.
Ibn H azm . 1948. Jamharat ansb al-arab. Ed. . Levy-Provencal, Cairo.
Ibn Mut arrif at -T araf , Ab Abdallh M. b. A. al-Kinn . 2003. Qis as al-anbiy (Te
Stories of the Prophets). Roberto Tottoli, ed. Berlin: Islamkundliche Untersuchungen
253.
al-Jumah Muh ammad b. Sallm 1916. Tabaqt a-uar. Joseph Hell, ed. Leiden.
Khalidi, Tarif . 1975. Islamic Historiography. Te histories of Masudi. Albano.
Kropp, Manfred . 1975. Die Geschichte der reinen Araber vom Stamme Qah t n aus dem
Kitb Nashwat at-Tarab f trkh jhiliyyat al-arab des Ibn Sad al-Maghrib, hrsg. u.
bersetzt von Manfred Kropp , Bd. I: Einleitung und Text, Bd. II: bersetzung und
Anmerkungen, Diss. Heidelberg.
Loucel, Henri . 1963. Lorigine du langage daprs les grammairiens arabes. Arabica 10 I
188208; II 253281; 11 (1964) III 5772; IV 151187.
Rubin, Uri . 1990. H anfyya and Kaba. An inquiry into the Arabian pre-Islamic back-
ground of dn Ibrhm. JSAI 13, 85112.
as-Suyt , Abdarrah mn Jalladdn . n.d. al-Muzhir f ulm al-luga. Muh ammad Ah mad
Jd al-Mawl Bek , Muh ammad Ab l-Fad l Ibrhm, Al Muh ammad al-Bjw, eds .
Cairo n.d., 2nd ed., Dr Ihy al-Kutub al-arabiyya, vol. 12.
Tottoli, Roberto, ed. 2003. Te Stories of the Prophets by Ibn Mutarrif al-Taraf, edited
with an introduction. Berlin, 253.
. 2002. Biblical Prophets in the Qurn and Muslim Literature. Richmond.
Weiss, Bernard G. 1974. Medieval Muslim Discussions on the Origin of Language.
ZDMG 124, 3341.
INFLECTION AND GOVERNMENT IN ARABIC ACCORDING
TO SPANISH MISSIONARY GRAMMARIANS FROM
DAMASCUS (XVIIITH CENTURY): GRAMMARS AT THE
CROSSROADS OF TWO SYSTEMS?
1
Otto Zwartjes
University of Amsterdam, NIAS
1. Introduction
Grammars of Hebrew written in Europe in the Renaissance by Chris-
tians could beneft from the Hebrew grammatical tradition . Johannes
Reuchlin (14551522) quotes in his De rudimentis hebraicis (1506)
Priscian and Rabbi David (i.e. Qimh i) as well (Law 2002, 247248). In
grammars of Sanskrit written in Europe we see also that the framework
of grammatical description has been derived from the Indian
grammatical tradition . Te German Jesuit Heinrich Roth (16201668),
as Hauschild (1988, 1314) observes, uses with perfect familiarity the
technical terms of Indian grammar [. . .] Roth stands entirely within
the Indian grammatical tradition , and probably he used the practical
grammar of Anubhti Svarpcrya , called the Srasvata Vykarana,
which was in general circulation in Hindustn , Bihr and Benares . [. . .].
Another candidate, though a less likely one, would be the grammar
called Mugdhabodha, which was composed in the second half of the
13th century by Vopadeva , but the usage of which was more common
1
Tis article is an elaborated version of paper delivered at the IIIrd International
Conference on Missionary Linguistics , Hong Kong and Macau , 12th15th March 2005.
Te organization of the conference and participation in Hong Kong has been made pos-
sible by fnancial support of the Norwegian Research Council (Norges Forskingsrd)
and the Language Centre of the Hong Kong University of Science and Technology. I
have to acknowledge Kees Versteegh who inspired me to do research in the feld of the
History of Linguistics and particularly Michael Carter for his valuable corrections and
suggestions. Tanks to my colleague Maria Cndida Barros , I came across the reference
to the grammar of Lucas Caballero . Research has been made possible by the Radboud
University (TCMO) where my 2 research on Pedro de Alcal started. I continued this
research topic at the University of Oslo , supported by the NFR-project OsProMil (Oslo
Project on Missionary Linguistics). I am grateful to the Rogge Library (Strngns) for
the reproduction of the MS. I gladly acknowledge Pierre Winkler for his translations
from Latin.
210 otto zwartjes
in Bengal . . . Te work of Roth was a masterpiece, which does not difer
considerably from current grammars , which similarly depend on the
linguistic achievements of India s own grammatical tradition (Hauschild
1988, 1314). In the grammars of Japanese written by the Portuguese
Jesuit Joo Rodrigues (15611634), particularly in his description of
particles and verbal endings , we can also fnd information concerning
the study of tenifa or tenivofa (the study of particles and verbal
endings) from contemporary Japanese scholars (Maruyama 2004, 155).
As has been demonstrated by Gregory James (2007), some missionaries
describing the Tamil language , such as Bartholomus Ziegenbalg (1682
1719), were familiar with some ancient Tamil works on grammar .
Tese missionaries worked under favourable conditions compared to
their colleagues who described languages, particularly Amerindian and
Austronesian , without any written tradition or an adequate indigenous
grammatical framework they could rely on.
Te use of Arabic grammatical terms in the frst grammar written in
Europe of vernacular Arabic , the Arte para ligeramente saber la lengua
arabiga (1505) of Pedro de Alcal (Order of St. Jerome ) has been the
subject of an article written by William Cowan (1981). In de Alcals
grammar, some technical terms were incorporated in the descriptive
framework, including terms such as damir , temiz , masdar , amr and
xucla . In this article, Spanish grammars of Arabicvernacular and clas-
sical written by Franciscans in Damascus and completed, copied, or
printed in Spain in the 18th century occupy our attention, particularly
the grammars of Francisco Caballero and Juan de la Encarnacin (18th
century) and Francisco Caes (17301795). Of the frst we have an
unpublished manuscript, which has escaped the attention of research-
ers until today,
2
and of the second a printed work has been conserved,
2
Bibliographical information concerning Francisco Caes can be found in Schnur-
rer (1811, 79, no. 113) BICRES III, and in Monroe (1970) we fnd some historical back-
ground. However Caes is not mentioned by Fck (1955), Dannenfeldt (1955), and
Killean (1984) and particularly the grammar of Caballero has been neglected by all.
Afer a century-long period of silence and total neglect, an important monograph on
Bernardino Gonzlez appeared recently, together with a facsimile edition of the diction-
ary (Intrprete arbico) and his grammar (Eptome) (Lourido Daz 2005), not long afer
this paper had been delivered in Hong Kong . When the proofs were almost ready for
publication, I received a copy of this monograph, courtesy of Emilio Ridruejo . Lourido
Daz (2005, I, 2122) had traced seven manuscripts of the dictionary, and six copies
of Eptome de la gramtica rabe made by Bernardino Gonzlez pupils, probably for
their own use. One copy was completed by Blas Francisco de Salamanca in 1704, the
second by Lucas Caballero and Juan de la Encarnacin between 1709 and 1710, the
so-called Tingstadius copy . Two copies were compiled in 1719, one from the El Escorial
inflection and government in arabic 211
studied sporadically by dialectologists describing the urban colloquial
Arabic of Damascus (Lentin 1997), but which also escaped the atten-
tion of researchers working in the feld of the history of linguistics . In
the Damascene Franciscan tradition , the authors used a great number
of Arabic grammatical terms , adapted in a hispanicised form, which
substituted, accompanied or sometimes superseded Latin terminology :
examples are: arabicacin (
l
munde
vocative
vocatiuo mund
z arf
adverb
ablatiuo z arf
As William Cowan observed:
10
they are not true cases in either the Arabic or the Latin sense of being an
infection added to a noun , a fact that Alcal was quite aware of, but are
regarded by him in the same way that modern theoreticians of case gram-
mar regard the syntactic positions in an utterance. Such relations are not
expressed by infections, since colloquial Arabic has none, but through
abitudines or conocimientos . Te markers of the equivalents of these Latin
cases are in fact combinations of prepositions and particles and the def-
nite article (lil, maal, lal, etc.) (Cowan 1981, 359).
9
La declinacion de todos los nombres arauigos es vna solamente. Porque todos los
nombres arauigos son inuariables (Alcal 1505, capitulo nono).
10
In the original text, a small hamza in superscript is placed on the frst vowel u.
inflection and government in arabic 215
Te information given by William Cowan , however, is not complete,
since translations from his dictionary are not included here. Although
we do not fnd all the terms in his dictionary, we do fnd some of them:
mubted is not only translated by Pedro de Alcal as nominative , but
also as principio de oracion (Corriente 1988, 12), which is not without
importance (see below). zm mudf is rendered as posesivo nonbre.
11
Te term abitud seems to be inherited from other grammars than
the Latin and Castilian grammars of Antonio de Nebrija (c. 14441522).
Te term habitudo is used by Ferdinandus Nepos in his Materies (com-
pleted between 1469 and 1485) and Juan de Pastrana Compendium
grammatica (1462).
12
Item per hanc regulam primo iuvenes component per unum casum tantum
sic dicendo: la tierra terra; de la tierra terre, etc.; uel per duos, sic: la tierra
del rey; uel per tres et quatuor et amplius, sic: o leyente la lection a los sco-
lares en el general de las escuelas componitur o legens lectionem scolaribus
in generali scolarum, dando cuilibet casui propriam habitudinem, inter-
rogando cuius casus, numeri et cuius declinationis hoc principio. (Codoer
2000, 90).
However, it is not so clear what Pedro de Alcal s exactly means with
the term abitud (pl. abitudines ), since he uses it as a synonym of the
defnite article (Nota que porque ay algunas abitudines en cada vno delos
casos que en alguna manera parescen preposiciones, porque se preponen
alos casos, avn que en verdad no lo sean, mas son articulos) (Alcal
1505[1971], 26). In another section, Alcal uses the term as a synonym
of preposicion , since according to his observations, the maxrr case
(see below) has the four abitudines la, lal, li, lil, whereas the term
abitud is not used for ba, bal, bi, bil, fa, etc. which are described in
the paragraph on the z arf case (Alcal 1505[1971], 27). Here they are
just called prepositions and not abitudines and there is no explanation
11
Other terms not mentioned by Cowen , are jezme , translated as consonante (letra
mazjun [sic] (Corriente (1988, 34), and mdde , translated as acento (Corriente 1988,
189).
12
With this rule, the novices first build constituents with only one case [= head of
the NP], saying la tierra terra, de la tierra, terre, etc., or with two cases [= head
+ complement of the NP], as la tierra del rey, or with four and so on, as: reading the
lesson in general for the scholars of the schools, assigning the appropriate habitudo
[= grammatical form] to whatever case [= grammatical function]. The term habitudo
casualis is also used by Nepos in relation with government: Haec enim regula maxima
est in construendo [. . .] quia talem casum regit dictio qualis fuerit habitudo casualis.
(ibid.).
216 otto zwartjes
why he used these diferent terms. In other sections, Alcal uses abitud
as synonym of conocimientos (see also Zwartjes 1993, and 1994).
In Arabic grammatical theory a nominal sentence can be divided into
topic for which mubtada
al-l-mubtada
(what is built
upon what is begun with) (Owens 1990, 45). Mudf is the word Pedro
de Alcal uses for the genitive , which in Arabic tradition means literally
what is added, i.e. possessed (Owens 1988, 34; 1990, 104). Majrr from
the same root as jarr (see below), means pulling, or governing the -i
infected form. Pedro de Alcal follows in his sections about the prepo-
sitions the Latin system and tries to apply Arabic terms to them, without
realizing that in the dialect he describes, case-endings are not used, and
without realizing that in classical Arabic nominal declension , there are
three infectional vowels , the -u, the -a, whereas for the verbal infection
the three vowels -u, -a and (zero ending) can be distinguished . Pedro
de Alcal did not take the Arabic infectional endings as starting point,
but the Latin prepositions in alphabetic order: prepositions + accusa-
tive , prepositions + ablative , etc. and at the same time he translated the
names of these Latin terminology into Arabic:
Capitulo XXXII. De las preposiciones.
Hallamos en el Arauia todas las preposiciones que en la gramatica [latina],
y ayuntadas a essos mesmos casos, que son maful y darf (que son acusatiuo
y ablatiuo ), y son las del acusativo las siguientes:
A ad apud circa circa ante longe ante [. . .]
A al aynd carib qued acbal bad cudim [. . .]
Las preposiciones del darf (que es ablativo) son las siguientes, conuiene
saber:
Con con con con en en en en
Ba bal bi bil fa fal f fl . . .
(Alcal 1505[1883], 26).
Chapter XXXII. About the prepositions.
We fnd in the Arabic language all the same prepositions as in Latin gram-
mar , and they are combined with the same cases, which are maf
l and
z arf (which are accusative and ablative ), and those which can be combined
with the accusative are the following:
A. ad apud circa circa ante longe ante [. . .]
A al aynd carib qued acbal bad cudim [. . .]
inflection and government in arabic 217
Te prepositions of the darf (which is ablative ) are the following:
con con con con en en en en
ba bal bi bil fa fal f fl . . .
As we can see, Pedro de Alcal did not separate the prepositions from
the defnite article (al).
Maf
l ma
ahu
accompaniment object, maf
l minhu
object from it. (Owens 1990, 160). For the locative object, instead of
maf
l fhi the term z arf is also used (Owens 1990, 51,141151), which
is the term Alcal uses here for the ablative . Mund is the direct trans-
lation of vocative . According to Cowan (1981, 360), Alcal was appar-
ently trying to make unfamiliar material intelligible to his audience, but
at the same time to avoid a direct equation with the Latin categories. In
fact, the frst might be true, but we must be aware that in his paradigm of
the case-system , we do fnd an equation with Latin cases, and we never
fnd any traces of the four traditional Arabic infectional endings , -zero,
-u, -a, and i and never the original Arabic names for these infectional
endings are introduced here.
Other technical grammatical terms in Alcal s grammar are: amir
(conocimiento ;
13
cl. Ar . damr conjunct pronoun ), temiz (cl. Ar . tamyz
accusative of specifcation),
14
amr (imperatiuo ; cl. Ar .
amr impera-
tive ), xucla seal ; cl. Ar . akl(a) orthographic sign :
15
Es otrosi de notar, que los Arauigos non tienen letras vocales como los Lati-
nos, mas tienen en lugar dellas ciertas seales, que ellos dizen xclas, con
las quales y con todos los caracteres suso dichos leen y escriuen lo necessario
(Alcal 1505[1883], 4).
It has to be observed that the Arabs do not have the letters for the vowels
as the Latins , but instead of them, they have certain signs, which they call
xuclas , and with all the above-mentioned characters they read and write
anything which is necessary.
13
Cf. Zwartjes (1994).
14
Tem is also translated as conocimiento in his dictionary (Corriente 1988, 197).
15
In his dictionary Alcal translates the word alma as (signo por) seal (Corriente
1988, 140). The seal is also used in his grammar for the article (seal de demostracin ),
so this term could be the technical grammatical term, but as happens often in these
grammars , it is not always possible to distinguish between language and metalanguage .
218 otto zwartjes
Te origins of the mnemonic vowel terms with and without tanwn
(the nunated forms) which expresses indefniteness in Arabic are
unknown to me and this could be a local teaching method, not recorded
elsewhere:
TABLE 2 THE VOWELS AND THE NUNATED FORMS
ACCORDING TO PEDRO DE ALCAL (1505[1883], 4)
16
minib fath a a
minib kasra i
minib damma u
minenbn fath atn an
mininbn kasratn in
minunbn dammatn un
minb sukn
16
alif skin
silent
t Dativus
al-mutak Accusativus
al-mund Vocativus
al-mustaqt i
Ablativus
23
3. Te grammars of Lucas Caballero, Juan de la Encarnacin, and
Francisco Caes : Grammars at the crossroads of two systems?
3.1 Introduction
Afer the foundation of a great number of missions in the East, the
importance of the Arabic language for preaching the Christian faith
continued to increase. Paul V in a papal bull dated 1610 had commanded
the various religious orders to teach Oriental languages in their colleges.
In the early 16th century, Arabic was taught in Seville at the Colegio
trilinge.
24
Te Franciscans decided to found colleges in Salamanca , Alcal ,
Paris and Toulouse for the teaching of Arabic , Greek and Hebrew (Monroe
1970, 26). According to Monroe, Bernardino Gonzlez (c. 16651735)
composed an Arabic dictionary in Seville, which was completed by
Franciscans in Jerusalem in 1709 (Monroe ibid.), an unpublished work.
Jos de Len began to compile a new dictionary of Arabic and Bernardino
Gonzlez was sent to Damascus in order to complete his work.
25
As Monroe
23
From the verbs samm to denominate, walada to give birth,
t to give, ak to
complain, nad to call and qat t a to cut of, to disjoin, the tenth derived form means to
deduct, which seems to be an approximate translation of auferre.
24
In this short account, there will be no space to summarize the study of languages
during the Middle Ages in Muslim Spain .
25
Franciscans had already arrived in 1233 in Damascus . Propagating the Christian
faith was not permitted by Sultan Malik al-Araf , but they took care of Europeans who
settled there. After several cases of martyrdom and imprisonment, the Cadi of the city
222 otto zwartjes
observed, the work was the most complete ever to have been composed
in Spanish afer that of Pedro de Alcal . [. . .] Te work of Gonzlez and
his collaborators was lost until Asn Palacios (1901) came across a copy
containing additions up to the year 1727 made by the friars of Damascus
(Monroe 1970, 27). One of the successors of Bernardino Gonzlez was
Francisco Caes who settled in Damascus at the Spanish Franciscan
College in 1757 (Monroe 1970, 28). Caes grammar of Arabic has been
printed in two diferent editions, a frst in 1775 (Madrid , Don Antonio
Perez de Soto , and a second in 1776 (Schnurrer 1811, no. 113, BICRES
959 and 971), entitled Gramatica arabigo-espaola, vulgar y literal. Con
un diccionario arbigo-espaol, en que se ponen las voces mas usuales
para una conversacin familiar, con el texto de la Doctrina Cristiana en
el idioma arabigo.
I came across another copy from the same Franciscan tradition , com-
posed by Lucas Caballero and Juan de la Encarnacin as we can read in
the colophon, which escaped the attention of scholars who have worked
in this feld. Te manuscript has been identifed by Magnus and Aare
Mrner in his Spanien i svenska arkiv. Te title of this manuscript is Com-
pendio de los rudimentos y gramtica rabe en que se da notizia de la len-
gua verncula y Vulgar y algunas reglas de la literal Iustamente, 1709, and
in the colophon 1710 (another author, Juan de la Encarnacin, fnished
the text San Diego , Seville ). Te work is based on Bernardino Gonzlez
as we can read in the title, and Lucas Caballero , lector actual Arabo en
el Colegio de Damasco composed (recopilado) this manuscript, which
has been donated by Johan Adam Tingstadius (17481827),
26
bishop of
Strngns , Sweden , from 1803, to the Rogge library, which has belonged
administratively to the Royal Library of Stockholm since 1968.
As the titles of the grammars of Caes and Caballero demonstrate, the
language under description is not only classical Arabic , but the urban
dialect of Damascus . Apart from Alcal s grammar of colloquial Ara-
bic of Granada , European scholars usually did not pay much attention
to lower registers, so the linguistic works of these Franciscans work-
granted the privilege of being able to open a public chapel and in 1668 the Francis-
cans established themselves in a Maronite church, which they left in 1719 when they
acquired a new church in the Christian Quarter of Bb Tma . The foundation of the
college where Arabic was taught dates from this period.
26
Tingstadius was a professor in Oriental languages at Uppsala . He published, for
instance in 1770, a treatise entitled Dissertatio philologica de natura et indole linguarum
orientalium communi (Uppsala: Johan Edman ) and in 1794 his Dictiones arabicae ex
carmine Tograi, hebraismum biblicum illustrantes. Uppsala: Johan Edman .
inflection and government in arabic 223
ing in Damascus are of great importance. However, they were not the
only grammarians who described non-Classical registers . Antonio ab
Aquila s grammar published in 1650 is not only a grammar of classical
Arabic (ad grammaticae doctrinalis intelligentiam) but also colloquial
Arabic (ad vulgaris dumtaxat idiomatic), probably the reason why he
called the grammar Arabicae linguae novae et methodicae institutiones.
Dominicus Germanus (Germanus of Silesia ; 15881670)
27
composed a
dictionary in 1636 with the title Fabrica overo Dittionario della lingua
volgare arabica et italiana, copioso di voci e locutioni, con osservare la
frase delluna e dellaltra lingua (Roma. Nella stampa della Sac. Congr. De
Propag. Fide) followed three years later by his Fabrica linguae Arabicae
cum interpretatione latina et italica, accommodata ad usum linguae uul-
garis et scripturalis (Roma. Typis Sac. Congreg. De Prop. Fid.).
In 1800 a work has been completed by Franciscus de Dombay (1758
1810) with the title Grammatica linguae Mauro-Arabicae juxta vernaculi
idiomatis usum, accessit vocabularium Latino-Arabicum (Vindobonae:
apud Camesina)
28
but, according to Schnurrer , this title is misleading;
although this grammar describes the common speech of the people in
the Maghreb (Arabicus sermo in Mauritania quo vulgus uti solet), he
observes that all words are good Arabic (non sunt vulgari idiomati pro-
pria, sed omnia bene Arabica) (Schnurrer 1811, 95).
29
27
Germanicus was a teacher of Arabic at the mission school St. Peter in Montorio ,
Rome . He assisted with the preparation of the Arab Bible , he published dictionaries,
and commentaries on the Qur
altu as faaltum (2
person pl.m.), the use of - instead of -na for the second person feminine singular in
taf
al (= Haztu fem.) (72). It is remarkable that sometimes we fnd even hybrid forms ,
such as antum faaltu. However, we fnd also classical elements , such as the use of the
feminine plural in the verbal paradigms , which is not used in colloquial urban speech
today. It is also signifcant that the order of the persons singular in the verbal paradigms
is not the traditional one 73v72r. Caes has 3 (masc. Sing, 3 fem. Sing, 3. plur. 2 masc.
Sing. 2 fem. Sing. 2 plur. 1 sing 1 plur. Instead of 3,3,2,2,1 (sing.), 3, 3, 2, 2, 1. I am grate-
ful to Manfred Woidich for his comments on this footnote.
224 otto zwartjes
Grammars of Classical Greek were available throughout Europe , but
Spanish priests were engaged in the writing of vernacular Greek too, as
we can read in the prologue of Francisco Caes (1775, s.n.).
30
In the following paragraphs we will treat the use of Arabic terminol-
ogy in both grammars separately.
3.2 Semantic and infectional vowels and the concept of
amal
Te questions to be answered now are: which Arabic grammatical
terms are used, particularly in the sections dedicated to infection and
government, what did they mean and why did Caballero and Caes use
them?
Lets start with the vowel system , according to the paradigm of Cabal-
lero are:
TABLE 4A THE VOWEL SYSTEM ACCORDING
TO CABALLERO
31
bu
ba
bi
Damma
31
Fatha
Kesra
O que inclina a V
A que inclina a E
E que inclina a I
Unlike Pedro de Alcal , Caballero gives also in an appendix of this
Chapter the Arabic technical terms for these vowels when they are used
as infectional endings , accompanied with translations into Spanish :
Los Gramaticos a estas mociones dan otros nonbres conbiene a saber que
signifcan, eleuacion, ereccion, y contraccion: al-raf
, al-nasb, wa-l-jarr y
estas mociones duplicadas llaman tanun (tanwn ) esto es nunacion que es
30
It must be emphasized that missionary sources , ofen written in Spanish , are in
many cases the only existent sources which can give us information of vernaculars once
spoken in early ages of languages of which we only have more detailed information con-
cerning the literary or classical register. Priests understood very well that in China the
teaching of Mandarin was not so useful in regions were other dialects were spoken.
Grammars of modern Greek circulated since 1638, the Grammatica lingu grc
vulgaris was printed by Simon Portius . The first Spanish grammar of modern Greek
was composed by Pedro Fuentes , as we can read in the following quotation from the
same prologue: Por lo tocante la lengua griega ha impreso su gramatica vulgar Fr.
Pedro Fuentes observante, que residi en el Seminario de Nicosa en Chipre, y ahora est
imprimiendo la gramatica literal.
31
In their tables of the vowel system , both authors also give the names of the vowels
in Arabic script, not reproduced here.
inflection and government in arabic 225
lo mismo que decir addicion de un sonido de nun o n y biene a sonar on, an,
en vg. Racholon, Racholan, Racholen, Homo, Hominem, Hominis [. . .] pero
esto ueras mas claro en el tratado del nonbre (77 r.).
Te grammarians give other names to these motions [h arakt ], namely
the names which signify elevation, raising, and contraction, ar-raf
,
an-nas b , wa-l-jarr , and these motions, when duplicated, are called tanwn ,
which is nunation which is the same as adding the sound of a nn, or -n,
which approximately sounds as on, an, en vg. Racholon, Racholan, Racho-
len, Man (nom.), man, (acc.), man (gen.), but this you will see more clearly
in the chapter about the noun .
Caes table resembles that of Caballero , although there are some
diferences. Te order of the vowels is diferent, Spanish translations of
the Arabic vowel terms are given, and instead of de verb inclinar we
fnd declinar:
TABLE 4B THE INFLECTIONAL VOWELS ACCORDING TO CAES
(1775, f. 7)
ba
bi
bu
Fatha , Apercion
Kesra , Fraccion
Domma , Colleccion
A que declina e, y a clara. Ba, Be.
E que declina en e, i claro.
O que declina en o, y u. como Bo, Bu.
Also here, we fnd almost the same observation as above, which
demonstrates that the works are closely related to each other and that
they probably derived from a common source (or sources):
Los gramaticos las tres mociones dichas les dan otros nombres, es saber:
al Fatha le llaman Nasbo , esto es ereccion, al Kesra Charro, contraccion,
y al Domma Rafo, elevacion. Suelen tambien duplicar las dichas mociones
de esta suerte (
) (
)
(
), and they give them the name tanwn which is nunation , or vocal nn
which sounds rougly like: an, en, on.
In Chapter 4, both authors deal with the cinco signos:
(1) Secun (quietud), which has according to both the synonym chezm
(caballero ), or chiasmo (Caes ), (jazm ) which Caballero translates
as anputacion and Caes as corte;
226 otto zwartjes
(2) Texdid (duplicacion, corroboracion);
(3) Maddo (extension produccion);
(4) Vasalo (union, conjuncion); and
(5) Hamza (punzamiento).
A chapter dealing with how to read Arabic without vowel signs was
appended by Caballero, not present in the grammar of Caes. Although
there are diferences, both grammars are from the same tradition and
both authors (re-)formulated probably the lost version of Bernardino
Gonzlez , or quoted directly from other sources, such as Agapito Valle
Flemmarum , who almost has the same defnitions as Caess, although
the vocalizations of the Arabic terms and the order is slightly diferent:
TABLE 4C THE INFLECTIONAL VOWELS ACCORDING TO
AGAPITO VALLE (1687, f. 79V.)
32
bu
ba
bi
Dzhamma , collectio
Fathha , aperitio
Kesra , fractio
o declinans ad u, & u claru
a declinans ad e, & a claru
e declinans ad i, & i claru
Te Tratado III deals with the noun . Here we fnd sometimes some
parallels with the grammar of Pedro de Alcal , particularly since
the term seal has been used in both sources, or notifcacin for the
article
33
and the abitudines of Pedro de Alcal resemble much the
seales del nombre. Since colloquial Arabic nouns are not infected
according to cases, we fnd in the grammars of Alcal , Caballero and
Caes equivalents of the Greco-Latin cases for didactic reasons:
Los arabes aunque en la lengua vulgar reconocen tres numeros en el nombre,
es saber: singular, dual , y plural no conocen distinction de casos. Y asi el
nombre en qualquier caso termina con una misma voz. (Caes 1775, 59).
Te Arabs , although they recognize in the colloquial speech three num-
bers in the noun, being, singular , dual and plural , they do not know the
distinction in cases. And thats why the noun ends with the same sound
in whichever case.
32
The vocalization of Martelottus is slightly different: Dzhammon, Phathhon and
Kafron.
33
Pedro de Alcal uses seal de demostracin. See Zwartjes (1992).
inflection and government in arabic 227
Elsewhere in his grammar, i.e. afer the Numeros cardinales and
Numeros ordinales we fnd a comparable description:
Los Arabos en la lengua Vernacular o Vulgar no tienen terminacion de casos
en el nombre assi como los Castellanos (16v.).
In the vernacular or vulgar speech , the Arabs do not have the ending of
cases in the noun as the Castilians have.
Te case system according to Caballero is as follows:
TABLE 5 THE INFLECTIONAL VOWELS ACCORDING TO
CABALLERO (1709, 14r.)
Nominativo Vocatiuo raf
un
(this column in Arabic script)
Rajulun Vir, o vir
Acusativo Vocatiuo nas b
un
Rajulan Virum o Virum
Genitiuo
Datiuo
Ablatiuo
jarr
un
Rajulin
Viri
Viro
A viro
What did these terms mean in the Arabic gramma tical tradition ? Before
Sbawayhi , no distinctions have been made between the vowels which
are used in classical Arabic for the declensional endings , and the other
vowels, for instance: there was no disctinction between both vowels i
in the genitive al-kitb-i. An important novelty of Sbawayhi is that
he distinguished the frst i that is non-declensional , from the fnal i
which is declensional (Versteegh 1997, 19).
When we analyze Arabic terminology in our 18th century grammars ,
the sections about the particles are even more interesting. Caes and
Caballero use both a metalanguage inherited from the Arabic tradition
that had already been developed by Sbawayhi :
Tese endings follow eight courses: accusative (nas b ), genitive ( jarr ),
nominative (raf
), apocopate ( jazm ), a-vowel ( fath ), i-vowel (kasr ),
u-vowel (damm ), zero-vowel (waqf ). (Translation by Versteegh 1997, 36).
Te main distinction is whether a certain vowel is declensional or not.
Te Arabic terms used by Caballero have all to be related to the Arabic
term
rb ):
For the nouns these forms are: u nominative (raf
); a accusative (nas b ); i
genitive ( jarr );
For the verb only the imperfect verb shows mode infection: u indicative
(raf
); a subjunctive (nas b ); jussive ( jazm ).(Owens 1988, 39).
Carter demonstrated in several publications (1991, 1993, 1994)dealing
with diferent grammatical theories that the translation of technical
terms are in many cases inexact, problematic, anachronistic and can lead
to misconceptions.
34
Carter argued that the basic meaning of the concept
of
amal
und regere: Es ist nicht recht verstndlich, wie man hier schwanken
kann.
38
conjugates in three ways: by
varying the endings according to persons. Tis variation results from the
fact that the Arabs put before the verb in the mudri
,
or nominative , so that there is a permutation from the nominative to the
accusative.
Caballero uses the same classifcation and terminology, although his
hispanicised forms are slightly diferent: particulas charrantes , nas-
suantes , and chesmantes . Although in Caes description, only classical
Arabic is commented on, Caballero also dedicates a paragraph on
the particulas Chesmantes, ojo, Vulgares (particles which correspond
with the optative Utinam and other Latin forms, such as quare, quia,
quoniam, etc.).
If we compare the use of Arabic terminology in the works of Caba-
llero and Caes with those used by Erpenius , we can conclude that the
latter did not use Arabic terms so extensively. Worthy of mention are for
instance: Fatha , damma , kesra in the frst book De elementis . . .. We fnd
a Latinized form of the term jazm : post gjezman constanter manent
inflection and government in arabic 231
(1620, 20), also used as the verb Gjezmare, gjezmant, or in the passive
form gjezmatur (47), verbum hamzatum (70). Te term motiones is not
used as a translation of h arakt but it for the change a noun undergoes
if the feminine ending is added to the masculine form.
39
Nunnatio is
used as well (141), but the Arabic terminology for infectional endings
as used by the Spanish missionaries is not recorded.
3.3 Possible sources
Te use of non-Western metalanguage in itself has its own tradition. Not
only Pedro de Alcal used non-Western terminology , but in Northern
Europe we see also that Hebrew grammars used Hebrew terminology
in an adapted form.
40
Which sources could the Franciscans have used?
In the prologue of the grammar, Caes informs us that he completed
an eclectic grammar in agreement with the taste of everyone (que
sea del gusto de todos), using the most useful aspects (he procurado
aprovecharme de lo bueno que en ellas he visto ibid.) of earlier sources
and adding material from his own 16 years long experience:
He procurado con el mayor cuidado y desvelo leer, y releer para el ajuste de
esta, las gramaticas de Fr. Pedro de Alcal , de Fr. Felipe Guadaoli , de Toms
Erpenio, de Fr. Antonio de Aguila , de Fr. Agapito de Valle fammarum, de Fr.
Francisco Gonzalez , & c. Asimismo me he valido de un considerable numero
de manu-scritos, que me han franqueado gustosos algunos afcionados la
lengua arabe. Finalmente he aprovechado lo que me ense el estudio, y la
experiencia por espacio de diez y seis aos, que estube predicando, y confe-
sando en arabe en las misiones del Asia (ibid.).
I have taken the greatest care to do my best to read and read over again the
grammars of Fr. Pedro de Alcal , Fr. Philip Guadagnoli , Tomas Erpenius ,
Fr. Antonio de Aguila , Fr. Agapito de Valle Flammarum , and Fr. Fran-
cisco Gonzlez , and others. Likewise I have used a considerable amount
of manuscripts, which some afcionados of the Arabic language have
passed to me. Finally I have benefted from what study and experience
39
Motio est nominis Masculini in Foemininum converse; sitque additione termina-
tionis foemininae (Chapter V).
40
Johannes Reuchlin , who published his Hebrew grammar one year afer the publi-
cation of Alcal s, introduces the verb dagessare: quando he uel aleph repellunt nun pas-
siue signifcationis, dagessatur prima ut . . . (1974 [1506] Liber III, 590). See also Geiger
(1871, 129). Te presence of Oriental elements in Western grammars is ofcourse not
only present when loans are used. Translations and mistranslations produced some-
times terms which are not longer recognized as from oriental origin. An example is the
history and development of the concept of radix (root ).
232 otto zwartjes
have taught me over the period of sixteen years that I spent predicating
and confessing in Arabic in the missions of Asia .
To start with the frst grammar mentioned, Pedro de Alcal s Arte, we
can conclude immediately that this grammar has not been the source
for the terms used for the infectional endings . Tomas Erpenius has
been apparently used, but we observed that Erpenius reduced the exotic
grammatical terms in his grammars ,
41
although he maintained them in
his Latin translation of Arabic treatises written by Ibn jurrm and
al-Jurjn , which could have been also the direct sources of Caes . Ibn
jurrm was born in Morocco in 12731274 and died in Fez in 1323.
He is the author of a grammatical compendium entitled Muqaddima
al-jurrmya Mabdi
ilm al-
awmil al-mia
n-nah wiyya (Book of the Hundred Regentia ) of
as elevatio, the
defnition of an-nas b is accusativus, quasi patiens positum sub agente
(without translation), al-xaf
amal as
efcientia or operatio, and al-
awmil as regentia:
Ratione autem, quo vnu quodque in suo vel Casu vel Modo & quacumque
dispositione locatur, quam Constructionem latini dicunt, Arabes interdum
al-i
amal
un
efcientiam, seu opera-
tionem appellant. Nomen autem illud, seu Verbum, seu Particula, eiusmodi
operationem exercens, scilicet cuius vi aliquid in tali vel tali dispositione
locatur, dicitur,
mil
un
operans, regens. Locatum ver ex eius dicitur
ma
ml
un
operatum, rectum. Exinde dicuntur, al-
awmil
u
regentia. Qui-
bus omnibus notis, nihil superest Grammatico. Collegit autem quidam,
inflection and government in arabic 235
cognomine Giargianius , in libello Regentia omnia, quem propterea nomina-
vit De Centum Regentibus. Regentia enim, vel sunt lafz iyya
tun
explicita, vel
ma
nawiyya
tun
implicita.
48
(Guadagnoli 1642, f. 248).
However, the system according to which a constituent has to be placed
in the appropriate case or mode and on which convenient position has
to be placed, this system which the speakers of Latin call Constructio
[= syntax ], is system is called al-
amal
un
efciens [an act
which produces a certain efect], or operation [an act caused by force of
an operator]. When a Noun , a Verb or a Particle produces such efects in
this manner, and when by force of this efect something has to be placed in
a certain disposition, we call this
mil
un
[producing a certain efect upon
something, to govern]. Te element which has been efected is called
ma
ml
un
[the governed]. Consequently, al-
awmil
u
are called regentia
[governers]. Although all this is well-known, nothing has been transmit-
ted by Te Grammarian,
49
but someone with the name Giargianius has
collected in a booklet all the regentia, which can be subdivided in two
subclasses, the lafz iyya (expressed) and the ma
nwiyya (abstract).
As we see, Guadagnoli s description is not only accurate and complete,
since the original Arabic is given and the literarly meaning had been
maintained, but he also mentions his source, which is the hundred
regentia written by al-Jurjn (Giargianius ).
50
Do we fnd traces of these translations of
rf )
(Guadagnoli, 1642, f. 255).
51
Scripsit autem eleganter admodum de dictionibus hisce, quemadmodum & de lit-
teris eruditissimus Orientalium linguarum in Leidensi Academia professor Thomas Erpe-
nius uniusquam leuasset quoq. nobis huiis secundi libri labores partem (1620, 38).
236 otto zwartjes
its technical terms be maintained, or should they be abandoned and
replaced by Latin terminology ? Martelottus decided to follow the tra-
ditional method when dealing with the single word classes (Priora
vero hic loci propri de singulis dictionibus singillatim ordine debito,
ac recta methodo explicanda sunt (1620, 37), but we can infer from his
preface, he obviously follows the Arabic model, and he does this explic-
itly.
52
However, Martelottus does not leave the Arabic terminology out
of his volume:
Ubi omisso Arabicorum grammaticorum ordine Latinis admodum dissono,
in rebus pluribus nobiscum, quoad methodum conuenisse comperimus.
Caeterum quamvis praedicta Arabum methodo defectentes, nostro nos
ordine Latinis magis consono procedamus, omnino grammaticalibus voci-
bus ubique utemur, eorumque ordines, ac procedendi rationes, divissiones,
ac series enucleabimus, ut faciliro cunctis ad eosdem perlegendos authores
pateat aditus. (Martelottus 1620, 38).
Whenever the diference between Arabic grammars and the Latin way of
construction was lef out of consideration, we discovered that, with res-
pect to the system of the language, Arab had many similarities with ours.
Afer all, to what extent we ourselves may difer from this language system
of the Arabs and are difering more and more as well from the Latin way
of construction, we should still use in our research the Arab terminology
and developing our knowledge we will explain the systems, the methods
of construction and concatenation, in order to make it easier for us all to
read the same authors.
Martelottus s methodology is in our eyes extremely modern. It tries to
bridge the gap between exo- and endo-grammaticalization . An eclectic
approach, combining the best elements of both traditions and their
corresponding technical terms is the best way to understand the Arabic
language. Martelottus also dedicates an entire chapter to infection (De
mila as:
1. particlae operantes in nomen simplex Giarrantes,
2. et in nomen simplex nasbantes,
3. et in summam,
4. et in verbum nasbantes,
5. et operantes in verbum Giezmantes (1687, 268).
Another Franciscan , Antonio ab Aquila , also used Arabic terminology
for the particles , although we do not fnd the same defnitions. In his
classifcation of the subclasses, we fnd a Latin terminology , such as
particulae coniunctae and particulae separatae, with subclasses, such
as particula interrogationis, iurandi, status in loco, ad determinationem,
optativi, conformandi, negationis, conditionalis (1650, Caput XXV,
388 f.). However, he also uses Arabic terms , such as particula Gizemeh
(f. 403).
And fnally, the question rises wether we fnd traces of the concept
of
(u ending, or subject ) or nominative , as in certainly to write to
Peter innam sayaktubu Bat rus.
To sum up, we have found the following authors as possible sources of
the two Franciscan grammarians of Damascene Arabic : Ibn jurrm
and al-Jurjn (through translations of the Medici edition, Erpenius ,
Obicini and maybe also Kirsten ) as the main Eastern sources. Western
sources mentioned by name are Erpenius, Golius (mentioned by Caes ),
Agapito Valle and Guadagnoli . It has to be observed that missionary
grammarians in Rome were familiar with some of the most important
Arabic authors. Te name of Al-Xall b. Ah mad al Farhid (died in 780)
who codifed and established a system of 15 meters has been mentioned
by Guadagnoli. Since the grammars of Caballero and Caes do not have
a fnal chapter on prosody, they did not deel the need to use this source.
Another prominent pioneer, who established the foundations of Arabic
grammar is of course Sbawayhi , whose work became known in the
West through translations of Jahn . It is surprising that we fnd already in
1620 his name in the grammar of Martelottus , which is evidence for the
fact that missionary-grammarians in Rome knew who he was. Dealing
with the verba ternaria, Martelottus explains that the forma masdari is
irregular, and in that context he mentions Sbawayhi:
In verbis ternarijs, ut iam diximus, forma masdari irregularis est, omnes
autem quidem Author Arabicus, nomine Sibauai ad 32. reuocauit, videlicet
(1620, 213).
In the ternary verbs , as we said before, the mas dar-form is irregular;
indeed, the Arabic author with the name Sbawayhi reduced all these to
32.
Te same author, Martelottus , also mentions another Arabic source in
his chapter dealing with the Constructio Particularum where he divides
the particulas operantes in fve subclasses, in agreement with an Arabic
work called Lucerna, or Al-Mis bh , the particulae operantes in nomen
simplex giarrtes, particulae operantes in nomen simplex nesbantes,
in verbum nasbantes, in verbum gezmantes, and the original text in
inflection and government in arabic 239
Arabic script is given in the same table as well. Although Martelottus
does not give the name of the author, we think this work is probably
the treatise written by al-Mut arriz , entitled Al-Mis bh f
ilm an-nah w.
Al-Mut arriz (11441213) compiled this treatise which became a text-
book in the madrasas of the East. Te Mis bh itself wa sbased on three
small grammatical monographs of al-Jurjn (Lichtenstdter 1936, 847
and Sellheim (EI [New edition] 7, 773).
4. Conclusion
Summarizing, we can conclude that Pedro de Alcal s use of Arabic
terminology seems to be unsystematic and the reasons why he used
them remain unclear, particularly when he uses the Arabic names for
the cases. Tey do not refect the Arabic infectional endings and his
model was obviously Latin grammar . Te use of the term damr is an
exception, since the sufxed pronouns can be used in a diferent way-
attached to verbs , nouns and particles compared to Latin . Pedro de
Alcal s mnemonic terms of the vowels stand alone, and we do not fnd
any use of them in other works and in his dictionary we do not fnd the
terms fath a , kasra , and damma . Although we fnd in Alcals dictionary
the translation obrar for the Arabic root
mil has followed its own course in grammatical theory and in the
20th century it was absorbed by anachronistic terms such as govern-
ment and dependency, as Carter demonstrated. Nevertheless, we have
240 otto zwartjes
demonstrated that the Franciscans were totally aware of the right con-
notations of
iqra
n . He stated that there are two basic features of waqf . Tese are
wajh manner and mah all place. Te purpose of mah alli l-waqf place of
pause is to regulate the speech event and leads to improving its division
where juncture or pause may occur. Te wajh , plural awjuh, provides
the speaker or the reader with the guide for waqf so the speech will
appear harmonized in its utterances. He added that waqf has a waz fa
task which serves for clarity of pronunciation and meaning that help in
understanding of what is being read or spoken (Farrj 2004, 1223).
1.4 Tamm H assn (1973)
H assn stated that waqf through its various means by its nature is a
mifs al , separator of speech where it is possible that the chain of speech
may broken into spoken groups. Every one of these, when its meaning
is complete, is considered a speech event . However, if the speech is not
250 salman h. alani
complete then the speech event may consist of more than one event
(H assn 1973, 270).
2. Te Arabic linguistic analysis of and rules for waqf
2.1 Te waqf status of the t marbt ah
One of the most common modes of waqf is applied to words that end with
a t marbt ah . Tis t marbt ah is considered to be a morpheme that
primarily marks the feminine endings of nouns and adjectives . However
there are some feminine nouns and adjectives that do not always end
in t marbt ah . In Arabic script some feminine nouns and adjectives
may be written with t mabst ah . Tis is the regular t that appears at
the end of words. Tis is especially true in the script of the Qur
n . In
fact we fnd sometimes the same word written with t marbt ah and in
other contexts it is written with a t mabst ah . Te word rah ma mercy
for example is written in either t marbt ah or t mabst ah .
Tere are nouns and adjectives that end in a t marbt ah that have a
masculine meaning. Proper names such as H amzah, Talh ah, and several
others are written with t marbt ah . Te rules of waqf apply to them as
they apply to the feminine nouns and adjectives . Terefore it is not the
gender of the word but rather the form of the word that determines the
application of the waqf rules.
Te t marbt ah in nouns and adjectives that appears in prepause
form is deleted and replaced by /-h/. Tis deletion, of course, includes
the vowels : /-i/, -/u/ and /-a/ in the defnite and /-in/, /-un/ and /-an/
in the indefnite . Te deletion takes place in all three cases: nominative ,
accusative and genitive . With prejunctural , on the other hand, the t
marbt ah and the vowels both in the defnite and indefnite are retained.
Tis is illustrated with the word madrasah school as in the prejunctural
forms below:
Nominative Genitive Accusative
Indefnite madrasa-tun madrasa-tin madrasa-tan
Defnite al-madrasa-tu al-madrasa-ti al-madrasa-ta
linguistic analysis and rules of pause in arabic 251
Te case endings in the indefnite /-tun, -tin, -tan/ and the defnite /-tu,
-ti, -ta/ all will be deleted in the prepausal forms . Te word madrasah
school and the word al-madrasah the school both in defnite and
indefnite forms will have the same endings in the prepausal forms . Te
reason is that the /-h/, placed between slashes, is sometimes weakened to
the point that one really cannot even hear it. Tis is especially the case of
Modern Standard Arabic , read aloud or spoken by radio and television
broadcasters. I have examined and analyzed the speech segments of
several announcers and observed both the dropping of of the /-h/ and
the retaining of it. In careful delivered speeches especially of religious
nature the /-h/ is almost always retained.
2.2 Te deletion of fnal short vowels
Te fnal short vowels , in waqf , are deleted. Tis deals with al-h arakti
l-
irbiyyah, the vowel marks that signal case endings , tense and mood .
Te following rules apply to nouns and adjectives derived from strong
verb roots :
(a) Te case endings in the defnite nouns are indicated by /-u/ in the
nominative case , /-i/ in the genitive case, and /-a/ in the accusative
case. All of these short vowels that mark the case endings are omit-
ted in waqf . Te following sentences illustrate the prejunctural and
prepausal forms :
Prejunctural forms Prepausal forms Meaning
ja-l-walad+u ja-l-walad the boy came
mina-l-walad+i mina-l-walad from the boy
raaytu-l-walad+a raaytu-l-walad I saw the boy
(b) When the indefnite nouns and adjectives that end in nunation,
the sound /-n/ pronounced but not written. Te case endings in the
indefnite nouns and adjectives are indicated by /-un/ in the nom-
inative case , /-in/ in the genitive case and /-an/ in the accusative
case . Both nominative and genitive endings are omitted in prep-
ausal forms . However the accusative case maker is changed into an
alif /-/ that requires a special treatment. Te word walad boy in
the following sentences illustrates both prejunctural and prepausal
forms .
252 salman h. alani
Prejunctural forms Prepausal forms Meaning
ja walad+un ja walad a boy came
min walad+in min walad from a boy
raaytu walad+an raaytu walad+aa I saw a boy
(c) Te indefnite accusative case ending is /-an/. Also certain types of
adverbs end in /-an/. Tis ending is replaced by /-/ in prepausal
forms as illustrated in the table above. In the case of adverbs some
speakers retain the nunation while others replaced it with /-/.
However, in prepausal and words in isolation may be retained but
ofentimes it changes to /-/ following the normal rules of waqf . Te
word id an meaning if is written orthographically either with nn
/n/ or with the alif tanwn. In prepausal form pronounced as /-/.
(d) Tense and mood in verb-vowel endings are omitted.
Prejunctural forms Prepausal forms Meaning
katab+a katab he wrote
yaktub+u yaktub he writes
lan yaktub+a lan yktub he will not write
2.3 Words that end in long vowels
Tere are two main grammatical categories of words that end in long
vowels . Tese are called in traditional Arabic grammar as the al-manqs
defective as in the word al-wd the valley and al-maqs r shortened
as in the word al-fat the youth.
1. Te rules of waqf on words that end in the long vowel /-/ such as
al-muh m the lawyer which is derived from a fnally weak verb h am
and its imperfect yah m to defend ends in /y/. Nouns that are derived
from fnally weak verbs like h am of the pattern of al-muh m always end
/-/. Tis category of words should not be confused with words ending in
y-n-nisbah as in words like lubnniyyun Lebanese or even with words
like z abyun deer both of these words are written in Arabic script with
/y/, however the source of this /y/ is not a fnal radical as the nouns
linguistic analysis and rules of pause in arabic 253
derived from fnally weak verbs like h am and its imperfect yah m to
defend which ends in /y/ as a radical (Farrj, 2001, 7172).
Te vowel endings of al-manqs defective are determined by the case
endings and whether they are defnite or indefnite . When words of the
manqs defective are in the accusative case and indefnite they end in
/-/ in the waqf as qbaltu muh my I met a lawyer afer the omitting
of the nunation. When they are defnite they end in /-/ as in qbaltu
l-muh m I met the lawyer. When the manqs words are in genitive or
nominative case they end in the omitting of the fnal vowel /-/ as in fl-
wdin the valley and hd wd this is a valley. However when they are
defnite the words of the manqs end in long vowel /-/ as in fl-wd in
the valley and hd l- wd this is the valley.
2. Te rules of waqf of words that end in the long vowel /-/ such as
al-fat the youth and which are primarily derived from fnally weak
verbs are called in traditional Arabic grammar al-maqs r the shortened.
Te waqf on these words is always end in an alif /-/ in all three case
endings and regardless how orthographically they are written with alif
maqs rah or regular alif. What matters here is the pronunciation and
not the script.
3. Concluding Remark
It is worthwhile to mention that the rules of waqf are not always adhered
to by readers and speakers of Modern Standard Arabic . Te Arabic
language for sometime has been going through processes of change.
Some readers and speakers are not using the al-h arakta-l-irbiyyah
case endings. Te ofen heard statement that states sakkin taslam use
sukn and you will be safe refects the state of afairs of the on going
change of the Arabic language. Terefore the rules of waqf outlined
above when considered should be used as guidelines not to be applied
in a rigorous and strict manner.
254 salman h. alani
4. References
4.1 Primary source
Ibn Ya . n.d. arh al-mufas s al. Cairo: Maktabat al-Mutanabb.
4.2 Secondary sources
Farrj , Muh ammad. 2001. Al-waqf wa-waz
. Annals
of Arts and Social sciences, Monograph 159 Volume 21. Kuwait: Kuwait University.
H assn , Tammm. 1973. Al-lua al-arabiyya mabnh wa-manh. Cairo: al-Hay
a
al-mma lil-Kitb.
Holes , Clive. 2004. Modern Arabic Structures, Functions and Varieties. Revised edition.
Georgetown, Washington, D.C.: Georgetown University press.
Mitchell , T.F. 1990. Pronouncing Arabic. Oxford: Clarendon Press.
THE EXPLANATION OF HOMONYMY
1
IN THE LEXICON
OF ARABIC
Georges Bohas and Abderrahim Saguer
ENS Lettres et Sciences Humaines, Lyon
1. Introduction: the theoretical framework and earlier analyses
Tis article is a follow-up to our previous paper entitled Sur un point de
vue heuristique concernant lhomonymie dans le lexique de larabe.
2
By
adopting a heuristic viewpoint, we take into account the fact that we have
not yet fnished exploring all the matrices of Arabic , and thus proceed
by means of successive evaluations and provisional hypotheses.
3
Certain
points remain to be clarifed; nevertheless, the level of explanation we
can achieve has become clearer, as well as the explanatory methods that
we are able to ofer within the framework of the Teory of Matrices and
Etymons (TME ). Although there are already numerous publications on
this subject, it is worth underlining that, within the TME framework,
the lexicon is organized on three levels:
4
1 / The matrix : a non linearly-ordered combination of a pair of pho-
netic feature vectors linked to a notional invariant ; for example,
{[labial ], [coronal ]} to strike a blow.
5
1
It is worth defning homonymy in opposition to polysemy . Polysemy is a word
which brings together several meanings between which users can recognize a link
(Nyckees , 1998: 194); the meanings are diferent but related. Homonymy is distinct
from polysemy in that, in the case of homonymy, it seems impossible to re-establish a
plausible semantic relationship (Nyckees , 1998: 194) between the diferent meanings,
for example: fies certain insects and fies the opening at the front of a pair of trousers
or to sound to make a noise and to sound to measure the depth of waterdiferent non
related meanings.
2
Bohas and Saguer (2006).
3
Tat is: in a not-rigorously demonstrated manner but justifed by reasons of inter-
nal coherence (see the website www.memo.fr Einstein, Albert); and accepting that you
cannot explain everything.
4
See Bohas (1997, 2000), Dat (2002).
5
Tis is a property of the language that was proved both formally and semantically
by Bohas and Darfouf (1993), developed in Bohas (1997), which consists in the fact that
256 georges bohas and abderrahim saguer
2 / The etymon : a non linearly-ordered bi-consonantal base made up
of two phonemes taken from a given matrix exhibiting both the fea-
tures of this matrix and its notional invariant ; for example, {b,t} to
strike a blow with a sharp object.
3 / The radical : an etymon that has developed by diffusion of the final
consonant or by incrementation , or that results from the blending of
two etymons ; the radical includes at least one vowel and vectors the
notional invariant ; for example: /bvtar/, to cut, to cut the tail (Bohas
2000, 9). The radical is the domain in which diverse morphological
and Ablaut processes take place (Guerssel and Lowenstamm 1993,
Segeral 1995).
So far, ten matrices have been accounted for; most of these have already
been subjected to in-depth studies (see footnote
14
).
Matrix 1 {[labial ],
6
[coronal ]}
Notional invariant : to strike a blow
Matrix 2 {[labial ], [-voiced ]
}
7
[+continuant ]
8
Notional invariant : movement of air
Matrix 3 {[labial ], [pharyngeal ]}
9
Notional invariant : (a) tightening
Matrix 4 {[coronal ], [pharyngeal ]}
[-dorsal ]
10
[-voiced ]
a binary combination {a, b} is realized in the order a+b and in the order b+a while keep-
ing the same notional invariant .
6
[labial ] characterizes sounds produced with a constriction of the lips. For matrices
1, 2, 3, 6 we integrate on-going research which shows that the feature [labial] should not
be restricted by [-sonorant ] (see Mansouri , 2006).
7
[voiced] Sounds produced with vibration of the vocal cords are said to be voiced
([+voiced ]), whereas other sounds are said to be non-voiced ([-voiced ]), see Dell (1973:
56).
8
[continuant] Sounds with the feature [+continuant ] are produced without inter-
rupting the fow of air through the oral cavity, those with the feature [-continuant ] are
produced with total interruption of the fow of air at the oral cavity, see Halle (1991:
208).
9
[pharyngeal ] characterizes segments that the Arabic tradition calls gutturals , that
is:
, h, , h , x, and q. For the problems posed by the characterization of this class, see
Kenstowicz (1994: 456f).
10
[dorsal ] characterizes sounds produced with a constriction created with the back
of the tongue between the sof palate and the uvula (velar and uvular consonants ; rear
vowels ).
explanation of homonymy in the lexicon of arabic 257
Notional invariant : low voice, mufed, hoarse noise
Matrix 5 {[coronal ], [dorsal ]}
Notional invariant : to strike a blow
11
Matrix 6 {[labial ], [dorsal ]}
Notional invariant : curvature
Matrix 7 {[dorsal ], [pharyngeal ]}
Notional invariant : the cries of animals
Matrix 8 {[+sonorant
12
], [+continuant]}
[+lateral
13
]
Notional invariant : the tongue
Matrix 9 {[+nasal ], [+continuant ]}
Notional invariant : the nose
Matrix 10 {[+nasal ], [coronal ]}
Notional invariant : traction
14
Te data on which we have based our study are to be found in the
Kazimirski , and have been checked in the Qms and/or the Lisn .
When they are based on another source, this is mentioned.
2. Explanatory methods
In the paper quoted above, we demonstrated that the homonymy of a
radical may be attributed to three causes:
A. the fact that it is the result of blending : it manifests the meanings of
both the etymons that are its source.
11
See Diab (2005) who brings a modifcation to the formulation of the notional
invariant of this matrix .
12
[sonorant] Sounds with the feature [+sonorant ] are produced with a constriction
which does not infuence the capacity of the vocal cords to vibrate spontaneously. Tose
with the feature [-sonorant ] have a constriction which reduces the global fow of air and
makes voicing more difcult. Tus the natural state for sonorants is [+voiced ] and for
non sonorants (termed obstruents) is [-voiced ], see Kenstowicz (1994: 36).
13
[lateral] A [+lateral ] sound is produced by making a constriction with the central
part of the tongue while lowering one or both lateral edges so that air escapes around the
side(s) of the mouth, see Kenstowicz (1994: 35).
14
For matrices 1 to 6, see Bohas (2000), Dat (2002), for an in-depth study of matrix 6,
see Serhane (2003), Bohas and Serhane (2003), for matrix 7, see Bohas and Dat (2005),
for matrices 8 and 9, see Bohas (to be published) and for matrix 10, see Saguer (2003).
258 georges bohas and abderrahim saguer
B. the fact that its etymon is the realization of several matrices : it mani-
fests the meanings of these matrices.
C. the fact that two etymonial analyses are possible, such as [nX]Y and
n[XY].
Below, we illustrate each case with an example from Bohas and Saguer
(2006).
A. Homonymy resulting from blending
Let us consider the verb araza, which attests two meanings (hereafer
senses ):
S1 to prick something with a needle, to drive in, to plunge (a sharp
instrument), to plunge a tail into the ground to lay eggs (of
locusts);
S2 to give but very little milk (of pregnant camel).
Te same semantic load is found in rizun;
S1 that drives in, plunges a sharp instrument, a goad into something;
that plunges a tail into the ground to lay eggs (of locusts);
S2 that gives but little milk (camel).
Since it is not possible to establish a plausible semantic relationship
between the two senses , this is an obvious case of homonymy . And yet
we observe the existence of the following words:
arra
15
F. III : to be found in small quantities (of milk of a female);
irrun : a small quantity, generally, such as a small quantity of
milk in a female
murrun : that has little milk in the udders (camel)
Phonetically , the etymonial analysis can only be { ,r}, since their radical
has no other consonant,
16
and it is obvious that they attest sense S2.
Moreover, the precise meaning of the verb razza is: to plunge a tail into
the ground to lay eggs (of locusts); to stick, drive in and fx frmly one
object into another or into the ground. Tus it clearly attests sense S1
and is analysed as the etymon {r,z}.
17
Terefore, the explanation is that
araza includes senses S1 and S2 because it results from blending of the
15
We use boldface for the segments that make up the etymon .
16
We call these non ambiguous radicals .
17
Non ambiguous radical .
explanation of homonymy in the lexicon of arabic 259
two etymons {,r} and {r,z}. Te way this blending occurs is represented
in model A:
18
A
C
j
C
i
C
i
C
k
S
i
S
j
C
j
C
i
C
k
19
S
i
+ S
J
More explicitly :
r rz
lack of milk x
20
to drive a sharp object into
S
i
S
j
araz
1920
Henceforth, in similar cases, we shall talk of explanation through
blending .
B. Homonymy through the realization of several matrices
1) An easy case
Let us take the verb mata
"
a, which means:
S1 to strike somebody with a stick
S2 to tighten, stretch out a rope
Te m is [labial ], the t is [coronal ]: the etymon {m,t} can thus be a real-
ization of matrix 1:
{[labial ], [coronal ]}
Notional invariant : to strike a blow
And for this reason assumes the sense S1 to strike somebody with a
stick.
But the m is also [nasal ] and the etymon {m,t} can also be a realization
of matrix 10:
18
See Bohas (1997: 175f and 2000: 49).
19
Te obligatory contour principle (OCP) explains the fusion of the two Ci into a
single segment. See McCarthy (1986) for the defnition and various applied examples
referring to Semitic languages of this principle which forbids adjacent identical elements
at the same level. Since, OCP has given rise to a multitude of studies the list of which
would be superfuous here.
20
We use x to indicate blending .
260 georges bohas and abderrahim saguer
{[+nasal ], [coronal ]}
Notional invariant : traction
And for this reason assumes the sense S2 to tighten, stretch out a
rope.
2) A more complex case
Te verb nat ara attests the following meanings:
S1 to disperse
F. I : to scatter, disperse, disseminate
F. II : to scatter a lot, in large quantities: intensive of F. I
F. V : to be scattered, dispersed, to disperse
F. VI : to be scattered, dispersed, to disseminate, to spread
S2 actions concerning the nose
F. I nat ura : to blow ones nose
F. II : to blow ones nose
F. II : to draw up water through the nostrils
F. VIII : to blow ones nose
F. VIII : to draw up water etc. through the nostrils and to
expel it through the nostrils
S3 to pull, tear of/out
F. I : to remove, to take the clothes of the body of some-
body, to strip
S4 to strike a blow with a sharp object
21
F. IV : to pierce somebody with a sharp instrument and to
make the blood fow
Te frst hypothesis we can formulate is that nat ara develops the ety-
mon : {n,t} and that for this reason, it is a realization of matrix 9:
22
{[+nasal ] [+continuant ]}
Notional invariant : the nose
Te phonetic substance of this matrix comprises, on the one hand, the
two nasals, m and n, and, on the other, the various fricatives .
21
We will propose no explanation for this sense ; as we said in the introduction, we
have not yet explored all the matrices of Arabic , the notional invariant sharp is without
doubt important but as yet we know nothing of it.
22
See Bohas (to be published) for a detailed study of this matrix .
explanation of homonymy in the lexicon of arabic 261
Te ramifcations of the notional invariant are as follows:
1. the nose
1.1. the organ itself and what afects it
1.2. the specifcation of parts (the top, the sides)
1.3. to be sharp, > protruding, > to precede
2.1. specifcations of the organ (big, small . . .)
2.2. animal or human presenting these specifcations
3. to lif the nose : movement of pride or contempt
4. the nose and air : to breathe in; to breathe out; to perceive odours,
to smell
5. the infuence of the nose on the voice: nasal sound; similar animal
cries (buzzing-grunting)
6. various liquids (mucus, phlegm) which pass through the nose
We may note that, in this organization, senses S2 to blow ones nose,
to draw up water through the nostrils, ft into both 6. and 4.
But, since n is [+nasal ] and t [coronal ], {n,t} can also be a realization of
matrix 10:
{[+nasal ], [+coronal ]}
Notional invariant: traction
Moreover, other realizations of the {n,t} etymon can be found in:
natala [nt]l : to remove the garment or breastplate
nanaa [n]n : to remove (ones clothes)
Sense S3 to pull, tear of/out of F. I: to remove, take the clothes of the
body of somebody, to strip is thus incorporated into this matrix .
We can therefore understand why the {n,t} etymon is homonymic : if we
take into account the t the feature [coronal ], then we see that it realizes
matrix 10 traction, while if we take into account the feature [+continu-
ant ], then we see it realizes matrix 9 the nose.
In this case we shall speak of ambiguity originating from the fact that
an etymon is the realization of several matrices .
C. Homonymy due to the possibility of two etymonial analyses
Let us pursue the analysis of nat ara. We have seen that it also attests the
sense to disperse, to scatter. It is true that we can establish both a pho-
netic and a semantic relation with:
tarra [tr]r : to disperse, disseminate
the etymon of which can only be {t,r},
23
and with:
23
No ambiguity.
262 georges bohas and abderrahim saguer
tra t[w]r : to be lifed and scatter in the air
tart aratun [tr]t r : dispersion, dissemination.
farata f[rt] : to be dispersed, disseminated (of a tribe)
Tis leads us to analyse nat ara by positing {t,r} as etymon and the n as
an initial crement . Terefore, the homonymy stems from the fact that in
A and B the form is analysed as [nt]r, which may have two matrix links,
and in C as n[tr]. We shall say in the latter case that homonymy arises
because several etymonial analyses are possible.
3. Te two levels of explanation
In Bohas and Saguer (2006) we studied cases concerning radicals
containing an n. In this paper, we shall extend the study to include the
analysis of radicals containing an l. According to previous studies,
24
l may
have the status of initial crement /prefx
25
or of matrix segment . Earlier,
Hurwitz (1913, 5560) had already recognized this prefx status:
Te preformatives are thus seen to possess a fairly defnite, though remote,
relationship to each other. Te sibilants and gutturals are to be traced to
causative stems ; the dental t and liquid n to refexive stems ; the liquids m,
l, and i are to be connected etymologically with the refexive n, and the
preformative y may be considered to be a denominative stem .
When it has a matrix function, l may form an etymon with the second
radical : [lx]y, or the third radical : l[x]y.
26
In other words, a radical [lvxy]
may present all its possibilities, as can be seen in the table below:
Etymon : {x,y} {l,x} {l,y}
Analysis: l[xy] [lx]y l[x]y
Radical : <lvxy>
24
Te most recent are those of Saguer (2000, 2002, 2002b).
25
We use prefx when l has a semantic -grammatical value and initial crement when
it does not.
26
Tis expresses the fact that in this form the etymon is [ly] and x is an inset
crement .
explanation of homonymy in the lexicon of arabic 263
3.1 Te explanation through attribution to identifed matrices
Te verb labaxa demonstrates eight senses :
S1 to be feshy (of the body)
S2 to beat, strike somebody
S3 to kill somebody
S4 to take a thing from somebody, from the hand of somebody
S5 to get something out of somebody using trickery
S6 F. III: to slap somebody in the face
S7 F. V: to perfume oneself with musk
S8 to say something foolish, insulting to somebody
It should be noted that several senses are polysemically related; in other
words, a plausible semantic relation can be established between them,
as follows:
S2, S6 and S3 are manifestations of the notional invariant to
strike a blow ; to kill is related to this concept through the
cause>consequence relationship and S6 specifes the mode of
action .
S4 and S5 come under the notional invariant : to pull , bring some-
thing to oneself.
Tus fve homonymic senses remain:
A S1 to be feshy (of the body)
B S2+S6+S3 to strike a blow
S2 to beat, strike somebody
S6 F. III: to slap somebody in the face
S3 to kill somebody
C S4+S5 to pull, bring something to oneself
S4 to take a thing from somebody, from the hand of some-
body
S5 to get something out of somebody using trickery
D S7 F.V: to perfume oneself with musk
E S8 to say something foolish, insulting to somebody
We are going to attribute each of these homonymic senses to a source
matrix .
Take sense A S1labaxa: to be feshy (of the body).
264 georges bohas and abderrahim saguer
It is easily noted that this form is semantically linked to:
rabxun r[bx] : fat, thick and with a sof and loose body
habayyaxun h[bx] : young chubby
at a l[
a [sb]
a : to peel, to skin
wafala : to peel something by removing the bark
We introduce an etymon {l,f }; l is [coronal ] and f [labial ]. Tis etymon
can thus be a realization of Matrix 1:
explanation of homonymy in the lexicon of arabic 285
{[labial ], [coronal ]}
Notional invariant : to strike a blow
Loss, harm comes under the heading B.3,
48
global consequence, as in:
h afata : to destroy, to lose
talifa : to perish
As for S4 F. IV: to pull out (e.g. somebodys nail) it is a realization of
a Matrix under study:
{[+approximant ], [+continuant ]}
[coronal ]
with the notional invariant : to bring something to oneself.
Tis manifests itself in the words quoted earlier in section 3.1. As f and
h are both continuants , it seems reasonable to consider the radical as a
blending of the two etymons that both realize this matrix : lh x lf, B type
blending .
49
Tere remains sense S5 F. IV: to burn, to have something consumed in
fre. In this meaning, we can relate lahafa to:
lafaha : to burn, to cause harm through its intensity (of fre, or
a very warm wind)
fayh : heat caused by a star
sahafa : to burn, to have something consumed in fre
Te above relation reveals the etymon { f,h}; the semantic relation with
to blow remains to be established in order to link it to Matrix 2.
As for the residual cases
F. II : to let the bottom of ones garment scrape the ground,
to wear it very long so that it trails, by extension: to
walk proudly
F. IV : to come to the foot of a mountain and lih f: the foot of
a mountain
At this stage in our research, all that can be noted is that the latter is
perhaps to be related to h fatun: edge, margin, extremity, although we
cannot establish this with certainty. All these comparisons are of limited
interest, but it is worth remembering, as we said at the start, that certain
points are still unclear.
48
See Bohas (2000: 79).
49
See Bohas (2000: 50).
286 georges bohas and abderrahim saguer
Consequently, we shall limit ourselves to those senses that are clear, in
order to construct the lexicogenetic tree :
M8 M1 M3 M7
{[+lat ], [+cont ]} {[labial ], [coronal ]} {[phar ], [lab ] } {[+approx ], [+cont ]}
tongue strike a blow tightening to bring to oneself
{l,h} {l,f } {h,f } {l,h} x {l,f }
[lh ] f l[h ]f l[h f ] [l,h ] x [l,f ]
lah af
5. Conclusion
Our method is thus distinct from that of the partisans of the
triconsonantal root . For them, it is sucient to identify the three
consonants in order to consider the analysis complete, even if
semantic incongruities and incompatibilities are evident, and even
if this identication provides no explanation of phono-semantic
links between words, such as homonymy and enantiosemy . What
does reassure them, however, is that they can pride themselves
on having reached a state of certainty.... that the root of maktab =
ktb is a certainty! Yet why should the root of istadaytu be dw rather
than dy, and, if it is dw, what phonetic motivation is there to be
found in istaf altu to justify the passage from w to y in istadaytu?
Indeed, this certainty is not as denitive as they would have us believe.
With our approach based on argumentative reasoning, we might make
mistakes: one word might, perhaps, be matched up with another on
the basis of such and such a property without us having noticed this
relation. Given the explanatory results of our approach, which adopts
a heuristic point of view, this is a risk we assume.
Te reader will have noticed that all our research has been carried
out taking the lexicon of Arabic as a synchronic whole. As we have ofen
repeated, in the study of the lexicon, it is vain to go back to a previous
explanation of homonymy in the lexicon of arabic 287
biliteral stage in order to diachronically derive a triliteral stage. In other
words, the old debatebiliteral or triliteralis immaterial. Te binary
composites (the matrices and the etymons ) and the ternary composites
are there before our eyes: you only have to open a dictionary to fnd
them. Every radical is, according to the level of explanation, binary or
ternary, as the explanation of homonymy demonstrates all too well.
Admittedly, no one is obliged to overstep the conception, invented
by the Arab grammarians, of the tri - or quadri-consonantal root . How-
ever, to restrain oneself to this is to forego the opportunity to provide
explanations for the phono-semantic phenomena contained in the lexi-
con , phenomena that everyone may observe. To work with a theory that
posits that the Sun rotates around the Earth enables one to explain a
certain number of observable phenomena;
50
but to work with a theory
that posits that the Earth rotates around the Sun enables one to explain
a greater range of phenomena, and fts in better with other knowledge
we have of the movement of the stars.
6. References
Ass al-bala =
Ass al-bala,
Abdarrah m Mah md, ed. Bayrt: Dr at -t iba wa-n-nar.
Bahri , A. 2003. Lnantiosmie en arabe, Doctoral thesis. University Paris 8.
Bailly , A. 1950. Dictionnaire grec franais. Paris: Hachette.
Bohas , Georges. 1997. Matrices, tymons, racines, lments dune thorie lexicologique du
vocabulaire arabe. Paris: Peeters.
. 2000. Matrices et tymons, dveloppements de la thorie. Lausanne: Editions du
Zbre.
. 2006. De la motivation corporelle de certains signes de la langue arabe et de ses
implications. Cahiers de linguistique analogique, 3, 1141.
, and N. Darfouf . 1993. Contribution la rorganisation du lexique de larabe, les
tymons non-ordonns, Linguistica Communicatio, 5/12, 55103.
, and A. Janah . 2000. Le statut du dd dans le lexique de larabe et ses implications.
Langues et Littratures du Monde Arabe, 1, 1328.
50
We fnd in Te Legend of Alexander: For the Sun is the servant of Te Lord, which
interrupts its course neither day nor night. Tis idea that the stars are in the service of
God seems to date back at least to Bardesane (born in 154): Neither the Sun, nor the
Moon, nor the other beings which are superior to us in any thing have received power
over themselves, they are on the contrary subject to a law and, consequently, they do
what they have been ordered and never anything else. Te Sun never says I shall not
rise at the given hour, nor the Moon I will no longer have phases, I will neither wax nor
wane, . . . All these creatures are servants and remain subject to a law: they are instru-
ments of the wisdom of Te Lord, Who is infallible. (See Teixidor 1992).
288 georges bohas and abderrahim saguer
, and M. Dat . 2005. La matrice acoustique {[dorsal], [pharyngal]} en arabe clas-
sique et en hbreu biblique, premire esquisse. Mlanges de lUniversit Saint-Joseph,
LVIII, 125143.
, and M. Dat . 2007. Une thorie de lorganisation du lexique des langues smitiques :
matrices et tymons, Lyon: ENS ditions.
, and A.R. Saguer . 2006. Sur un point de vue heuristique concernant lhomonymie
dans le lexique de larabe. In Edzard, L. & J. Watson (eds), Grammar as a Window
onto Arabic Humanism. A Collection of Articles in Honour of Michael G. Carter, Wies-
baden, Harrassowitz Verlag, 130154.
, and R. Serhane. 2003. Consquences de la dcomposition du phonme en traits.
In: Angoujard, J.-P. and S. Wauguier-Gravelines (Eds.): Phonologie. Champs et per-
spectives. Lyon: ENS ditions, 131-155.
Cantineau , Jean. 1951. Le consonantisme du smitique. Semitica, IV, 7994.
Dat , M. 2002. Matrices et tymons. Mimophonie lexicale en hbreu biblique. Doctoral
thesis, Lyon: Ecole Normale Suprieure Lettres et Sciences Humaines.
Dell , F. 1973. Les rgles et les sons: introduction la phonologie gnrative. Paris: Her-
mann.
Diab , S. 2005. La matrice {[coronal], [dorsal]}, Les tymons impliquant le jm, Masters 2
Research Paper. Lyon: cole normale suprieure lettres et sciences humaines.
Guerssel , M. and J. Lowenstamm . 1993. Te Derivational Morphology of the Classical
Arabic Verbal System. ms. UQAM and University Paris VII.
Halle , M. 1991. Phonological Features. In W. Bright (ed.): Oxford International Ency-
clopedia of Linguistics. Oxford: Oxford University Press, 207212.
Hurwitz , S. 1913 [1966]. Root-Determinatives in Semitic Speech, a Contribution to Semitic
Philology. New York: Columbia University Press.
Joon , P. 1923. Grammaire de lhbreu biblique. Rome: Institut biblique pontifcal.
Kazimirski , A. de Biberstein. 1860. Dictionnaire arabe franais, Paris: Maisonneuve et
Cie.
Kenstowicz , M. 1994. Phonology in Generative Grammar. Oxford UK : Blackwell.
Khatef , L. 2003. Statut de la troisime radicale en arabe: le croisement des tymons, Doc-
toral thesis. University Paris VIII.
. 2004. Le croisement des tymons: organisation formelle et smantique. Langues
et Littratures du Monde Arabe, 119138.
Lisn = Jaml ad-Dn Ab l-Fad l Muh ammad b. Mukarram b. Al b. Ah mad b. Ab
l-Qsim b. H abqa Ibn Manz r, Lisn al-Arab , s.d. Abd Allh Al al-Kabr, Muh ammad
Ah mad H asab Allh, Him Muh ammad a-dil, eds. Cairo: Dr al-Marif.
Lipinski , E. 1997. Semitic Languages. Outline of a Comparative Grammar. Leuven:
Peeters.
Mansouri , W. 2006. La place du trait [sonorant] dans les matrices de larabe. Masters 2
Research Paper, Lyon, cole normale suprieure lettres et sciences humaines.
McCarthy , J.J. 1986. OCP Efects: Gemination and Antigemination. Linguistic Inquiry,
17,2. 207263.
Nyckees , V. 1998. La smantique, Paris: Belin.
Qms = Majd ad-Dn Muh ammad b. Yaqb al-Fayrzbd, Al-Qms al-Muh t .
Bayrt: Mu
assasat ar-Risla.
Quiniou , Y. 2006. La mort scientifque de Dieu. Le nouvel observateur. Hors-srie,
3841.
Le Petit Robert , dictionnaire alphabtique et analogique de la langue franaise. 1967
[1993]. Paris: Dictionnaires Le Robert.
Saguer , A.R. 2000. Lincrmentation des prfxes dans le lexique de larabe. Le cas du n.
Actes du colloque Journes de linguistique arabe et smitique, Langues et littratures du
monde arabe, 1, 5782.
. 2002a. Lincrmentation des prfxes dans le lexique de larabe. Le cas du m.
Langues et littratures du monde arabe, 3, 2957.
explanation of homonymy in the lexicon of arabic 289
. 2002b. Z hirat al-isbq f l-jud r al-arabiyya. Agadir: publications de lUniversit
Ibn Zuhr.
. 2003. La matrice {[+nasal), [coronal)}, traction en arabe. Premire esquisse.
Langues et littratures du monde arabe, 4, 138183.
Segeral , Ph. 1995. Une thorie gnralise de lapophonie, Doctoral thesis. University
Paris 7.
Serhane , R. 2003. Etude de la matrice {[labial], [dorsal]} en arabe, Doctoral thesis. Uni-
versity Paris 8.
Teixidor , J. 1992. Bardesane dEdesse, la premire philosophie syriaque. Paris: Les ditions
du Cerf.
Yeou , M. and S. Maeda . 1994. Pharyngales et uvulaires arabes sont des approxim-
antes: caractrisation acoustique. 20me Journes dtudes sur la Parole, Trgastel,
409414.
THE PERIPHRASTIC BILINGUAL VERB CONSTRUCTION AS A
MARKER OF INTENSE LANGUAGE CONTACT .
EVIDENCE FROM GREEK , PORTUGUESE AND
MAGHRIBIAN ARABIC
Louis Boumans
University of Nijmegen
1. Introduction
I would like to start this paper with some terminological notes. I use
the dichotomy community language versus superimposed language
to refer to the typical unequal social-economic status of the bilingual
speakers languages. Tese terms refer to local as well as global power
relations and their sociolinguistic consequences. For instance, whether
in Portugal, Brazil or the United States, Portuguese speaking people
learn English in order to gain access to valuable information and upward
social mobility , i.e. education, media and employment.
I will use the term socially dominant for the language the individual
speaker is most exposed to in her daily life. Tis could be either the
community or the superimposed language , depending on the local situ-
ation. Tus English is more likely to be socially dominant for a particular
Portuguese /English bilingual member of the Portuguese community in
the US, while Portuguese will be socially dominant for most bilinguals
living in Portugal.
Te terms matrix language and embedded language are grammati-
cal notions referring solely to local syntactic units of analysis in bilingual
speech. Te higher order constituent is the matrix in which lower order
constituents are embedded. In mixed sentences higher and lower order
constituents are in diferent languages. In most instances the community
language functions as the matrix language and superimposed language
elements are embedded. However, the reverse occurs in a minority of
cases so the terms should not be confused.
1
1
Since the early 1990s, Carol Myers-Scotton has been the most infuential promoter of
the insertion approach to code-switching and the terms matrix and embedded language .
292 louis boumans
Te present article deals with verbs from a superimposed language
that function as embedded elements in community language discourse.
Te community language may or may not be socially dominant , and I
argue that this makes a diference for the way in which foreign verbs are
embedded.
2. Te integration of foreign verbs
Tere are three ways in which foreign verbs are integrated in the matrix
language, two of which are common. One is the complete morphologi-
cal integration . Some basic form of the foreign verb , typically the verb
stem or the infnitive , is treated as the verb stem of the receiving matrix
language , and verbal categories of the latter are expressed by matrix
language morphology . Gloss (1) is a Moroccan Arabic /French example
showing the French verb stem montr- (from montrer [m
~
tre] to show)
with an Arabic prefx and sufxes .
2
(1) wa adi y-montr.i-w-l-ek . . .
Q FUT 3-show-PL-to-2SG
Are they going to show you . . .? MA
3
/French (Wernitz 1993, 308)
Te insertion of foreign verb stems without overt morphological inte-
gration in matrix languages lacking verbal morphology, such as various
Austronesian languages (Van Staden 1999) can be considered as a sub-
category of the morphological integration strategy, even if the integra-
tion is not overtly expressed by ML morphemes.
I concur with the fundamentals of her original Matrix Language Frame (MLF) model
(Myers-Scotton 1993), except for the defnition of the matrix language . In my view, all
syntactic constituents function as a matrix for lower-order constituents , whereas in the
MLF model only the Complementizer Phrase functions as a matrix. I refer to earlier
work for more details on this approach to code-switching (Boumans 1998, Boumans
and Caubet 2000, Boumans 2002).
2
Te vowel i is not a proper sufx . Embedded French verbs are modelled on a class
of Arabic verbs ending in a vowel. Tis vowel is subject to a/i ablaut. Cf. Caubet (1993),
Boumans (1998), and Boumans and Caubet (2000).
3
Te following abbreviations are being used: in the main text: MA Moroccan Arabic;
in the glosses to numbered examples: 1,2,3 frst, second, third person; ACC accusative;
AGR agreement; ART article; AUX auxiliary; DEF defnite article; FUT future tense;
IMPF imperfective; INF infnitive; M masculine; NEG negation; PASTPART past parti-
ciple; PL plural; PROGRPART progressive participle; PRT preterit; REL relative clause
marker; SG singular.
the periphrastic bilingual verb construction 293
Te other common strategy is the periphrastic construction . Tis
strategy is equally common in the worlds languages and can also be
illustrated with Moroccan Arabic , this time in contact with Dutch :
(2) ba y-dir li-h aanvall-en
want 3-do to-3SG attack-INF
It [the bird] wanted to attack him. MADutch (Jamal, 16, Utrecht 2000)
In his book Bilingual Speech (2000), Muysken presents a typology of
verb integration with special attention for the periphrastic type. In the
periphrastic construction , he argues, the foreign verb can be a nominal-
ization , an infnitive or an adjunction . In the frst case, the nominalized
verb is the complement of the helping verb, in the case of an infni-
tive the helping verb must be analysed as an auxiliary . In the adjunction
analysis, the foreign verb and the helping verb form a kind of verbal
compound .
Which of these three analyses is most appropriate does not only depend
on the language pair involved. As I discussed elsewhere (Boumans 1998;
2000), strategies and grammatical constructions difer among individu-
als belonging to the same bilingual community. Compare examples (2)
and (3). In Jamals utterance in (2), the patient participant of the Dutch
transitive verb aanvallen to attack is expressed as the MA indirect object
li-h to-him. Te Patient role is expressed as an indirect object , because
the Dutch verb occupies the Direct Object position of the MA transitive
verb dar to do. Tis is in fact the common construction in the Dutch
Moroccan community . Now in Samirs example in (3), the patient of the
transitive verb controleren to supervise surfaces as the direct object suf-
fx -hom them. Te construction in (2) can be described as a helping
verb plus a verbal noun object , while the one in example (3) is in fact
very similar to the monolingual Dutch auxiliary construction . Compare
(3) with its Dutch translation in (4). Note that the pronominal direct
object of the main verb cliticizes to the fnite auxiliary , even if an adjunct
constituent (morgen tomorrow) comes between the two verbs .
(3) kun adi y-dir-hom controler-en?
who FUT 3do-3PL supervise-INF
Who will supervise them? MA/Dutch (Samir 20, Nijmegen 1991)
(4) wie gaa-t hun (morgen) controler-en
who go-2SG 3PL tomorrow supervise-INF
Who will supervise them (tomorrow)? Dutch translation of (3)
294 louis boumans
Te variation between speakers, and even between utterances of the
same speaker, can be explained by the varying levels of competence in
both languages, and by the efect of grammaticalization and conven-
tionalization (Backus 1996).
Te third, less common strategy for the incorporation of foreign
verbs deserves mentioning here for the sake of completeness. Tis stra-
tegy consists in inserting infected verb forms rather than verb stems .
Te infected foreign forms are mapped onto the ML paradigm and
express ML verbal categories. A famous case in point is Mednyj Aleut ,
also known as Copper Island Aleut (Tomason and Kaufman 1988,
2338). In this variety of Aleut , Russian infectional patterns in fnite
verbs replaced Aleut ones while most other grammatical subsystems
remained intact. Crucially, the Russian verbal infections express Aleut
tense and aspect categories. Similarly, Igla (1991) reports on a dialect of
Romani nowadays spoken in a suburb of Athens in Greece. Te speakers
of this dialect of Romani moved in from Turkey in the 1920s but no lon-
ger speak any Turkish . However there are still approximately 30 verbs of
Turkish origin that continue to be infected with Turkish sufxes while
following the Romani verbal paradigm, cf. Table 1. Tat is, the Turkish
infectional sufxes express Romani infectional categories.
TABLE 1 THE PRESENT TENSE OF TURKISH VERBS AND OF VERBS
OF TURKISH AND NON-TURKISH ORIGIN IN AJIA VAVARA ROMANI
(FROM IGLA 1991, 52)
Turkish Ajia Vavara Romani
present to write to write to bring
sg 1 yaz-ar-m yaz-ar-um an-av
sg 2 yaz-ar-sn yaz-ar-sun an-es
sg 3 yaz-ar yaz-ar an-el
pl 1 yaz-ar-z yaz-ar-us an-as
pl 2 yaz-ar-snz yaz-ar-sunus an-en
pl 3 yaz-ar-(lar) yaz-ar-(lar) an-en
(5) i thagarni kurta-du len e rom-en
ART queen save-PRT REL ART gypsy-ACC.PL
(..) the queen saved (them) the gypsies. Romani/Turkish (Igla 1991, 53)
Tis latter strategy is actually very rare in language contact situations.
However, it may be more common in the case of two closely related
languages or varieties. For instance Malkiel (1986) draws attention to
the Spanish conjugation pattern of a number of Portuguese verbs . Also
the periphrastic bilingual verb construction 295
non-tensed verb forms such as participles , infnitives and imperatives
are quite readily interchangeable between languages, cf. (6).
(6) Le estaba poniendo atencin qu estaba
3.IO was.3SG put.PROGRPART attention what was.SG
recorded
record.PASTPAR
He was paying attention to what was recorded. Spanish/English (Pfaf
1979, 300)
3. Explanations for the distribution of the two common strategies
Why do speakers opt for one strategy rather than the other? Te two
common strategies in particular, morphological integration and peri-
phrastic constructions , allow us to compare the languages and linguistic
situations involved.
3.1 Characteristics of the matrix language
When one makes a list of all language contact situations and the attested
strategies for the integration of foreign verbs , it becomes clear that
genetic or areal typological factors play an important role. In the Indo-
Iranian and Turkic languages, for instance, the periphrastic construc-
tion appears in virtually all language contact situations. In connection
with this, Muysken (2000, 194) speaks of a large linguistic area, in
this case, stretching from Sri Lanka to Greece. On the other hand, most
western European languages seem to prefer the morphological integra-
tion strategy. A plausible explanation for the areal bias is that speakers
who are accustomed to a certain strategy of incorporating foreign verbs
will reuse this strategy in new contact situations. Some examples of this
kind of bilingual knowledge will be discussed below. Still, the two MA
examples cited in (1) and (2) show that the matrix language is not the
only factor deciding which incorporation strategy speakers will use.
3.2 Characteristics of the embedded language
One way to explain the diference between MA /French and MA /Dutch
code-switching is to look for diferences in the superimposed embed-
ded languages . Heath (1989) argues that the phonological shape of the
French verbs facilitates their incorporation in Arabic . Bilingual speakers
associate the stressed fnal vowel /e/ of most French infnitives and other
296 louis boumans
verb forms with the fnal /i/ of the prefx conjugation (imperfective ) of
a class of MA verbs .
Dutch infnitives typically end in an unstressed sufx -en that is pro-
nounced as a schwa . It is not obvious whether this makes them phono-
logically less similar to the MA fnite verb in /i/ than French infnitives .
Firstly, MA phonology does not allow for the schwa in open syllables.
Tis may lead Moroccan listeners to ignore word-fnal schwas in Dutch ,
and interpret the infnitives as consonant-fnal. Alternatively, however,
Moroccans may interpret the Dutch fnal schwa as a full front or back
vowel . Both tendencies can be observed in the speech of Moroccan
learners of Dutch .
Moreover, the vocalic ending in French verbs cannot be decisive,
since Spanish , Italian and English verbs are morphologically integrated
in North African varieties of Arabic in the same way as French verbs .
Many of these Romance and English infnitives end in a consonant.
Arabic /Spanish language contact is still common in the (formerly)
Spanish occupied northern parts of Morocco (cf. Heath 1989, Her-
rero Muoz-Cobo 1996). Cohen (1912) notes interesting observations
on the Jewish dialect of Algiers, where the Spanish (or Lingua Franca)
infnitive ending -ar is even extended to embedded French verbs. E.g.
refuzarit from French refuser [rfyze] in (8). Numerous verbs of Italian
and English origin are found in Maltese , another variety of Maghribian
Arabic (Aquilina 1965 [1987], Camilleri 1994, Mifsud , 1995).
4
Te stem
extension -ja- in Maltese verbs of English origin, as in (10), is a refec-
tion of the Sicilian ~ Italian infnitive sufx -are.
(7) frin-ar -t ~ frin-ar .i-t
brake- INF-1SG ~ brake- INF.STEM EXTENSION -1SG (Sp. frenar)
I braked Tetouan Arabic/Spanish (Heath 1989, 184)
(8) refuz-ar-it
refuse-SPANISH INF-1SG (Fr. refuser)
I refused Jewish Arabic of Algiers/French (Cohen 1912, 432)
(9) ti-ppartiipa-w
3-participate-PL (It. partecipare)
you (pl) participate Maltese/Italian~ Sicilian (Camilleri 1994, 437)
4
As a matter of fact, the type of verb integration illustrated in (1) does not occur in
the Middle Eastern varieties of Arabic . Instead the periphrastic construction is more
common in that region.
the periphrastic bilingual verb construction 297
(10) ni-bbli.ja-ha
1-bleach.STEM EXTENSION -3SG
I bleach it Maltese/English (Camilleri 1994, 443)
Tus, verb stems from various embedded languages and with diverse
phonological characteristics can be morphologically integrated into
Moroccan or Maghribian Arabic , and phonology does not seem to be an
explanation for the periphrastic construction in the case of MA /Dutch .
A second way in which the embedded language might infuence the
selection of the verb integration strategy is the use of periphrastic verb
constructions in the embedded language itself. Following this line of
thought, the bilingual MA /Dutch periphrastic construction could be
inspired by periphrastic constructions with doen to do in standard or
non-standard Dutch .
5
Again, the comparison with the other language
pairs speaks against this hypothesis. Periphrastic constructions with a
do verb are much more common in English than they are in Dutch ,
while their abundance in French and Spanish may be similar to that in
Dutch .
6
Terefore, the occurrence of do constructions in monolingual
Dutch likewise does not explain why the periphrasis strategy is chosen
to insert Dutch verbs in MA matrix clauses.
3.3 Characteristics of the sociolinguistic setting
In the case of the North African language contacts , neither the host lan-
guage nor the embedded language is the sole factor determining the way
in which the loan verbs are integrated in Arabic . For this reason I con-
jecture that the sociolinguistic setting in the Netherlands, where MA is
a minority language , is responsible for the periphrastic construction in
MA /Dutch . As a more general hypothesis, I suggest situations in which
language contact is more intense, like migration to an urban industria-
lised society, favour the use of the periphrastic construction .
7
5
Jacomine Nortier and Roeland van Hout made this suggestion when we discussed
this paper at the SS14 workshop on borrowing in Gent, April 4, 2002, and on earlier
occasions.
6
With the exception of certain infrequent constructions, do-periphrasis in Dutch is
associated with non-standard regional varieties or child language. Cf. Nuijtens (1962,
15457), Giesbers (1984), and Cornips (1994).
7
One could argue that the morphological integration of French verbs is found both
in North Africa and in the North African diaspora in Francophone Europe and Can-
ada. In this particular case, however, code-switching strategies that were already estab-
lished in the homeland have simply been maintained in the diaspora.
298 louis boumans
In order to shed some light on this issue and to test the intensity of
contact hypothesis, the following two sections survey verb integration
in two additional language pairs for which both integration strategies
are attested: Greek /English and Portuguese /English . In both cases the
infuence of English has been described for immigrant communities in
Anglophone countries as well as in the Greek and Portuguese speaking
homelands.
4. Greek /English
4.1 Greece
Te morphological integration of Italian verbs consists of adding Greek
sufxes to the Italian infnitive in -are.
8
(11) It. posare to pose
It. schizzare to sketch
Mainland Greek/Italian (Van Dijk-Wittop Koning 1963)
(12) fundro It. fondare to anchor
barkro It. barcare to board
kompletro It. completare to fll up [cargo]
Mainland Greek/Italian (Hartley 2001)
According to Swanson the mildly productive verbal sufx -aro entered
the Greek language in late Byzantine times (1958, 40). Te Italian infni-
tive marker has been generalised to verbs of French and English origin
as well:
(13) v Fr. dbuter to make ones debut
Fr. lancer to launch
Mainland Greek/French (Contossopoulos 1978, 42)
(14) fertaro to firt
stoparo to stop (a machine)
sutaro to make a shot (soccer, basketball)
Mainland Greek/English (Swanson 1958)
8
Te authors on loanwords and codeswitching in Greek make use of diferent writ-
ing and spelling conventions. Te spelling of the source publication is retained in the
examples cited here.
the periphrastic bilingual verb construction 299
(15) /parkaro/ to park
v /manadzaro/ to manage
/triparo/ to trip
Mainland Greek/English (Apostolou Panara 1991, 50)
Interestingly, the recycling of a foreign verb marker like the Italian
-are to mark verbs from a third language is yet another example of how
bilingual experience infuences the way in which speakers treat words
from foreign languages. Note the close parallel with -ar in the Jewish
Arabic dialect of Algiers (8), and the -ja stem extension afer English
verb stems in Maltese (10).
9
Most sources on loan verbs in Modern Greek mention only this
-ar- extension with complete morphological integration of the foreign
word. However, Apostolou Panara , while asserting that verbs all adapt,
i.e. are morphologically integrated, mentions that in some cases, along
with the single lexeme we attest a periphrasis with the Greek verb /kano/
(to do, to make) preceding a noun or a gerund (1991, 50), cf. the exam-
ples in (16). A few verbs occur exclusively as part of the periphrastic
construction (17), whereas the periphrasis is judged ungrammatical in
most cases (18).
(16) /fertaro/ besides /kano fert/ to firt
/manadzaro/ besides /kano manadzing/ to manage
/stokaro/ besides /kano stok/ to stock
Mainland Greek/English (Apostolou Panara 1991, 50)
(17) /kano kambing/ but not */kambaro/ to camp
/kano serfng/ but not */serfaro/ to surf
Mainland Greek/English (Apostolou Panara 1991, 51)
(18) /triparo/ but not */kano trip/ to trip
/parkaro/ but not */kano park/ to park
Mainland Greek/English (Apostolou Panara , 1991, 51)
It is possible that a closer study of the English vocabulary in Modern
Greek would reveal regularities that govern the use of periphrasis .
Periphrasis with kno could be associated with particular types of events,
for instance habitual or durative activities . Te periphrastic construc-
tion may be more common with recent English loans , or in the speech
9
In Dutch a similar situation obtains with the French infnitive marker -er [er] (Tref-
fers-Daller 1994), which is sometimes used for English loanwords as well, e.g. formatt-
er-en to format.
300 louis boumans
of speakers who are more exposed to English. kano kambing may be a
calque afer the French expression faire du camping (cf. kno ski in (24)
below, possibly afer French faire du ski). Te available data do not allow
for such conclusions at the moment. Still, I think it is safe to conclude
from the literature that in Greece the periphrastic construction occurs
as a minor pattern with some verbs only, while morphological integra-
tion is the rule.
4.2 Cyprus
Cyprus has been under British rule for more than eighty years, during
which English was the language of administration. First, from 1878 on
Britain administered the island in agreement with the Ottoman Empire.
Ten Cyprus was annexed by Britain when the Ottoman Empire enters
into World War I on the side of Germany, and subsequently the island
became a British Crown colony under the British rule. In 1960 Cyprus
gained independence, and (Modern Standard) Greek became the lan-
guage of administration. Te English language remained infuential
through the tourism industry and the large international community
on the island.
Tere are a number of studies dealing specifcally with the infuence
of English on Cypriot Greek . Papapavlou (1997) cites a list of English
loan words found in written sources and tape-recorded speech. His list
includes 23 verbs in the -aro conjugation. He makes no mention of the
periphrastic construction .
(19) frdro to firt
rejistrro to register
riskro to take a risk
arcro to change (money)
ak
h
ro to check (inspect)
Cypriot Greek/English (Papapavlou 1997)
Similarly, Evripidou , also on the basis of written sources and recorded
speech, lists 19 English -origin Cypriot Greek verbs , all of them inte-
grated in the -aro class (2001). Goutsos discusses Greek/English code-
switching among members of a middle-class Cypriot family in Limassol
(2001). Tis study difers from the other ones, as Goutsos does not list
English loan words but rather focuses on the discourse functions of
language choice in the conversations. He does not comment on inte-
grated verbs of the -aro conjugation. He does cite three instances of the
the periphrastic bilingual verb construction 301
periphrastic construction . As in Apostolou Panara s data, there are two
types of construction, one in which the English verb is represented by
the infnitive (or verb stem ), see (20) and (21), the other with the ger-
und (22).
(20) a kmno shower
I will shower
Cypriot Greek/English (Goutsos 2001, 203)
(21) i jti tan kmnis wash
no because when you wash
Cypriot Greek/English (Goutsos 2001, 203)
(22) . . . llun tus pu na rti i jaj su en kmni swimming
I was saying to them, when your grandma comes, she will swim.
Cypriot Greek/English (Goutsos 2001, 203)
Te periphrastic construction is not a pervasive phenomenon in the
speech of Goutsos informants, nor is code-switching in general (Gout-
sos 2001, 216). Te examples also show the absence of an established
convention on how to insert the English verb into the periphrastic con-
struction . Still it is noticeable that the do-construction is recorded on
Cyprus and not in mainland Greece. Goutsos comments: Compared
to SMG [Standard Modern Greek ], CG [Cypriot Greek ] draws more on
English resources for borrowing and creates mixed compound forms
with de-lexicalized verbs (2001, 204).
Although Apostolou Panara shows that the periphrastic construc-
tion does occur in mainland Greek , Goutsos s comment confrms the
impression that it is a marginal phenomenon. If it is indeed true that
the periphrasis is more common in Cyprus, it is attractive in the light
of the present discussion to explain this diference as resulting from the
greater impact of English on the language of the former crown colony.
Moreover, Goutsos s examples stem from a really bilingual setting,
whereas the loanwords collected by the other authors stem mostly from
written sources that do not presuppose a bilingual readership.
4.3 Mainland Greek in the Diaspora
Te oldest study on American Greek mentions just one English -origin
verb : mouvaro to move (Lontos 192526). However, this particular
form probably goes back to an Italian form muevere, as Hartley (2001)
suggests in his paper on loan words in Greek nautical terminology.
302 louis boumans
A much more elaborated study of American Greek is provided by Sea-
man . He mentions the same form muvro (1972, 165), but it becomes
clear that the periphrastic construction with kno to do, make is the
productive way to incorporate English verbs in his data: In verbal loans
from English, /jno/ seems to be the auxiliary for passives and /kno/ for
actives (1972, 166). His examples of the passive construction with jno
to be, become are ambiguous, as they might also be analysed as cases
of a copula plus predicate, and there are only a handful of examples. On
the other hand, Seaman cites 47 examples of the periphrastic construc-
tion with kno. In 33 cases the embedded English element is unambigu-
ously a verb ; 11 cases may involve either a verb or a noun , e.g. control,
welcome. Some examples are reproduced below (see also Muysken 2000,
212).
10
(23) kno, cover up I cover up
knis, brush ta ndja su You (sg) brush your teeth
kane, punch s/he punched [a meal ticket]
American Greek (Seaman 1972, 16768)
Tamis found the same construction in Australia (1986, cited in Muysken
2000, 212). Te construction is also known from Canada (24) and Brus-
sels (25). Compared with Seaman s study, the information on the latter
diaspora communities is rather scarce. It is not clear whether in these
cases the periphrastic construction is really the productive strategy for
incorporating English or French verbs . Te Canadian examples cited
in (24), involving computer terminology and sports, remind of the
use of the periphrastic construction in Greece described by Apostolou
Panara .
(24) kno ski
kno save [computer]
kno print [computer]
kno jogging
Montreal Greek/English (Muysken 2000, 213, based on Hatzidaki p.c.)
(25) kno dmnager to move house
Greek/French (Muysken 2000, 213, based on Hatzidaki p.c.)
10
In Table 7.7 on p. 212 of Muysken s book the words jno and kno have been
reversed.
the periphrastic bilingual verb construction 303
4.4 Cypriot Greek in London
Tere are a number of publications (and many unpublished theses) on the
language of the Greek Cypriot community in London. Gardner- Chloros
(1992) mentions the occurrence of morphologically integrated verbs
like tekaro to check, which is not surprising as these words are found
in Cyprus (and Greece) as well. In addition, however, she points out the
periphrasis with kno as an innovative construction, occurring with Eng-
lish adjectives, nouns , participles . . . (Gardner-Chloros 1992, 127).
(26) kamno use to use
kamno respect to respect
kamno developed to develop
kamno spelling to spell
Cypriot Greek in London (Gardner-Chloros , 1992, 127)
Zarpetea compares the language use of three generations of London
Cypriots (1995). She notices the use of both integrated loan verbs and
the periphrastic construction in the frst as well as the second-generation
immigrants. Te periphrastic construction still occurs in the speech of
the third generation, young children who use only little Greek :
(27) wash cups, do you want me to do the plates now mum?
I have washed the cups, . . .
Cypriot Greek/English (Zarpetea 1995, 581)
As a fnal note on morphologically integrated English verbs in immi-
grant Greek , I would like to draw the attention to the fact that there is
much overlap between the lemmas mentioned in the various sources.
Tis is particularly striking in Zarpetea s paper. Out of the seven verbs
she cites, fve also occur in other sources, cf. Table 2.
TABLE 2 OVERLAP BETWEEN LOAN VERBS IN ZARPETEA (1995)
AND OTHER SOURCES
Zarpetea 1995, 578 other sources
to cancel Gardner-Chloros 1992, 126, Papapavlou 1997,
232, Evripidou 2001, 24
to check Papapavlou 1997, 240, Evripidou 2001, 24
to charge Papapavlou 1997, 240
to park Apostolou Panara 1991, 50, Papapavlou 1997,
236
to move Lontos 192526, 309, Seaman 1972, 167, Hartley
2001
to book
to packet
304 louis boumans
Tere is no such overlap in the examples cited for the periphrastic con-
struction with kmno (mainland Greek kno).
11
Tis indicates that the
morphologically integrated verbs found in London may well have been
coined in the more monolingual setting in Cyprus. As noted above,
to move is probably not even of English origin. Tis is a fur-
ther indication that in the Cypriot diaspora, the periphrastic construc-
tion replaces morphological integration as the most productive strategy
for the incorporation of new English verbs . Te chances are that the
same development has taken place in the American Greek community
studied by Seaman (1972).
5. Portuguese /English
I have been able to trace surprisingly few studies on English loan words
in Portuguese , whether European or Brazilian . Compared with Greek,
there are also few studies on Portuguese speaking communities in
Anglophone countries. Te picture that arises from the available data is
approximately the same as for Greek.
5.1 Portugal
Te website of the Instituto Superior Politcnico de Viseu in Portugual
hosts a short list of English loanwords in Portuguese , compiled by two
students and their English teacher (Queiroz , Rodrigues and McKenny
1999, accessed June 2002). Te authors are not explicit about their data
sources, except that visitors of the web page are invited to contribute
observations. As it concerns a Portuguese project, however, I assume
that most of their sources are likewise of Portuguese, rather than Brazil-
ian , origin.
Teir list of English loan words contains four verbs integrated in
the Portuguese conjugation class ending in -ar (28), in addition to the
11
Apostolou-Panara, Goutsos, Gardner-Chloros and Zarpatea cite 14 diferent Eng-
lish verbs in the periphrastic construction . kmno wash is shared between Goutsos and
Zarpatea. kmno use is cited in the papers by Gardner-Chloros and Zarpetea. But these
two papers must be based on (partly) the same data. I infer this from the fact that the
example very busy I know that you have (are) very busy is cited in both
(Gardner-Chloros 1992, 128; Zarpatea 1995, 578).
the periphrastic bilingual verb construction 305
denominal form stressar (from stress) and the de-adjectival compactar
(to compress from compact, computer terminology).
(28) drafar to draf
driblar to dribble
linchar to lynch
snifar to snif [drugs]
European Portuguese/English (Queiroz et al. 1999)
In addition this list contains six deverbal English nouns denoting activi-
ties in the domains of sports and information and communication tech-
nology: jogging, kickboxing, surf; scan, upgrade and zap. Te web site
does not indicate how these words are used in Portuguese . Nicholas
Hurst (p.c. 18 Apr 2002) cites examples of periphrastic constructions
involving the Portuguese verb fazer to do (29). Te examples show that
there is much variation: as in Greek contexts, the English verb is used in
either the citation form (infnitive ) or the gerund . Moreover, the indef-
nite article um may or may not be present.
(29) fazer zapping
fazer um scanning
fazer surf
fazer um upgrade
European Portuguese/English (p.c. Nicholas Hurst 18 Apr 2002)
5.2 Brazil
Kennedy (1971) and Harmon (1994) quote a number of English verbs
occurring in Brazilian dictionaries and other written sources. All are
integrated into the so-called frst conjugation class ending in -ar:
(30) chutar to shoot [ball game]
drenar to drain
ranquear to rank
treinar to train
Brazilian Portuguese/English (Kennedy , 1971)
Harmon does not mention the construction with fazer. Kennedy cites
just one example of a somewhat enigmatic periphrastic construction :
fazer o footing, translated in English as take a stroll.
306 louis boumans
5.3 American Portuguese
Tere is a classical study of the language of Portuguese immigrants
in the US by Pap (1949). He gives many examples of morphologically
integrated English verbs . Tese are added to the most frequent en most
productive Portuguese conjugation class in -ar, with the necessary pho-
nological adaptations . In some cases the English verb stem is extended
with an -e- such that the infnitive ends in -ear.
(31) raid-ear to ride
canec-ar to connect
damp-ar to dump
s(e)leir-ar to slide
chinj-ar to change
raiv-ar to drive
North American Portuguese/English (Pap 1949, 95100)
While morphological integration of English verbs is the most common
pattern, Pap also lists a large number of instances of a periphrastic con-
struction with fazer. Te nominalized English verb stem occupies the
direct object position. In contrast with the contemporary European
examples cited by Hurst , the foreign verb stem is preceded by the Portu-
guese masculine defnite article o.
(32) fazer o boda to bother
fazer o spoil to spoil
fazer o save to save
fazer o fnd out to fnd out
fazer o give up to give up
North American Portuguese/English (Pap 1949, 105106)
6. Discussion
Te data from North African varieties of Arabic in contact with Euro-
pean languages suggest that diaspora communities are more inclined
to use the periphrastic alternative for incorporating foreign verbs . Te
studies on Greek and Portuguese in contact with English do not unam-
biguously confrm this. Morphological integration as well as periphras-
tic constructions are reported for all contact situations, whether in the
Greek and Portuguese speaking countries or in the respective immi-
grant communities in Anglophone countries.
the periphrastic bilingual verb construction 307
Broadly speaking, however, the literature on Greek and Portuguese
does point in the same direction. Integration in the productive conju-
gation classes, Greek -ro and Portuguese -ar, is predominant in the
studies on English loanwords in Greece, Cyprus, Portugal and Brazil.
Do-constructions, on the other hand, feature prominently in the stud-
ies on migrant communities.
A caveat is in order: studies on the infuence of English as a foreign
language are traditionally oriented toward loan words , i.e. words in
isolation, in written sources. Tese studies tend not to identify foreign
verbs in the periphrastic construction as verbs at all, which can easily
lead to the suggestion that embedded verbs are either absent or consis-
tently integrated into the morphology of the matrix language . For obvi-
ous reasons, studies on bilingualism in migrant communities focus on
oral communication. Tanks to the recognition of code-switching as
an interesting linguistic phenomenon, the latter studies also pay more
attention to the higher degrees of linguistic organisation like clauses and
utterances. Hence the emphasis on one verb strategy or the other may be
to some extent an artefact of the diferent research traditions.
Still, at least the Greek case supports my hypothesis, in particular
the studies by Apostolou Panara on Greece and Seaman on the Greek
community in the US. Both authors discuss the diferent types of verb
integration , and designate one as the default strategy: morphological
integration in Greece and periphrasis in the US. Te studies on the Lon-
don Cypriots suggest that periphrasis is the most productive strategy
there as well.
With respect to the sociolinguistic factors in verb integration , I con-
clude that, all other things being equal, morphological integration is
the norm in homeland settings where the matrix language is socially
dominant , while the periphrastic construction is found in migrant com-
munities where the embedded language is dominant. Te migrant com-
munities concerned here are actually undergoing language shif . Stated
in more general terms, the periphrastic construction characterises situa-
tions of intense contact with the language of the embedded foreign verb .
As mentioned above, other factors may override the sociolinguistic
ones, as in the case of matrix languages that opt for the same integration
strategy in no matter what sociolinguistic setting.
308 louis boumans
6.1 Automatization of superimposed language practices
Te next question is why speakers more ofen revert to periphrasis
in situations of more intense contact . I can think of the following
explanations:
Firstly, automatization of superimposed language practices: In gen-
eral the degree of integration of foreign words is inversely proportional
to the speakers knowledge of the embedded language . With increasing
practise and competence , the linguistic habits characteristic of that for-
eign language become entrenched in the speakers brain, and can more
easily be reproduced. Tis even results in the disintegration of formerly
integrated foreign words, a process known as denativization (Haugen
1950). In Morocco, for instance, French -Moroccan Arabic bilinguals
now prefer to reborrow the same French terms in shapes closer to the
French prototype (Heath 1986, 114). Denativization has ofen been
described with respect to phonology of loan words from English , a lan-
guage whose infuence is expanding (cf. Khubchandani 1968, Hasselmo
1969, De Reuse 1994, Van Ness 1994). I would like to stress this point,
because it is a commonly held belief that loanwords become, on the
contrary, more integrated over time (Nortier and Schatz 1988, Heath
1989).
Te replacement of morpho-phonological integration by periphrasis
as the productive strategy for the integration of foreign verbs fts into
the denativization process, since the periphrastic construction entails
little or no phonological or morphological adaptation to the matrix lan-
guage . Tis type of change could be taking place in for instance Greece
or Portugal, where exposure to English is still increasing.
Denativization refers to a diachronic change in situations where lan-
guage contact increases over time. However, the underlying mechanism
of automatization of superimposed language practices also explains the
diferences between immigrant and homeland bilingualism .
6.2 Loss of community language practices
Secondly, attrition or loss of community language practices: the mor-
phological integration of foreign verb requires the application of matrix
language morphological procedures. Tese procedures are typically the
same as those used for deriving verbs from nouns or adjectives . Te
loss or erosion of community language morphology , a common efect
of decreased use of the language and incipient shif (Andersen 1982,
the periphrastic bilingual verb construction 309
El Aissati 1997), cuts of morphological integration as a strategy for
embedding foreign verbs .
6.3 Collocational complements
Tirdly, triggering by collocational complements of the embedded verbs :
Many examples of periphrasis with foreign verbs involve a collocation
with a foreign direct object noun or other complement, cf. wash and
cups in Greek /English example (27) above and diep, gesprek and voeren
in the Arabic /Dutch example below.
(33) ma ne-qder eh n-dir
NEG 1-can NEG er 1-do
diep-e gesprekk-en voer-en
deep-AGR conversation-PL conduct-INF
I cant er carry out deep conversations.
MA/Dutch, Jamal, 20, Eindhoven 1991 (Boumans 1998, 245)
Te periphrasis leaves the original embedded language collocation
intact. Terefore the collocation might trigger this construction. Te
usage of foreign collocations presupposes a higher level of competence
in that language and hence more intense exposure to it.
7. Conclusions
In nearly all cases, foreign verbs are combined with matrix language
infection in either of two ways: they are morphologically integrated and
infected with ML afxes (if any), or they are embedded in a periphrastic
construction with an infected ML verb . Some MLs, such as Magribian
Arabic , Portuguese and Greek , allow for both solutions. A comparison
of Magribian, Portuguese and Greek bilingual communities suggests
that the periphrastic integration of foreign verbs is favored in situations
of more intense contact with the superimposed language .
8. References
Andersen , R. 1982. Determining the Linguistic Attributes of Language Attrition. In
R. Lambert and B. Freed, eds. Te Loss of Language Skills. Rowley, MA: Newbury
House, 83118.
Apostolou Panara , A. 1991. English Loanwords in Modern Greek: an Overview. Termi-
nologie et Traduction, 1.91, 4560.
310 louis boumans
Aquilina , J. 1965 (1987). Teach Yourself Maltese. Valetta: Hodder and Stoughton.
Backus , A. 1996. Two in One: Bilingual Speech of Turkish Immigrants in the Netherlands.
Tilburg: Tilburg University Press.
. 2001. Synchronic Register Variation in Monolingual and Bilingual Turkish (project
description).
Boumans , L. 1998. Te Syntax of Code-switching. Analysing Moroccan Arabic/Dutch
Conversations. Tilburg: Tilburg University Press.
. 2000. Periphrastic Verb Constructions in Moroccan Arabic/Dutch Code-switch-
ing. In A. Fenyvesi and K. Sandor, eds. Language Contact and the Verbal Complex of
Dutch and Hungarian. Working Papers of the 1st Bilingual Language Use Teme Meet-
ing. Szeged: University of Szeged, Teacher Training College, 6784.
. 2002. Repetition Phenomena in Insertional Codes-witching. In A. Rouchdy, ed.
Language Contact and Language Confict Phenomena in Arabic. London: Curzon,
279316.
, and Caubet , D. 2000. Modelling Intrasentential Code-switching: a Comparative
study of Algerian Arabic/French in Algeria and Moroccan Arabic/Dutch in the Neth-
erlands. In J. Owens, ed. Arabic as a Minority Language. Berlin: Mouton de Gruyter,
pp. 11380.
Camilleri , A. 1994. Language contact between Maltese and English: code-switching
and cross-linguistic infuence. In D. Caubet and M. Vanhove, eds. Actes des premires
journes internationales de dialectologie arabe de Paris. Paris: INALCO, 43149.
Cohen , M. 1912. Le parler arabe des juifs dAlger. Paris: H. Champion.
Contossopoulos , N.G. 1978. Linfuence du franais sur le grec: emprunts lexicaux et
calques phrasologiques. Athnes: s.n.
Cornips , L. 1994. De hardnekkige vooroordelen over de regionale doen+infnitief-
constructie. Forum der Letteren, 35.4, 28294.
De Reuse , W. 1994. English Loanwords in the Native Languages of the Chukotka Pen-
insula. Anthropological Linguistics, 36.1, 5668.
El Aissati , A. 1997. Language Loss among Native Speakers of Moroccan Arabic in the
Netherlands. Tilburg: Tilburg University Press.
Evripidou , D. 2001. Lexical Borrowing: A Study of English Loanwords in the Greek Cypriot
Dialect. Lancaster: Lancaster University.
Gardner-Chloros , P. 1992. Te Sociolinguistics of the Greek Cypriot Community in
London. Plurilinguismes, 4, 11236.
Giesbers , H. 1984. Doe jij even lief spelen? Notities over het perifrastisch doen. Mede-
delingen NCDN, 19, 5776.
Goutsos , D. 2001. A Discourse-analytic Approach to the Use of English in Cypriot
Greek Conversations. International Journal of Applied Linguistics, 11:2, 194223.
Harmon , R.M. 1994. Aspectos lingsticos dos emprstimos em portugus. Hispania,
77:3, 463469.
Hartley , A.H. 2001. Loanwords in Modern Nautical Greek. (2002).
Hasselmo , N. 1969. On Diversity in American Swedish. Svenska Landsml och Svenskt
Folkliv, 92, 5372.
Haugen , E. 1950. Te Analysis of Linguistic Borrowing. Language, 26:2, 21031.
Heath , J. 1986. Hasta la Mujerra! and other Instances of Playful Language Mixing in
Morocco. Mediterranean Language Review, 2, 1136.
. 1989. From Code-Switching to Borrowing. A Case Study of Moroccan Arabic. Lon-
don: Kegan Paul International.
Herrero Muoz-Cobo , B. 1996. El rabe marroqu: aproximacin sociolingstica. Alme-
ra: Universidad de Almera.
Igla , B. 1991. On the Treatment of Foreign Verbs in Romani. In P. Bakker and M. Cor-
tiade, eds. In the Margin of Romani. Gypsy Languages in Contact. Amsterdam: Inst.
ATW, Univ. Amsterdam, 505.
the periphrastic bilingual verb construction 311
Kennedy , J.H. 1971. Te Infuence of English on the Vocabulary of Brazilian Portu-
guese. Hispania 54:2, 327.
Khubchandani , L.M. 1968. Te Gender of English Loanwords in Sindhi. In B. Krish-
namurti, ed. Studies in Indian Linguistics. (Professor M.B. Emeneau sastipurti volume).
Poona: Deccan College, 18088.
Lontos , S.S. 192526. American Greek. American Speech, I, 30710.
Malkiel , Y. 1986. A Spanish Conjugational Model Superimposed on Portuguese. Medi-
terranean Language Review 2, 5166.
Mifsud , M. 1995. Loan Verbs in Maltese. A Descriptive and Comparative Grammar.
Leiden: E.J. Brill.
Muysken , P. 2000. Bilingual Speech. A typology of Code-Mixing. Cambridge: Cambridge
University Press.
Myers-Scotton , C. 1993. Duelling Languages. Grammatical Structure in Codeswitching.
Oxford: Clarendon Press.
Nortier , J. and Schatz , H. 1988. Van nwoordwisseling naar ontlening, een vergelijk-
end onderzoek. Forum der Letteren, 29, 16178.
Nuijtens , E.T.G. 1962. De tweetalige menseen taalsociologisch onderzoek naar het
gebruik van dialect en cultuurtaal in Borne. Assen: Van Gorcum.
Pap , L. 1949. Portuguese-American Speech. New York: Kings Crown Press.
Papapavlou , A. 1997. Te Infuence of English and its Dominance in Cyprus: Reality or
unfounded fears? Journal of Mediterranean Studies, 7:2, 21849.
Queiroz , A., Rodrigues , J. and McKenny , J. 1999. Emprstimos da lngua inglesa para a
portuguesa. English loan words and phrases in Portuguese. Escola Superior de Edu-
cao de Viseu. 2002.
Seaman , P.D. 1972. Modern Greek and American English in Contact. Te Hague:
Mouton.
Swanson , D. 1958. English Loanwords in Modern Greek. Word, 14, 2646.
Tamis , A. 1986. Te State of the Modern Greek Language as Spoken in Victoria. Mel-
bourne, University of Melbourne.
Tomason , S.G. and Kaufman , T. 1988. Language Contact, Creolization and Genetic Lin-
guistics. Berkeley: Univ. of California Press.
Trefers-Daller , J. 1994. Mixing Two Languages. French-Dutch contact in a comparative
perspective. Berlin: Mouton de Gruyter.
Van Dijk-Wittop Koning , A.M. 1963. De continuteit van het Grieks: onderzoek naar de
herkomst van de woordvoorraad van het Nieuwgrieks. Zwolle: W.E.J. Tjeenk Willink.
Van Ness , S. 1994. Die Dimensionen lexicalischer Entlehnungen im Pennsylvanien-
deutschen von Ohio (USA). Zeitschrif fr Dialektologie und Linguistik, 61.3,
27997.
Van Staden , M. 1999. Where does Malay end and Tidore begin? In R.A.C. Dam, ed.
Perspectives on the Birds Head of Irian Jaya, Indonesia. Proceedings of the conference
Leiden, 1317 October 1997. Amsterdam: Rodopi, 691716.
Wernitz , C.J. 1993. Bedingungen und Voraussetzungen fr Sprachwechsel. Eine Untersuc-
hung zum Sprachwechsel bei bilingualen Marokkanern in Frankreich. Frankfurt: Peter
Lang.
Zarpetea , P. 1995. Code-switching and Lexical Borrowing (loanwords) in the Speech
of Tree Generations of Greek Cypriots in London (Harringey). Studies in Greek
Linguistics. Proceedings of the 16th Annual Meeting of the Department of Linguistics.
Tessaloniki: Aristotle University, 57687.
FAULA , FAILA , FAALA :
DISPERSION ET RGULARITS SMANTIQUES
DANS LES TROIS SCHMES SIMPLES DU VERBE ARABE
Joseph Dichy
Universit Lumire-Lyon 2
Pour Kees Versteegh
Aut cuspis sic vita fuit, dum stare videtur.
1. Introduction
1
Les schmes du verbe augment (mazd ) en arabe sont traditionnelle-
ment associs des valeurs smantiques de base, telles que litration ,
lintensivit , la causativit et la factitivit , la rfexivit , etc. Si les lis-
tes dexceptions qui accompagnent ces descriptions indiquent que le
problme, malgr des avances signifcatives, est loin dtre rsolu de
manire satisfaisante, la question des valeurs smantiques associables
aux trois schmes simples (mujarrad ) ou non-augments faula , faila et
faala
2
semble ofrir encore plus de rsistance lanalyse.
La description de ces schmes dans les grammaires arabisantes est,
lorsquelle existe, bien plus sommaire que pour les verbes augments.
3
1
Je reprends ici, sous une forme largement revue, le chapitre 3 de mon cours de pr-
paration lagrgation darabe du Centre national denseignement distance (CNED),
2002 et 2003. Par souci dexplicitation, jai prsent quelque peu en dtail le cadre concep-
tuel ncessaire au traitement de cette difcile question, et conserv des dfnitions ou
des indications qui pourront paratre videntes, mais ne le sont pas ncessairement.
Ce travail doit beaucoup au cadre thorique et aux travaux de Jean-Pierre Descls ,
ainsi quaux sminaires que nous avons anims, et continuons danimer ensemble,
depuis 2001, sur les valeurs associes en contexte aux formes verbales en arabe et en
franais. Je remercie galement les membres de son quipe de recherche, notamment
Brahim Djioua, pour de nombreuses et fcondes discussions.
2
La convention qui prvaut dans la tradition linguistique arabe dsigne ce que nous
appelons les consonnes radicales 1, 2 et 3 (de la racine smitique) respectivement par
f, et l.
3
Ainsi pour les grammaires en langues occidentales : Neyreneuf et Al-Hakkak (1996,
3542) signalent des nuances de sens . . . pour chaque forme drive , mais non pour
les verbes simples (id., 2829). De mme, les descriptions consacres par Blachre et
314 joseph dichy
Il suft dobserver quils demeurent dsigns dans la tradition arabi-
sante occidentale comme la Forme I. Cette dsignation relve dune
ide gnrale qui rpond, dfaut de dcrire les donnes linguistiques,
une cohrence interne. La Forme I est appele verbe primitif par A.
Sylvestre de Sacy , par rfrence au verbe nu mujarrad des grammaires
arabes (1831, I : 123). A cette forme primitive, Sylvestre de Sacy associe
les grandes divisions smantiques gnrales des verbes (id. 121122) sur
lesquelles il fondera ensuite son analyse des formes drives (id. 129143).
Mais il nidentife pas de valeurs smantiques associes en propre aux
schmes faula , faila et faala , dans lesquelles il ne voit quune forme
primitive ou nue (mujarrad ) partir de laquelle les formes drives ou
augmentes (mazda) sont construites.
Si la tradition arabisante a pu ainsi rapporter ces trois schmes une
seule forme, cest sans doute aussi pour une autre raison : ces derniers
apparaissent en efet comme instables, tant du point de vue formel, o lon
observe des modifcations de la voyelle de la 2e radicale (section 2 ci-des-
sous), que du point de vue smantique, o les notions trs gnrales de
transitivit , de verbes daction ou dtat par lesquelles on a essay de les
caractriser rencontrent un nombre lev de contre-exemples. Cest ainsi
que les trois pages dindications essentiellement formelles consacres
cette question par la grammaire da-artn sont prcdes de cet aver-
tissement : On ne peut tablir [la forme des verbes relevant de] ces schmes
(awzn ) quen ayant recours des ouvrages lexicographiques (1910, 11).
Faut-il donc renoncer ? Il y a de solides raisons de penser que non. Plu-
sieurs essais de mise en ordre et danalyse ont t tents. La synthse la
plus rcente de ces tentatives est, ma connaissance, celle de Pierre Lar-
cher (2003, chap. II), qui comporte en outre des propositions origina-
les. Signalons galement les analyses incontournables dAndr Roman
(1983, II : 886900). Dautres auteurs, au premier rang desquels Marcel
Cohen , Paul Joon , Jean Cantineau et Henri Fleisch ont t utiliss. En
outre, un nombre considrable dobservations et danalyses a t efectu
dans les textes des sciences linguistiques arabes mdivales . Je ferai lar-
gement usage de ces travaux, sans toutefois les discuter en dtail, pour
Godefroy-Demombynes (1952, 4970) aux formes drives contrastent avec labsence
danalyse de la Forme I. Belot , (1922, 15), Badawi , Carter et Gully (2004, 60) ne signa-
lent pour la forme primitive ou de base (al-fl al-mujarrad ), de manire trs pru-
dente, que la transitivit ou son absence en fonction de la voyelle mdiane de faala ,
faila et faula . On trouve en revanche des propositions de description du sens dans
Caspari-Uricoecha (1881, 3233), Wright (189698 I : 30), Brockelmann (1948, 35) ou
Boormans (1967, 10), ainsi, naturellement, que dans la tradition linguistique arabe (voir
la synthse de Nr ad-Dn 2002, 177186).
fa
ula, fa
ila, fa
ala 315
viter dallonger indment cette tude. A lenttant problme pos par la
dispersion smantique des verbes de schme simple en arabe jai cherch
apporter une solution fonde sur une dfnition aussi rigoureuse que
possible des termes utiliss, ainsi que sur une conception, me semble-t-il,
renouvele de la relation entre sens et forme dans les schmes (au sens
que prend ce terme dans le domaine smitique) en arabe (Dichy 2003).
Les propositions qui suivent sappuient en outre sur un dpouillement
systmatique des donnes lexicales dans le roman de Tawfq al-H akm ,
Yawmiyyt nib f l-aryf (abrg en Yawm.) dune part, et dans plu-
sieurs dictionnaires arabes de lautre (notamment Hans Wehr-Cowan
1979 et al-Mujam al-Wst ). Jai fait largement usage de la base de don-
nes lexicale DIINAR .1 (Dictionnaire informatis de larabe , version
1 Dichy , Braham , Ghazeli et Hassoun 2002 ; Dichy et Hassoun 2005 ;
http ://www.elda.org), du concordancier informatis construit par
R. Abbs (2004), ainsi que de la liste indexe de 20.000 verbes rali-
se pour les besoins du Bescherelle des Verbes arabes (Ammar et Dichy
1999). Les rfrences ces dernires sources ne sont pas donnes, la
difrence des textes arabes mdivaux , des travaux arabisants de lpo-
que moderne ou, au besoin, des dictionnaires ; Yawm est indiqu avec
le numro de la page suivi de la ligne lorsque le sens doit tre interprt
en fonction du contexte.
2. Les schmes du verbe non augment , aspects formels
2.1 Lalternance vocalique de la voyelle de la 2
e
radicale
Les trois schmes de base du verbe simple ont t identifs dans
les sciences mdivales arabes du langage partir de la voyelle de la
deuxime radicale ou voyelle mdiane du verbe au paradigme du md
(ou sufx ),
4
respectivement u, i et a. Les trois familles morphologi-
ques correspondantes sont numres sous forme dexemples (amt ila)
4
Les termes daccompli et inaccompli , dachev et dinachev ou de perfectif et imper-
fectif , pourtant reus, comportent un problme, celui de dsigner la forme morpholo-
gique par un terme dnotant lune des valeurs aspectuo-temporelles que cette forme
peut prendre en contexte. Do le choix de dsigner ces paradigmes verbaux par leur
nom arabe ou par le trait morphologique qui les caractrise. Ainsi, le md (tradition-
nellement accompli), dans lequel le morphme de personne prcde la base du verbe,
correspond ici au sufx ; le mudri (traditionnellement inaccompli), dans lequel ce
morphme prcde la base, est dsign du terme de prfx (cf. Moscati , d. 1964, 1312 :
prfx-conjugation vs sufx-conjugation).
316 joseph dichy
par Sbawayhi (m. vers 180/796) ds lintroduction du Kitb (I : 12), et
reprises au cours de ce trait au moyen des conventions faala , faila et
faula (dsignes comme des formes construites , abniya ) avec des indi-
cations morphologiques , syntaxiques et smantiques (Kitb IV : 567,
notamment 38). Je laisse de ct ici la question, qui a t pose, de savoir
si la forme sujet non exprim (dite passive ) fuila ne constituerait pas
un schme part entire.
5
Selon la reconstruction du proto-systme de la langue arabe , telle
quelle a t propose en diachronie (et en tout cas dans lorganisation
qui sous-tend le systme de la langue), la forme simple du verbe com-
porte trois schmes, reprsents dans le tableau ci-dessous (voir notam-
ment Cantineau 1950, 77 ; Fleisch 1957 et 1968, 115119 ; Roman 1983,
II : 894) :
Md
.
( sufx ) Mudri
'
( prfx )
Schme simple 1 faula yaf ulu
Schme simple 2 faila yaf alu
Schme simple 3 faala yaf ilu
FIGURE 1 LES SCHMES POSTULS DU VERBE SIMPLE DANS
LE PROTO-SYSTME DE LARABE
Selon Moscati , d. (1964 : 122), qui se situe dans la perspective des tudes
smitiques compares, lantiquit de ce schma vocalique trois ter-
mes en arabe est confrm par quelques-unes des manifestations les plus
anciennes du groupe smitique du Nord-Ouest, cest--dire, par lamorite ,
lougaritique , et les gloses de Tell Amarna . . .
6
Dans cette reconstruction,
comme le souligne Roman (loc. cit.), les schmes sont toujours identifa-
5
Voir sur ce point Cantineau (1950) ; Roman (1983 II : 897900 ; 2005, 33), qui voit
dans la premire voyelle la marque de la diathse (subjective ou objective) ; Fleisch
(1957 ; 1968, 246 ; 1979) ; Larcher (1996 ; 2003, 2628). Nr ad-Dn (2002, 185186)
signale des traits tardifs, dans lesquels la forme fuila est considre comme un qua-
trime principe (as l), i.e., un quatrime schme du verbe simple , et non une forme de
la conjugaison (comme dans , m. en 643/1245, arh al-Mulk : 30).
6
Te antiquity of this threefold vocalic scheme in Arabic is confrmed by some of
the oldest manifestations in North-West Semitic, i.e. Amorite, Ugaritic, and the Tell
Amarna glosses. In the prefx-conjugation the variation in the second vowell is at least
partly paralleled : u or i corresponding to a, and a to i, while u generally remains (Mos-
cati , d. 1964, 122).
fa
ula, fa
ila, fa
ala 317
bles en fonction de la voyelle de la deuxime consonne radicale (en gras
dans la transcription), et ce, dans chacun des deux grands paradigmes
de la conjugaison . Dans ltat de langue arabe tel quil nous est parvenu,
ds les poques classiques, les trois schmes simples ci-dessus ne sont
plus directement identifables, comme le rappelle le tableau suivant :
Md ( sufx ) Mudri' (prfx )
Schme simple 1 faula yaf ulu
Schme simple 2
faila
yaf alu
Sous-catgorie formelle yaf ilu
Schme simple 3
faala
yaf ilu
Sous-catgorie formelle 1 yaf ulu
Sous-catgorie formelle 2 yaf alu
FIGURE 2 LES SCHMES DU VERBE SIMPLE ET LEURS SOUS-CAT-
GORIES FORMELLES, TELS QUILS NOUS SONT PARVENUS
Cette situation continue de permettre lidentifcation au md (suf-
fx ) de trois schmes distingus au moyen de la voyelle qui afecte la
2
e
radicale , mais non au mudri (prfx ). Lorsque lon traite daspects
formels, comme dans la conjugaison ou dans un dictionnaire , il est
donc ncessaire de caractriser le verbe par lalternance vocalique de la
voyelle mdiane aux deux paradigmes du sufx et du prfx , exemple :
daraba/yadribu frapper, dalternance vocalique a/i. (Jcrirai ici, selon
une convention reue : daraba i.)
2.2 Les zones de stabilit ou dinstabilit formelles de lalternance
vocalique
Comme le montre la fgure 2, le schme simple 1 (alternance u/u) est
formellement stable. Dans les schmes faila et faala en revanche, on
trouve des sous-catgories formelles dans lesquelles la voyelle de la 2
e
radicale varie. Ces variations sont souvent corrles la prsence, au
sein de la racine , dune cause de transformation (illa ) :
7
deuxime et
7
Sur le statut pistmologique de cette notion dans la tradition linguistique arabe
318 joseph dichy
troisime radicales identiques, consonne radicale /w/ ou /y/, ou encore
radicale // en premire, deuxime ou troisime position. Les relations
entre ces causes de transformation et les variations observes dans les
schmes du verbe simple ont t, sinon expliques, du moins largement
inventories dans les sciences linguistiques arabes mdivales (voir
rsum dans a-artn , 1910 : 1014 ; Qabbwa , 1998 : 8594).
2.2.1 La sous-catgorie formelle faila /yaf ilu du schme simple 2
La sous-catgorie formelle du schme simple 2 ( faila /yaf ilu) est corr-
le dans la plupart des cas une racine de premire radicale w ou y. Elle
ne concerne quun petit nombre de verbes . Trois dentre eux, repris de
Sbawayhi (Kitb IV : 3839) et rgulirement cits dans les traits lin-
guistiques arabes mdivaux ou par les grammaires, relvent de racines
sans cause de transformation : h asiba i ou a croire, estimer ; baisa i ou
a tre misrable ; naima i ou a tre doux, fn. Ils peuvent tre consid-
rs comme ngligeables, car ils supportent galement une ralisation a
de la voyelle mdiane du mudri prfx .
8
2.2.2 La sous-catgorie formelle faala /yaf alu du schme simple 3
Une seule modifcation de la voyelle mdiane est sufsamment rgu-
lire. Elle est largement signale dans les ouvrages arabes mdivaux
(cf. par exemple, al-Zajjj , Jumal, 396397). Il sagit de la sous-cat-
gorie formelle 2 du schme simple 3 ci-dessus ( faala /yaf alu) : elle se
produit pour simplifer si la 2
e
ou la 3
e
radicale du verbe est lune
des six consonnes darrire (h urf al-h alq ), i.e. appartient lensemble
{, h, h , , , x}. Cette rgle sapplique selon une gradation, de manire
hautement frquente pour les deux consonnes les plus antrieures (les
consonnes glottales et h), et nettement moins frquente pour les deux
mdivale, et notamment chez Ibn Jinn (m. en 393/1002), voir Guillaume 1984. La
liste et les combinaisons des causes de transformation (il peut y en voir plus dune dans
une racine donne) ont t prsents sous une forme modlise dans Dichy (1993) et
Ammar et Dichy (1999, 3134).
8
Voir par exemple Ibn Xlawayh , m. en 370/981, Laysa f kalm al-arab Point ne
se trouve dans le langage des Arabes . . . (sous-entendu hormis, lexception de), chap.
10 ; az-Zajjj , m. en 337/949, Jumal : 398 (plus cohrent que le prcdent). Sur le statut
de la classe h asiba/yah sibu, notamment chez Ibn Jinn , m. en 392/1002, cf. Guillaume
(1984, 427433).
fa
ula, fa
ila, fa
ala 319
consonnes les moins antrieures ( et x) (cf. Ibn Ya , arh al-Mufas s al
VII : 153154, repris in Ammar et Dichy 1999 : 24).
2.2.3 La sous-catgorie formelle faala /yaf ulu du schme simple 3
La sous-catgorie formelle 1 du schme simple 3 ( faala /yaf ulu) ne
parat pas, quant elle, pourvoir tre lobjet de corrlations systmati-
ques entre la voyelle de la 2
e
radicale et des traits phontiques afectant
les consonnes du verbe. On trouve ainsi un nombre non ngligeable de
verbes dont le prfx (mudri ) admet les deux voyelles mdianes i et
u, ainsi : azafa/yazifu et yazufu (an) se dtourner (de) ; qafala/yaqflu
et yaqfulu revenir (de voyage), rentrer, etc. La prsence, dans de tels
exemples, des deux ralisations met en dfaut toute corrlation entre les
traits phontiques des consonnes radicales et la voyelle mdiane dans la
langue arabe telle quelle nous est parvenue, avec ses normes stabilises
par les travaux des lexicographes et lautorit des grammairiens.
3. Aspects smantiques , concepts de base
Considrons maintenant les valeurs smantiques associables aux sch-
mes simples . Les deux notions qui ont t traditionnellement utilises
constituent un cadre conceptuel dont on verra quil ne peut sufre. Il
sagit du trait syntaxique de transitivit ( 3.1.1) et de la distinction entre
verbes daction et verbes dtat ( 3.1.2). Ces notions devront tre repri-
ses dans un cadre conceptuel largi (sections 3.2 3.4). Les descriptions
proposes ci-dessous sappuieront sur une srie dexemples, choisis pour
illustrer la plupart des types de valeurs smantiques qui caractrisent
ces schmes.
3.1 Les deux principales notions traditionnellement utilises pour
dcrire le sens des schmes simples
3.1.1 Le trait syntaxique de transitivit /intransitivit
La premire distinction que lon rencontre dans les grammaires arabisan-
tes comme dans les sciences mdivales arabes du langage est de nature
syntaxique. En fonction de la voyelle a, u ou i afectant la deuxime
consonne radicale , ces verbes sont dits :
320 joseph dichy
pour le schme faala , transitifs , bien que comportant un nombre
considrable dintransitifs ,
transitifs et intransitifs pour le schme faila , et
toujours intransitifs pour les verbes en faula .
9
Les traits syntaxiques de transitivit ou dintransitivit sont cruciaux
dans la thorie de la rection qui prvaut dans la tradition linguistique
arabe mdivale. Leur incidence sur le sens des verbes nest toutefois
pas directe : elle en est plutt une consquence quune cause. Une autre
question que la thorie de la rection laisse dans lombre est celle de la
transitivit indirecte, dans laquelle le complment du verbe est un syn-
tagme prpositionnel. La transitivit, envisage sous ses deux formes,
directe et indirecte sera naturellement incluse dans les valeurs smanti-
ques associes aux verbes des schmes simples , particulirement en ce
qui concerne le schme faala .
3.1.2 La partition traditionnelle entre verbes daction et verbes dtat
Les arabisants et les smitisants occidentaux ont gnralement fond
leurs analyses des verbes des schmes simples sur une partition gn-
rale en verbes daction , ou actifs , et verbes dtat ou statifs , les seconds
tant ncessairement intransitifs , et les premiers, transitifs ou intransi-
tifs selon les cas. Dans son principe, le schme faala aurait ainsi corres-
pondu des verbes daction (transitifs ou intransitifs ), le schme faula
des verbes statifs toujours intransitifs exprimant un tat permanent
ou une qualit inhrente naturelle , et le schme faila des verbes gn-
ralement intransitifs indiquant soit un tat ou une condition tempo-
raire, soit une qualit purement accidentelle dune personne ou dune
chose (Wright 189698 I : 3638). Dans cette description, le schme
faila pose un problme, qui a t trait en recourant la notion de verbe
moyen dE. Benveniste (section 3.4 ci-dessous).
Si la notion de verbe dtat , une fois redfnie, rsiste lanalyse, celle
de verbe daction ne laisse pas, dans lusage qui en a t fait, dapparatre
comme indifrencie. Que faire, par exemple de verbes comme z anna
u croire, saqat a u tomber, halaka i prir, wajaba i devoir, amala u
englober ou ra a voir, qui relvent tous du schme faala , rput cor-
9
Cf. Sbawayhi (Kitb IV : 38), ainsi par exemple que az-Zamaxar (Mufas s al, 277
279). Lintransitivit des verbes dtat est rappele notamment par Wright (189698,
vol. I, 3638) ou Fleisch (1957 ; 1979). Pour lhbreu , voir par ex. Gesenius -Kautzsch -
Cowley (1910, 4143) ou Joon (1923, 4041).
fa
ula, fa
ila, fa
ala 321
respondre aux verbes daction ? Et que dire des paires akala u manger,
de schme faala vs. ariba a boire, de schme faila , ou ra a voir, vs.
samia a entendre ? La rponse consistant poser une catgorie inter-
mdiaire ou moyenne (Marcel Cohen 1929 ; Joon 1930 ; Fleisch 1957),
qui concerne au premier chef le schme faila tout en incluant des verbes
en faala , ne rsout que trs partiellement le problme. Les processus de
glissement smantique entranant le passage dun verbe daction un
verbe dtat , sils ont donn lieu quelques fnes observations,
10
nappor-
tent pas non plus de solution de porte gnrale. Mais le principal pro-
blme, comme la section 4.3 lillustrera, est que lapparente transparence
de la notion de verbe daction aura masqu la complexit des donnes
et la forte dispersion smantique du schme faala . On verra dans la
suite de ce travail quil nest de fait pas ncessaire de recourir aux verbes
daction pour dcrire les verbes en faala , malgr lintrt que prsente
par ailleurs cette notion.
Passons maintenant la dfnition des proprits fondamentales
ncessaires lanalyse des schmes simples : lagentivit (section 3.2), les
notions de procs , dvnement , de processus et dtat (sect. 3.3), celle de
verbe moyen ou de diathse interne (sect. 3.4).
3.2 La proprit dagentivit
Lagentivit se dfnit comme la participation de lagent au procs dnot
par le verbe, le critre de cette participation tant le contrle (Lyons
1978/90, 3.4 ; Descls 1994) de lagent sur ce procs. Il y a en arabe trois
degrs dagentivit :
11
lagentivit pleine dans laquelle lagent contrle
efectivement le procs ; partielle , dans laquelle il nexerce quun con-
trle imparfait , et la non agentivit , dans laquelle il ne jouit daucun con-
trle (lagent existe, mais il ne concide nullement avec le sujet gramma-
tical ). Ces trois degrs sopposent eux-mmes lagentivit neutralise ,
qui correspond des procs dans lesquels il ny a aucun agent , ce qui se
10
Cf. par exemple pour lhbreu , Joon 1923, 41, pour larabe , Larcher 2003,
2728.
11
Cf. Roman (1990, 4243 ; 1999/2005, B-2.3.2). La dfnition ci-dessus de lagenti-
vit est en partie difrente. Cf. galement lanalyse de Fleisch (1968, 116117), qui pose
une bipartition du verbe arabe : 1 le verbe agent (le sujet tant considr en tant
quagent) ; le verbe de qualit (le sujet tant simplement le qualif), et distingue, dans la
premire catgorie lagent pur et simple de lagent intress.
322 joseph dichy
prsente typiquement dans les verbes valeur dtat caractristique (
4.1.1 ci-dessous). Exemples (verbes de schme simple) :
12
(a) verbes dagentivit neutralise (valeur smantique tre x) ou non-
agentifs (valeur : devenir x) :
schme faula : karuma u, tre, devenir karm noble-et-gnreux ;
arufa u devenir, puis tre arf , i.e. noble, de premier rang social ;
s aura u devenir, tre s ar, petit ou : tre trs jeune ; lat ufa u tre,
devenir lat f, subtil, fn, gracieux ;
schme faila : baliha a devenir ou tre ablah, faible desprit, stu-
pide ; s alia a devenir ou tre chauve (as la) ; karia a tre ou devenir
rid, craquel (notamment : peau) ;
schme faala : xadaba i devenir, tre vert (plante, herbage, lieu . . .) ;
xafa i devenir, tre xaff, lger (antonyme de schme faula : t aqula
u devenir, tre t aql, lourd) ;
(b) verbes non-agentifs (le sujet grammatical nest pas lagent du procs
dont il est le sige) :
schme faila : dasiqa a se remplir ( ras bords), dborder (bas-
sin . . .) ; marida a tomber malade ; qarima a avoir faim de viande
(par efet de privation) ;
schme faala : sakana u, se calmer (mer, vent, bruit . . .) ; waqaa a
tomber et galement : se trouver (ville, montagne, lieu . . .) ; halaka i
prir ;
(c) verbes partiellement agentifs :
schme faila : adiba a tre, se mettre en colre ; h azina a tre ou
devenir triste, prouver de la tristesse ; wamiqa i chrir ; aiqa a
tre ou tomber passionnment amoureux ; samia a entendre (sens
de loue) ;
schme faala : ra a voir (vue) ; amma u sentir (odorat) ;
(d) verbes pleinement agentifs :
schme faila : ariba a boire ; rakiba a chevaucher, monter (pour se
dplacer) ; h amida a louer (louange),
12
Certains des exemples donns dans ce travail existent aussi avec des formes corres-
pondant dautres schmes du verbe simple , quelquefois avec le mme sens . Ces autres
formes ne sont pas reprises ici, malgr lintrt quet reprsent la discussion.
fa
ula, fa
ila, fa
ala 323
schme faala : daraba i frapper ; jalasa i sasseoir ; da u revenir ;
d ahaba a partir, aller ; t abaxa u (ou a) faire cuire ou bouillir, cui-
siner ; t alaa a monter ; akala u manger ; qla u dire.
La sous-catgorie (a) inclut des verbes des trois schmes et les sous-cat-
gories (b) (d), des verbes des schmes faila et faala . Ces exemples, qui
anticipent par leur classement sur la suite de lexpos, montrent que la
proprit dagentivit , malgr limportance qui est la sienne dans la des-
cription smantique des schmes simples , ne peut, elle seule, consti-
tuer le principe qui en permettrait le classement.
3.2.1 Agent , patient , sujet grammatical
Il est essentiel, pour bien comprendre la notion dagentivit , de distin-
guer les termes dagent (et corollairement, de patient ) du sujet gramma-
tical du verbe. Les formes verbales de larabe incluent en efet toujours
leur pronom sujet, qui correspond aux morphmes de personne sufxs
au md et prfxs au mudri , et concide avec le sujet grammatical du
verbe. Ainsi : jat l-amra correspond mot--mot : elle est venue la
princesse (Ammar et Dichy 1999, 1214). Cette structure est occulte
par lemploi indifrenci, dans la tradition linguistique arabe , de fil
pour lagent comme pour le sujet grammatical .
Lagent ( fil ) est ce qui efectue le procs dcrit par le verbe. Dans les
verbes dcrivant une action , lagent est lauteur de celle-ci. Dans les ver-
bes de perception ou de sentiment , par exemple, il est le sige de celui-ci.
Dans les verbes dcrivant un tat , le procs , comme on vient de le voir,
ne rfre pas un agent : le sujet grammatical inclus dans la forme verbale
est le sige de cet tat , quil sagisse dune personne comme dans raufa u
tre, devenir ou se montrer doux, compatissant ou dune chose, comme
dans daula u, devenir/tre troit ou petit (quantit). Dans les verbes
dagentivit partielle ou entire , lagent concide avec le sujet grammati-
cal . Ce dernier, la voix dite passive , concide avec le patient (qui subit
le procs dcrit par le verbe), exemple : suiltu jai t questionn. Dans
les formes non-agentives , le sujet grammatical subit galement le pro-
cs ou en est le sige, exemples :
saqatat al l-lard elle est tombe par terre (quil sagisse dune per-
sonne ou dune chose) ;
darasa r-rasmu la trace (du campement) sest eface : lagent est ici le
vent du dsert. (Ce thme est classique dans la posie arabe ancienne ;
324 joseph dichy
cf. la fn du 6
e
vers de la Muallaqa grande ode du pote emblma-
tique de lant-islam Imru al-Qays : Wa-hal al rasmin drisin min
muawwal, Sur traces efaces, qui sen irait pleurant ?)
3.3 Les proprits dvnement , de processus et dtat ( partir de
Descls 1994)
Le terme gnral de procs (auquel correspond en anglais celui de
situation)
13
traduit le fait que le sens du verbe est inscrit dans une dure ,
cest--dire, dans un intervalle de temps. Cette dure peut tre perue de
trois manires, correspondant aux notions aspectuelles dtat , dvne-
ment ou de processus . La dfnition de ces termes premiers est emprunte
directement J.-P. Descls.
14
premire vue, ces dfnitions pourraient
apparatre comme dtournes : les trois termes aspectuels dvnement ,
de processus ou dtat , qui dcrivent des phnomnes afectant des ver-
bes en en contexte sont rutiliss, dans ce travail comme dans ceux dont
la publication suivra, pour dsigner dans la smantique des schmes de
larabe des primitives incluses dans le sens grammatical associ aux sch-
mes simples et certains schmes augments . Mais y regarder de plus
prs, sagissant de valeurs smantiques primitives, il ny a rien dtonnant
ce quelles puissent galement tre observes dans les proprits lexi-
cales de familles morphologiques , telles que les schmes simples et les
schmes augments. La mise en lumire de ces observations constitue
une premire contribution thorique du travail prsent ici la dfni-
tion de ces valeurs primitives. Leur confrontation aux donnes de larabe
ma en outre conduit en proposer des extensions, particulirement en
ce qui concerne les sous-catgories de ltat ( 3.3.3).
13
Le terme de procs ne se traduit pas directement en anglais. Comrie (1985, 5),
Lyons (1978/1990, 3.4) ont recours au terme de situation pour recouvrir les tats dune
part, les vnements et les processus de lautre.
14
Mais voir aussi Comrie (1976, 1985), Lyons (1978/1990), ou, difremment, la syn-
thse de Maingueneau (1994). La thorie des invariants cognitifs du langage et des pri-
mitives smantico-cognitives de J.-P. Descls est sous-jacente aux analyses prsentes
ici, notamment : lide que les primitives sont des oprateurs abstraits dtermins par
des proprits formelles, lhypothse selon laquelle elles sont ancres sur la perception
et laction, ainsi quun nombre important de points qui dcoulent de ces conceptions
(Descls 1990 ; 1994 ; Descls et Guentchva paratre). Le lecteur familier de ces tra-
vaux notera enfn que le terme de schme est employ ici la manire des travaux
arabisants et smitisants. On distinguera donc cet usage de celui des schmes sman-
tico-cognitifs (Descls , 1990 ; Abraham , 1995 ; Descls et al., 1998).
fa
ula, fa
ila, fa
ala 325
3.3.1 vnement
Laspect vnement [. . .] est lexpression dune occurrence qui apparat
sur un arrire-fond stable [. . . et] tablit une coupure entre lavant v-
nementiel (pas encore) et laprs vnementiel (ne plus). La dure de
lvnement est donc borne gauche par un avant et droite par un
aprs. Cet aprs peut concider avec un tat engendr par lvnement ,
ou tat rsultatif (Descls 1994, 7375).
15
Les exemples ci-dessous (verbe
concern en gras) illustrent quelques proprits fondamentales de las-
pect vnement :
m kidtu hd s -s abh a arifu fnjna l-qahwa al maktab h att
waradat ira tilifuniyya . . . (Yawm., 123), peine, ce matin-l,
avais-je sirot le caf [pos] sur mon bureau, quun tlgramme tl-
phon survint . . . (mot--mot, lorsque apparut). La premire pro-
position de cette phrase (m kidtu . . .arifu peine . . . avais-je sirot)
constitue larrire-plan sur lequel se dtache lvnement h att wara-
dat lorsque apparut, vnement qui dtermine au sein de la situation
dcrite, un avant et un aprs.
wasala dayfun mund u sa notre invit est arriv depuis une heure :
de lvnement est arriv rsulte un tat , dont la dure englobe le
moment de lnonciation. Le locuteur rapporte lvnement de larri-
ve de linvit, et exprime par les mmes mots ltat dans lequel son
interlocuteur et lui se trouvent, et qui correspond laprs du procs :
leur invit est l depuis une heure. On dira, en simplifant, que lv-
nement engendre ltat qui rsulte de lui.
ariqa r-rajulu lhomme sest noy (ou se noya) : lvnement dcrit
par le procs engendre ltat rsultatif noy, exprim en arabe par
ladjectif arq, associ ce verbe.
3.3.2 Processus
Laspect processus exprime un changement saisi dans son volution
interne. Tout processus exprime ncessairement un changement initial
(. . .) qui indique le dbut du processus. Le processus peut tre dcrit
15
Je fais dans ce travail un usage gnrique (ou hypronymique) du terme dtat rsul-
tatif . Ce dernier inclut ici ltat rsultant, qui est par ailleurs trs utile pour lanalyse des
valeurs aspectuo-temporelles en contexte, par exemple pour la description de ltat-pr-
sent , concomitant lacte dnonciation (Descls , 1994, 73). Ce choix, limit au prsent
travail, est d au fait que les valeurs smantiques associes aux familles morphologiques
constitues par les schmes ne sont pas sensibles cette distinction contextuelle.
326 joseph dichy
comme une succession dtats . Il peut, selon le cas, sorienter ou non vers
un terme explicite, qui correspond alors au dernier tat du processus ou
tat rsultatif (Descls 1994, 7580). Exemples :
Dans la phrase ci-dessus, le syntagme m kidtu arifu fnjna l-qahwa
peine avais-je sirot le caf exprime un processus dont le terme est
spcif : un moment donn, qui correspond ltat fnal du proces-
sus , le breuvage est entirement bu. La phrase dit mot--mot : . . . sirot
la tasse de caf ; la valeur smantique associe cette expression dans
le lexique de larabe implique que le contenu de la tasse a t bu en
entier.
16
Dans arabati -amsu le soleil sest couch ou hal turdu an
tujanninan ? tu veux me rendre fou (folle) ? cest galement dans
la valeur lexicale du verbe que le terme du processus (le coucher du
soleil, la folie) est inscrit.
Le sens de amiya a perdre la vue comporte un terme, qui corres-
pond dans le lexique ladjectif am aveugle. Ce verbe dcrit, selon
le contexte, soit un vnement , soit un processus .
3.3.3 tat, types dtats et catgories drives ou corrles
Laspect tat exprime la stabilit de la situation rfrentielle reprsente,
cest--dire quil rfre une absence de mouvement et de changement.
Ltat dcrit un procs qui nest born ni gauche (en son commence-
ment) ni droite (le terme nest pas spcif), et ne fait rfrence aucun
agent (Descls 1994, 7173).
17
Il y a plusieurs types dtats . Deux dentre
eux, ltat caractristique ( 3.3.3.a) et ltat rsultatif ou tat acquis
( 3.3.3.b) intressent directement les schmes du verbe simple . Le second,
qui est produit par un changement dtat , correspond un glissement
smantique troitement corrl lusage des verbes dtat en discours.
Lanalyse ci-dessous dbouche sur ladjonction, la valeur smantique
de ltat caractristique , de celle de lacquisition dtat . Cette dernire
16
Ainsi, le dictionnaire glose lexpression raifa l-ina, mot--mot siroter un rci-
pient comme en aspirer le contenu jusquau bout (al-Mujam al-Wast : racine /r--f/) ;
de mme, ariba l-ina boire un rcipient correspond ariba kulla m fhi en boire
tout le contenu (Ilys et Ns f 1995, 164).
17
Comparer avec la dfnition de Maingueneau (1994, 64) : Les tats . . . nont ni
dbut, ni fn, ni milieu, ils ne supposent ni agent ni changement (ex. Luc est paresseux).
la difrence de Comrie (1976, 1985), Lyons (1978/1990), Descls (1994) et du prsent
travail, cet auteur noppose pas tat , vnement et processus .
fa
ula, fa
ila, fa
ala 327
notion relve de celles, plus gnrales, dtat rsultatif et de changement
dtat . Lacquisition dtat constitue en efet la principale valeur smanti-
que corrle ltat caractristique ( 3.3.4).
3.3.3.a La notion dtat caractristique
Les donnes de larabe mont conduit proposer une sous-catgorie de
la notion gnrale dtat , celle dtat caractristique (Dichy 2002/2003).
Ce dernier est exprim par un verbe associ dans le lexique un adjectif ,
traditionnellement appel adjectif assimil ou analogue [au participe]
(s ifa muabbaha ci-aprs 4.1.2). Ltat caractristique par oppo-
sition ltat non caractristique , dcrit une proprit ou un attribut
de lobjet , pos, selon le point de vue, comme dominant, intrinsque,
inhrent ou inscrit dans la nature de celui-ci. Cet tat peut, comme on
le verra au paragraphe suivant, tre acquis lissue dun vnement ou
dun processus .
Le terme dtat caractristique , dont le contenu sera confront aux don-
nes du schme faula la section 4.1, est prfrable lemploi habituel
des grammaires arabisantes qui oppose, dans les verbes du schme sim-
ple , qualit permanente ou durable qualit temporaire (par exemple,
Wright 189698, I : 3638). La dure relative dun tat tout comme
celle dun processus ou dun vnement correspond en efet lune
des valeurs aspectuo-temporelles attribues une forme verbale donne
en fonction du contexte, et ne doit pas tre confondue avec une valeur
smantique morpho-lexicale , associe comme telle un schme donn.
Ainsi, h azina a tre ou devenir triste (de schme faila , adjectif cor-
respondant : h azn) est traditionnellement cit comme exemple de
qualit temporaire ou transitoire .
18
Il rfre toutefois une qualit
durable dans les noncs suivants : fa-h azina il xiri umrihi Et il
fut triste jusqu la fn de ses jours ; inna llad na man . . . l xawfa
alayhim wa-l hum yah zann Ceux qui ont eu foi . . . aucune crainte
18
Ce verbe est cit par Caspari-Uricoecha pour illustrer un tat transitoire et passa-
ger ou bien une proprit ou une qualit qui nafecte lobjet du verbe que pendant peu
de temps (1881, 32) ; par Wright , comme illustrant a temporary state or condition,
or a merely accidental quality in persons or things (189698 I : 30) ; par Brockelmann
(1948, 35) comme exemple de verbes exprimant toujours une qualit ou une situation
accidentelle ou temporaire (Die Form faila steht durchweg fr zufllige, vorberge-
hende . . . Eigenschafen und Zustnde) ; ou Boormans (1967, 10), comme dcrivant un
tat passager .
328 joseph dichy
[ avoir] pour eux, il ne connatront [ternellement] aucune tristesse
mot--mot : et aucunement ne sattristeront (Coran 2 : 62). H azn sera
donc mieux dcrit comme un adjectif dtat non caractristique .
Inversement, le verbe xalua u mriter le nom de dprav, tre un
dprav (adjectif correspondant : xal), qui relve du schme faula,
rput renvoyer des verbes de qualit permanente ou durable
est glos dans le dictionnaire par lexpression taraka l-h ay wa-ra-
kiba hawhu, abandonner [toute] honte et se livrer ses passions
(al-Mujam al-Wast : racine /x-l-/). La qualit de xal dprav
(notoire) nest donc pas permanente. Xal peut en revanche tre
dcrit indpendamment toute rfrence une dure comme un
adjectif dtat caractristique.
Exemples de ces deux sous-catgories de la notion dtat :
verbes associs un adjectif dtat caractristique :
t la u tre t awl long ; hayifa a tre ahyaf , avoir la taille lance ;
awira a tre awar, borgne ; baxula u tre baxl, avaricieux (au
sens de tre avare par nature)
19
verbes associs un adjectif dtat non caractristique :
farih a a tre farih , joyeux (lun des contraires de h azina ci-dessus) ;
yaisa a tre yis dsespr ; at ia a tre at n ayant soif , avoir
soif ; marida a tre mard malade. . .
3.3.3.b Verbes dtat vs verbes de changement dtat produisant un
tat rsultatif
Pour H. Fleisch , cependant :
Le verbe de qualit nest pas un statif . Il signife acqurir une qualit , autre-
ment dit : devenir tel (daprs la qualit en question) : karuma = devenir
19
Les proprits de lagentivit ne relvent pas dinvariants linguistiques, mais de la
reprsentation du procs vhicule par le vocabulaire de chaque culture. Ainsi, dans la
culture arabe mdivale, on peut, soit (a) se montrer avare envers quelquun dans une
situation donne, ou (b) tre avare par nature ( f t -t ab) : cf. par exemple la prface du
Kitb al-Buxal Livre des avares dal-Jh iz , m. en 255/868 ; semblablement lavarice est,
lpoque de Molire considre comme pouvant tre inscrite dans le caractre de celui
qui, comme Harpagon, est un avaricieux. Le premier cas (a) est exprim au moyen du
verbe pleinement agentif baxila a al se montrer avare envers qqun, et le second (b),
par le verbe dagentivit neutralise baxula u tre un avaricieux (lavare est le sige du
vice inscrit dans sa nature).
fa
ula, fa
ila, fa
ala 329
karm (gnreux) ; ou bien, comme consquence de lacquisition : avoir
une qualit, soit : tre tel, cest un rsultatif , alors karuma = tre karm
(gnreux) (1968, 117).
Fleisch suit vraisemblablement ici la dfnition que donne al-Mubarrid
de la valeur smantique des verbes en faula :
innam huwa li-l-h l allat yantaqil ilayh l-fil . . . nah wu karuma wa-
z arufa wa-arufa elle (cette classe de verbes) dsigne uniquement ltat
dans lequel entre le sujet [du verbe] (mot--mot : ltat vers lequel se
dplace le sujet ), exemples : karuma, z arufa et arufa (Muqtadab I : 209
traduction des exemples : devenir et donc tre noble-et-gnreux, deve-
nir, puis tre rafn et devenir, puis tre prminent, illustre, noble).
Ces deux descriptions font de ltat caractristique une valeur seconde
par rapport la valeur devenir qqch. Or Sbawayhi dcrit les verbes de
schme faula comme dnotant les caractres qui peuvent tre dans les
choses (al-xis l allat taknu f l-ay Kitb IV : 28). Ces termes ont
t par la suite largement repris dans les sciences linguistiques arabes
mdivales (cf. par ex. az-Zamaxar , Mufas s al : 279). Dans cette concep-
tion, la valeur fondamentale des verbes de qualit est stative, alors que
chez al-Mubarrid et Fleisch , elle correspond au changement dtat . Mais
faut-il vraiment opposer ces conceptions ? Est-il possible de dcider de
la valeur smantique qui serait premire par rapport lautre ?
Considrons les exemples suivants :
H asuna u tre h asan, beau, qaduma u tre qadm, ancien ou hayifa
a tre ahyaf , qui a la taille lance dsignent la base un tat carac-
tristique .
amasa i ou u devenir/tre ensoleill (jour) ; asaqa i devenir/tre
noir, obscur (nuit) rfrent fondamentalement un changement
dtat . Il sagit de verbes signifant devenir qqch, dcrivant un proces-
sus qui aboutit un tat rsultatif non caractristique .
adiba a se mettre/tre en colre ou ja u avoir faim/tre afam (=
ayant faim) correspondent fondamentalement des verbes de chan-
gement dtat , dans lesquels le nouvel tat (non caractristique) qui
afecte le sujet grammatical rsulte du procs . Les adjectifs corres-
pondants, adib, adbn en colre et ji, jawn ayant faim sont des
tats rsultatifs non caractristiques .
Le verbe awira a devenir/tre borgne dsigne selon le contexte
soit un changement dtat (devenir qqch) dbouchant sur ltat carac-
tristique tre awar, borgne, soit cet tat caractristique lui-mme
(qui peut tre rsultatif ou non).
330 joseph dichy
Un verbe dtat caractristique correspondant fondamentalement
un tat, comme h asuna u tre beau ou hayifa a avoir la taille lance
ci-dessus peut aussi tre amen par le contexte dsigner le rsultat
dun vnement ou dun processus , exemple : hayifa l-na l-fat le
jeune homme est maintenant ahyaf , a maintenant une taille lance
(lnonc prsuppose : maintenant quil a grandi).
Ces exemples montrent qu la question pose (ce qui, de la valeur
smantique stative ou du changement dtat serait premier par rapport
lautre), il ny a pas de rponse gnrale, qui demeurerait valable dans
tous les cas. La rponse doit au contraire tre propose pour chaque
verbe , en fonction de son sens lexical (et des connaissances encyclop-
diques associes celui-ci). Cest ainsi que :
amasa i ou u devenir/tre ensoleill ou ja u avoir faim dcrivent
dabord un changement dtat (devenir qqch), et en second lieu, ltat
non caractristique correspondant.
Le sens des verbes dtat caractristique qaduma u tre ancien ou
jayida a avoir le cou long (signe de beaut) est peu compatible avec
un contexte dans lequel ces verbes prendraient la valeur devenir tel
(bien quun tel contexte demeure, comme on vient de lindiquer, pos-
sible).
Awira a, qui dcrit un tat caractristique , est compatible tant avec la
valeur stative tre borgne quavec celle du changement dtat (deve-
nir tel), sans quil soit possible de dcider de la valeur qui driverait
de lautre.
3.3.4. Lvnement de lacquisition dtat , cas particulier du changement
dtat et de ltat rsultatif
Mais il convient, pour comprendre la prsence dans ces verbes des deux
valeurs stative et dacquisition dtat , dinterroger plus avant la notion
dtat rsultatif . Ce dernier est engendr par un procs (vnement ou
processus ) correspondant un changement dtat (Descls 1994, 71 et
suiv. ; Descls et Guentcheva paratre, chap. II). Dans les donnes sui-
vantes, cet tat met en jeu des notions dans lesquelles les proprits de
ltat, de lvnement et du processus se trouvent combines :
(a) tat rsultatif (caractristique ou non) engendr par un vnement :
Le verbe amura u galement ralis amira a est glos dans les
dictionnaires comme devenir mir, notamment par succession (al
fa
ula, fa
ila, fa
ala 331
Mujam al-Wast , al-Munjid , Hans Wehr-Cowan 1979). Ce verbe
dcrit, dans cette acception, lvnement dun changement dtat , et
le nom-adjectif associ amr, qui dsigne ltat caractristique mir,
prince, celui qui gouverne un mirat, correspond un tat rsultatif .
Mais il peut galement, comme nous le verrons au paragraphe sui-
vant, dsigner ltat caractristique correspondant, indpendamment
dun vnement ou dun processus qui laurait prcd.
H azina a tre ou devenir triste, sattrister dsigne ltat non caract-
ristique h azn triste ou lentre dans cet tat. Lvnement que consti-
tue cette entre peut tre caus par un autre vnement, par exemple
larrive dune mauvaise nouvelle : lamm samia xabara axhi, h azina
adda l-h uzni lorsquil entendit les nouvelles de son frre, il tomba
dans une profonde tristesse (mot--mot : il sattrista dune intense
tristesse). Ladjectif h azn, dcrit ci-dessus ( 3.3.3.a) comme un
adjectif dtat non caractristique peut donc en outre correspondre
un tat rsultatif , h azina relevant, dans ce dernier cas, des verbes de
changement dtat .
(b) tat rsultatif (caractristique ou non) correspondant au dernier tat
dun processus :
Comme rappel ci-dessus, un processus correspond une succession
dtats . Ltat rsultatif peut donc concider avec le dernier tat du pro-
cessus qui le produit. Exemples :
az uma anuhu ayan fa-ayan sa situation sociale est devenue peu
peu (ou devint peu peu) de grande importance (az m) ;
sawida a devenir ou tre aswad noir ; saura u devenir ou tre
(socialement ou quantitativement) s ar petit ;
xadira z-zaru la rcolte a verdi ou verdit (Munjid , racine /x-d-r/ ;
Ilys et Ns f 1995, 113).
Ces donnes montrent que la notion dtat rsultatif est loin daller de
soi. Elles appellent deux remarques gnrales :
3.3.4.a Valeur aspectuo-temporelle du sufx (md ) et vnement
engendrant un tat exprim ou non par un adjectif assimil
On connat en franais la valeur daccompli du prsent que peut pren-
dre le pass compos et qui indique dans une nonciation qui inclut
la fois le pass compos et le prsent de lnonciateur le rsultat
prsent dun procs accompli antrieurement (Maingueneau 1994,
332 joseph dichy
6768). Descls (1994, 73) dcrit ce rsultat comme un tat-prsent
concomitant lacte dnonciation. Comparons par exemple :
(1) Alors tu te dpches ? Je fnis !
(2) Alors tu te dpches ? Jai fni ! (Leeman-Bouix 1994, 6162).
En (1), linterlocuteur est en train dachever quelque chose, le verbe
conjugu au prsent a donc valeur dinaccompli ; en (2), le pass com-
pos exprime un processus la fois accompli et achev (voir, pour cette
distinction, Descls 1994, 7680).
En arabe, la conjugaison au md (sufx ) comporte galement la pro-
prit dengendrer un tat rsultatif , exemples :
saimtu taklifa l-h ayti . . . je suis las des fardeaux [dont nous charge]
la vie (Muallaqa grande ode du pote prislamique Zuhayr b. Ab
Sulm , 46
e
vers) : la forme conjugue au sufx (mot--mot : jai t
ou je fus las . . .) engendre ltat rsultatif tre las.
La traduction franaise : je suis las est contrainte dexprimer cet tat-
prsent (au sens de Descls 1994, 73), mais elle ne peut rendre par une
forme verbale lvnement qui lengendre. En arabe le md (sufx ) de
saimtu englobe ces deux interprtations (cf. lexemple de was ala au
3.3.1) ;
at itu jai soif : la forme au sufx (mot-a-mot : jeus ou jai eu soif )
engendre ltat avoir soif (en traduction : jai soif mme commen-
taire que ci-dessus), qui pourrait tre glos comme : la soif sest main-
tenant installe en moi (ou sinstalla dsormais en moi).
Il est essentiel, pour comprendre le sens des schmes associs la notion
dtat , dobserver que ces exemples illustrent la prsence en arabe de deux
catgories difrentes de verbes , selon que ltat rsultatif est exprim ou
non par un adjectif associ au verbe par la drivation morpho-lexicale.
Ltat rsultatif engendr par saimtu nest pas lexicalis : le cas est donc
comparable mutatis mutandis lusage du pass-simple franais.
Pour at itu, ltat qui rsulte du procs correspond une entre lexicale
(i.e. un adjectif assimil ). Ainsi :
fa
ula, fa
ila, fa
ala 333
En reprenant les exemples (a) du 3.3.4, qui dcrivent des vnements :
Lnonc amura -iru (le pote est devenu/devint le prince
[rgnant] ou le pote rgne/rgna) peut tre glos comme : la posi-
tion de prince est maintenant (ou fut dsormais) occupe par le pote.
Le sens du verbe amura u est donc bien devenir, puis tre mir, et
non uniquement devenir mir, comme dans les dictionnaires cits
plus haut.
lamm samia xabara axhi, h azina adda l-h uzni lorsquil entendit
les nouvelles de son frre, il tomba dans une profonde tristesse cor-
respond : lorsquil entendit . . . ltat de tristesse sinstalla en lui.
Les exemples donns en (b), qui portent sur des processus , mritent un
mot dexplication : lachvement du processus parvenu son dernier tat
constitue lui-mme un vnement (Descls , 1994, 72, 7780) :
Dans xadira z-zaru la rcolte a verdi ou verdit cet vnement cor-
respond au moment o la rcolte parvient au degr de mrissement
attendu (ltat de verdeur).
Dans az uma anuhu ayan fa-ayan sa situation sociale est deve-
nue/devint peu peu de grande importance, cet vnement est celui
de lachvement du processus devenir peu peu, qui atteint alors son
dernier tat (celui de grande importance).
Dans tous ces cas, (1) il y a un avant et un aprs de lvnement du
changement dtat et de linstallation du nouvel tat , et (2) cet tat est
dsign dans le lexique par un adjectif assimil (s ifa muabbaha ). Lv-
nement pourrait de ce fait tre glos comme atteindre ou acqurir ltat
x, dans lequel x correspond un adjectif ou un nom-adjectif : at n
qui a soif ; amr mir ; h azn triste ; axdar vert ; az m de grande
importance.
3.3.4.b Les relations asymtriques tre x o devenir x et devenir x
tre x
La prsence dans le lexique dun adjectif assimil permet de mieux sai-
sir le lien entre tre x (valeur stative ) et devenir x (valeur dacquisition
dtat ) qui stablit entre les verbes associs un adjectif dtat caract-
ristique ou non caractristique dune part, et les verbes de changement
dtat de lautre. Ainsi :
334 joseph dichy
qadumat al-yawma dawlatuhum leur dynastie est devenue aujourdhui
ancienne peut tre glos comme : leur dynastie est devenue tant
ancienne (ou devint tant ancienne). Il sagit dun exemple de verbe
dtat caractristique (qaduma u tre ancien) dans lequel on observe
un glissement smantique en contexte de la valeur stative (tre qadm
ancien) celle dacquisition dtat (tre devenu tant qadm). Cest la
rencontre au sein de lnonc du sens lexical tre ancien et de la valeur
aspectuo-temporelle dachvement associe au md (sufx ) elle-
mme renforce ici par le marqueur temporel al-yawma aujourdhui,
qui permet le glissement smantique tre x o devenir x observ.
adiba r-rajulu lhomme est (ou fut) en colre (adib) correspond
: ltat tre en colre sest accompli en lui. Cet exemple est celui dun
verbe dacquisition dtat non caractristique , dans lequel on ren-
contre un passage de la valeur de changement et dacquisition dtat
(devenir en colre) la valeur dtat rsultatif acquis (tre en colre),
la seconde tant engendre par la premire. Il peut tre glos comme :
tre devenu tant adib (en colre). Cest galement la rencontre en
contexte du sens lexical tre en colre et de la valeur dachvement
que prend le md (sufx ) qui permet lengendrement de ltat acquis
tre x par la valeur de changement et dacquisition dtat devenir x.
Il y a donc deux cas gnraux de modifcation smantique corrls aux
verbes dtat . Les verbes dont la valeur de base est celle de ltat carac-
tristique connaissent un glissement smantique tre x o devenir x
(glissement de ltat au changement et lacquisition dtat ). Dans les
verbes exprimant fondamentalement un changement correspondant
lacquisition dun tat (caractristique ou non), ltat acquis est engendr
par lvnement qui concide avec lachvement du processus dacquisi-
tion, ce qui peut tre reprsent schmatiquement comme : devenir x
tre x (dans lequel le symbole se lit engendre).
3.3.5 Les deux valeurs smantiques associes la notion dtat : ltat
caractristique et lacquisition dtat
Lanalyse ci-dessus conduit donc distinguer, pour les verbes associs
un adjectif dtat caractristique , dune part ltat caractristique (tre x)
et de lautre lacquisition dtat (devenir x), qui correspond un change-
ment dtat . Il leur est associ, respectivement, des adjectifs dtat carac-
tristique et des adjectifs dtat acquis .
fa
ula, fa
ila, fa
ala 335
Lexamen des donnes relatives aux verbes associs un adjectif dtat
non caractristique permet en outre dobserver que ces verbes relvent
tous du changement dtat : ils appartiennent de ce fait la catgorie des
verbes dacquisition dtat . A condition de tenir compte des glissements
smantiques dcrits plus haut (tre x o devenir x et devenir x
tre x), lanalyse de Fleisch et la conception dal-Mubarrid cites plus
haut ( 3.3.3.b), qui ne sapplique que partiellement aux verbes associs
un adjectif dtat caractristique , sapplique en revanche pleinement
aux verbes associs un adjectif dtat non caractristique , que la tra-
dition arabisante appelait tat temporaire . On en a examin ci-dessus
plusieurs exemples.
3.4 Benveniste revisit : la notion de verbe moyen (ou diathse interne ) ;
le mdio-passif
Lautre grande distinction sur la dfnition de laquelle il convient de
revenir est celle de verbe moyen . On appelle verbe moyen selon la df-
nition dE. Benveniste , un verbe qui dcrit un procs dont le sujet est
le sige :
20
Ici, le sujet est le lieu du procs , mme si ce procs, comme cest le cas
pour lat. fuor [jouir, avoir proft de] ou sanskr. manyate [prouver une
agitation mentale], demande un objet ; le sujet est centre en mme temps
quacteur du procs ; il accomplit quelque chose qui saccomplit en lui, na-
tre, dormir, gsir, imaginer, crotre, etc. Il est bien intrieur au procs . . .
(1950/1966, 172).
3.4.1 Ambigut de cette dfnition ; les verbes mdio-passifs
La formulation ci-dessus pose un double problme :
(a) Dans les verbes dtat aussi on pourrait dire que le sujet est le lieu du
procs , mme si lon voit bien quil ne peut en tre lacteur . Les verbes
dtat , dans lesquels lagentivit est neutralise, nont pas de diathse , et
ne constituent en aucun cas une sous-catgorie du verbe moyen . Cette
20
A propos de la notion de verbe moyen en arabe classique , outre M. Cohen (1929),
Joon (1930), Cantineau (1950) et Fleisch (1957 ; 1968, 116117), comparer la descrip-
tion ci-dessus avec P. Larcher (1995, 2956 ; 2003, 2226) ou, pour larabe dialectal gyp-
tien , de C. Audebert (2002).
336 joseph dichy
restriction, qui est implicite dans la dfnition ci-dessus, se dduit gale-
ment de cette remarque :
Il ne sagit donc nullement de faire concider la difrence de lactif au
moyen avec celle des verbes daction et des verbes dtat (Benveniste
1950/1966, 172).
(b) Si le sujet est acteur du procs dont il est le sige, sil accomplit
quelque chose, cela suppose quil jouit dun certain degr dagentivit .
La proprit dagentivit est incluse de mme dans la notion de diathse
laquelle Benveniste a recours, notamment lorsquil propose en conclu-
sion de larticle cit, de substituer diathse interne (par opposition la
diathse externe ) au terme de verbe moyen (par opposition au verbe
actif ou externe ). Or le passage cit plus haut donne comme exemples
de ce type de verbes natre, gsir et crotre, qui sont non agentifs .
Pour traiter cette difcult, il convient demprunter au mme article
(mais non sur ce point lanalyse) dE. Benveniste la notion de verbe
mdio-passif , pour dcrire les verbes non agentifs hors verbes dtat
bien entendu comme nam u crotre, grandir, waqaa a f se trouver,
tre situ ou dans, mad passer (temps), mta u mourir, etc.
21
3.4.2 Diathse interne vs diathse externe et proprits de transitivit /
intransitivit
Le verbe moyen soppose au verbe actif ou externe :
De cette confrontation [entre deux types de verbes] se dgage assez claire-
ment le principe dune distinction proprement linguistique, portant sur
la relation entre le sujet et le procs . Dans lactif, les verbes dnotent un
procs qui saccomplit partir du sujet et hors de lui.
22
Dans le moyen , qui
est la diathse dfnir par opposition, le verbe indique un procs dont le
sujet est le sige ; le sujet est intrieur au procs (Benveniste , 1950/1966,
172). Lactif est une production dacte, rvlant plus clairement encore la
position extrieure du sujet relativement au procs ; et le moyen servira
dfnir le sujet comme intrieur au procs (op. cit. 173).
21
Je fais mienne ici une suggestion de Jean-Pierre Descls , communication orale.
22
Il est intressant de noter titre de comparaison que Sbawayhi considrait que
la transitivit fait entrer un verbe dans le chapitre des actions visibles ou audibles ( f
bb al-aml allat tur wa-tusma Kitb IV : 6). Cette description est reprise dans le
commentaire dal-Srf (en note de la mme page) et dans Ibn Ya , arh al-Mufas s al
VII : 157.
fa
ula, fa
ila, fa
ala 337
Le verbe actif ou externe est par dfnition un verbe transitif :
Ainsi se constituent partir du moyen des actifs [. . .] qui se caractrisent
toujours par ceci que le sujet , pos hors du procs , le commande dsor-
mais comme acteur , et que le procs , au lieu davoir le sujet pour sige, doit
prendre un objet pour fn (ibid.).
Rappelons que la transitivit doit sentendre, en ce qui concerne les don-
nes traites ici, comme directe ou indirecte (i.e. complment prposi-
tionnel). Exemples de verbes de diathse externe :
h akama ayan matriser qqch. ; aml al fulnin ayan dicter
qqch. qqun ; dafaa a pousser, repousser ou payer ; kataba u
crire ; qla u dire . . .
Les verbes moyens ou de diathse interne , dans la dfnition de Benve-
niste , peuvent tre aussi bien intransitifs que transitifs directs ou indi-
rects. Exemples de verbes du schme simple :
verbes moyens intransitifs :
baqiya a, rester, demeurer lorsque le sujet est humain ; farih a a se
rjouir, prouver de la joie ;
verbes moyens transitifs directs :
malla (dalternance vocalique i/a) tre ennuy par, prouver de len-
nui [ cause] de ; xa i craindre ;
verbes moyens transitifs indirects :
aara u bi-, sentir, ressentir ; raiba a f dsirer qqch, et son contraire
raiba a an dsirer viter ou spargner qqch.
4. Valeurs smantiques de base des schmes simples
Le cadre conceptuel tant pos, considrons maintenant les valeurs
smantiques associes aux trois schmes faula , faala et faila .
4.1 Les deux valeurs smantiques du schme faula
4.1.1 Les faula valeur dtat caractristique
Les verbes de ce schme sont traditionnellement dcrits comme expri-
mant un tat ou une qualit durable ou constante , par opposition aux
338 joseph dichy
verbes en faila , qui dnoteraient un tat temporaire ou instable (cf.
par exemple az-Zamaxar , Mufas s al : 230). A cette rfrence la dure ,
jai propos de substituer la notion dtat caractristique , selon la dfni-
tion ci-dessus ( 3.3.3.a). Ces verbes dcrivent efet une qualit ou un
tat pos comme constitutif ou caractristique du sujet grammatical .
Wright dsignait ce dernier comme correspondant une qualit inh-
rente naturelle (a naturally inherent quality 189698, I : 3638),
et Sbawayhi , dj cit, comme les caractres susceptibles dafecter les
choses, mot--mot qui peuvent tre dans les choses (al-xis l allat
taknu f l-ay Kitb IV : 28).
Dans les faula valeur dtat caractristique , le procs tat est, comme
on la vu, dagentivit neutralise . Ces verbes sont, comme il est attendu,
intransitifs.
23
Exemples (hors glissement smantique de type tre x o
devenir x) :
qabuh a u tre qabh laid (valeur physique ou morale ) ; h asuba u
tre h asb noble, de haute naissance ; h aluma u tre h alm clment ;
s abuh a u tre s abh brillant, radieux (visage) ou beau, joli (garon,
personne) ; kat ufa u tre kat f pais, dense.
Faula est le schme simple le plus stable dun point de vue formel. Nous
avons vu (fgures 1 et 2) que la voyelle de sa deuxime consonne radicale
demeurait inchange au paradigme du prfx (mudri ). Ce schme est
aussi le plus stable smantiquement. Sa valeur de base, ltat caractristique ,
est toutefois lobjet dun glissement de sens qui lui est troitement cor-
rl, le changement et lacquisition dtat ( 3.3.3.b et 3.3.4 ci-dessus).
4.1.2 Les faula valeur dacquisition dtat caractristique
Ce glissement smantique , trs frquent, est illustr par lexemple :
ayra anna hd ihi l-jama . . . d
.
a
'
ufa h udruh l-alaniyy . . . mund u
an . . . Mais la prsence publique de ce rassemblement . . . sest afai-
blie (mot--mot, est faible) . . . depuis que . . . (Majallat an-Nahr, 2
juillet 2006, 13, col. 1).
23
Cf. Sbawayhi : Il nexiste pas dans le langage des Arabes de verbe transitif en
faula (laysa f l-kalm faultuhu mutaaddiyan Kitb IV : 38).
fa
ula, fa
ila, fa
ala 339
Des exemples similaires ont t vus aux 3.3.4. a et b. Le contexte indi-
que clairement un dbut du procs (depuis que . . .) : le verbe daufa u
(adjectif dtat caractristique : daf faible) ne rfre donc pas un tat.
La forme conjugue au md exprime ici un vnement , qui concide
avec lachvement du processus dcrit par le verbe (lafaiblissement) :
ce processus a atteint son dernier tat , ce qui dtermine un avant et
un aprs. Cet aprs correspond un tat rsultatif que lon peut dcrire
comme un tat acquis : cette prsence publique est devenue faible ou
est dsormais faible.
On observe, travers les usages illustrs par cet exemple et ceux qui
ont t analyss plus haut, que la valeur smantique tre x incluse dans
celle de ltat caractristique se trouve trs souvent ralise en contexte
comme un changement dtat (devenir x). Ce changement a, son tour,
la proprit dengendrer un tat rsultatif . Un tel phnomne nest nulle-
ment propre larabe . On peut comparer, au moins partiellement, avec
les exemples suivants, qui sont emprunts au franais :
Ah ! Maintenant, tu es beau ! (dit une maman son fls, qui revenait en
tranant les pieds de chez le coifeur).
Lucie est dsormais notre anctre.
Dans les deux cas, les prdicats tre beau et tre notre anctre paraissent
incompatibles avec un contexte qui leur assignerait un commencement
ou un terme. Or les marqueurs temporels maintenant et dsormais
prsents dans ces exemples assignent aux procs un dbut, ce qui a pour
efet doprer une translation des prdicats verbaux ci-dessus de la cat-
gorie de ltat vers celle du changement et de lacquisition dtat . Cest
comme si lon disait : tu es devenu beau (maintenant que tu as les che-
veux coups) et Lucie est devenue lanctre de lhumanit (depuis sa
dcouverte et les recherches qui ont suivi).
Le dictionnaire de Hans Wehr-Cowan (1979) traite le plus souvent
mais pas systmatiquement le glissement smantique de ltat caract-
ristique vers lacquisition dtat comme relevant des proprits lexicales
des verbes concerns, exemples :
(a) glissement smantique trait comme inclus dans le sens du verbe :
daufa u to be or become weak ; kat ura u to be . . . numerous suivi
de : to increase, augment ; s aura u to be or become small (1
er
sens
de ce verbe ) ;
340 joseph dichy
(b) glissement smantique non pris en compte dans le sens du verbe :
h asuna u to be handsome, beautiful ; qabuh a u to be ugly ; z arufu
u to be charming (. . .) witty ; s aura u (2
e
sens de ce verbe ) : to be
young ; qaduma u to be old, ancient.
Si ce dictionnaire semble hsiter, cest que la question de savoir si ce
glissement smantique est assez frquent pour tre inclus dans le sens
lexical du verbe nest pas toujours facile trancher. Les traductions pro-
poses dans Wehr-Cowan proviennent vraisemblablement des usages
attests dans les fches des auteurs. Ces donnes montrent en tout tat de
cause que, pour un verbe donn du schme faula , les deux cas prsents
plus haut ( 3.3.4.b) se rencontrent. Voyons de quelle manire :
4.1.2a Le glissement smantique tre x o devenir x
Dans les verbes en faula dont le sens est fondamentalement celui de
ltat caractristique , comme dans h asuba u tre h asb noble, de haute
naissance, un contexte dacquisition de cet tat est presque toujours pos-
sible. Cette valeur doit alors tre considre comme smantiquement
seconde, quelle soit inscrite dans le lexique ou quelle soit limite un
nonc donn. Le glissement smantique observ correspond alors au
schma : tre x o devenir x, exemples :
h asuba l-na jruna notre voisin maintenant est devenu un noble
(nonc ironique) ;
karuma r-rajulu lhomme sest montr noble-et-gnreux.
Dans ces deux noncs, la prsence dun adjectif dtat caractristique
associ par les structures morpho-lexicales de la langue aux verbes en
faula suggre une valeur attributive : notre voisin a acquis le titre de
h asb, i.e. de noble de haute naissance (do lironie) ; lhomme a mrit
le nom de karm, noble-et-gnreux.
4.1.2.b Lengendrement de ltat acquis par lvnement ou le
processus du changement dtat (devenir x tre x)
Dans les faula dont le sens de base est celui de lacquisition dun tat ,
comme dans badua u devenir incomparable, sans pareil (al-Mujam
al-Wast et al-Munjid , racine /b-d-/), le contexte associe au verbe un
tat acquis , qui est engendr, soit au terme du processus de changement
dtat (ltat acquis correspond alors au dernier tat du processus), soit
par un vnement . Exemple :
fa
ula, fa
ila, fa
ala 341
badua ibn sna bi-ilmihi xayran wa-arran Avicenne fut incompara-
ble par sa science, en bien comme en mal. Du vivant du philosophe-
mdecin, ltat acquis et pris une valeur dtat-prsent dont la dure
inclut le moment de lnonciation (Descls , 1994, 73), la traduction
tant alors Avicenne sest rvl . . . ou se rvle . . .
Comme dans les exemples du paragraphe prcdent, le verbe en faula
peut tre interprt comme prenant une valeur attributive dont le pr-
dicat est ladjectif dtat caractristique correspondant : Avicenne mrita
le nom de bad, incomparable et admirable. Ces exemples ralisent le
schma : devenir x tre x (dans lequel se lit engendre).
4.1.3 Schme faula et adjectifs assimils et/ou participes
Si lon carte le glissement smantique tre x o devenir x, le sens
des verbes valeur dtat caractristique exclut les participes actif (ism
al-fil ) et passif (ism al-maf l ). Les valeurs aspectuo-temporelles de
base des participes sont en fonction du contexte celles de lachve-
ment (qui peut engendrer un tat rsultatif ) ou du progressif (dont le
terme peut, ou non, tre spcif).
24
Ces valeurs ne sont pas compatibles
avec celles de ltat caractristique , dont la dure nest pas borne et qui
ne comporte aucune progression. En outre, le participe passif qui ne
peut, quant lui, tre driv que dun verbe transitif , est donc en quel-
que sorte doublement exclu.
Les adjectifs assimils ou analogues [au participe] (s ifa muabbaha )
qui sont associs aux verbes en faula ainsi quaux verbes dtat caract-
ristique ou non relevant dautres schmes compensent en quelque sorte
labsence de participes. Comparer par exemple les phrases nominales :
hum kuram ils sont gnreux ; dans cette phrase, karm, plur.
kuram, est ladjectif assimil associ au verbe dtat caractristique
karuma u ;
(a) hum t libn ils sont en train de demander/ils ont demand
ou (b) hum mat lbn ils sont demands ; t lib et mat lb sont les
24
Pour les deux valeurs de base des participes (dachvement et progressive), com-
parer avec Roman (1990, 3940). Pour une analyse dveloppe de la valeur smantique
des participes en contexte et notamment du progressif terme spcif ou non, voir
Dichy (2002/2003, 5.2.2). En ce qui concerne la relation morphologique entre certains
schmes et ladjectif assimil ou le participe actif , comparer la discussion mene ici avec
Larcher (2003, 2526).
342 joseph dichy
participes actif et passif de t alaba u demander, qui ne relve pas des
verbes dtat .
Cette division entre dune part les verbes dcrivant un tat et dots dun
adjectif assimil et de lautre les verbes auxquels sont associs des par-
ticipes , rencontre toutefois un problme. Au chapitre consacr ladjec-
tif assimil (s ifa muabbaha ), az-Zamaxar (Mufas s al : 230) associe
celui-ci la valeur smantique dimmutabilit man t -t ubt (qui renvoie
ce qui est t bit immuable, tabli en quelque sorte hors nonciation du
temps). Il oppose cette dernire la valeur doccurrence dans le temps
man l-h udt , valeur quil associe aux participes. Or il nillustre pas
ces derniers par des verbes dcrivant des vnements ou des proces-
sus (comme je lai fait ci-dessus avec t alaba u demander), mais par des
verbes associs un adjectif dtat caractristique . Le premier exemple
donn est celui de h sin, qui correspond un participe actif de h asuna u
tre beau (adjectif dtat caractristique h asan). Az-Zamaxar souligne
la valeur doccurrence dans le temps man h udt de la forme h sin en
insrant celle-ci dans les phrases :
huwa h sin al-na ou adan, mot--mot : il est en train de devenir/il
est devenu beau maintenant, ou il le deviendra demain.
Les exemples suivants sont : krim, participe actif de karuma u tre
noble et gnreux, t il, participe actif de t la u tre ou devenir long
et diq, participe actif du verbe dtat caractristique dqa i tre ou
devenir troit (adjectif dtat caractristique dayyiq), pour lequel
az-Zamaxar cite le verset : wa-diqun bihi s adruka, alors que tu
es angoiss par cela, mot--mot, ta poitrine tant oppresse (rendue
troite) par cela (Coran 11 : 12).
Ibn Ya commente, dans ce verset, le choix du participe actif diq au
lieu de ladjectif dayyiq en indiquant que le premier rfre une troi-
tesse (i.e. une oppression) [correspondant ] un accident [se produi-
sant] dans le prsent et non un [caractre] immuable (dq rid f l-h l
ayr t bit arh al-Mufas s al VI : 83).
Si les verbes associs des adjectifs dtat caractristique peuvent tre
compatibles avec le sens des participes actifs , cest en raison :
(a) soit du glissement de sens qui permet ces verbes, lorsque leur sens
premier est celui de verbes dtat , de prendre celui de lacquisition dtat
(glissement tre x o devenir x),
fa
ula, fa
ila, fa
ala 343
(b) soit du fait que ces verbes rfrent, fondamentalement, lacquisition
dun tat caractristique , cest--dire, un processus ou un vnement
engendrant un tat rsultatif (schma devenir x tre x).
Les quatre exemples cits par az-Zamaxar relvent de ces deux cas.
Si h asuna tre beau relve de la premire catgorie, on a vu qual-Mu-
barrid , cit au 3.3.3.b, considrait karuma comme ayant une valeur
dacquisition dtat , et que le dictionnaire de Hans Wehr-Cowan oscillait
entre la valeur de base attribuer chaque verbe. La valeur attributive
se distribue, comme on la vu, sur les verbes relevant tant de (a) que de
(b). Il apparat clairement, en consquence, que lon ne peut rduire les
valeurs smantiques du schme faula au seul sens de verbe dtat : il sy
ajoute lacquisition dtat .
4.1.4 Deux consquences de lagentivit neutralise ou de la non
agentivit des faula
Lagentivit neutralise des verbes valeur dtat caractristique , et la
non agentivit des verbes dnotant lacquisition dun tat caractristi-
que entranent galement limpossibilit pour ceux-ci de limpratif :
ce dernier nest possible que si un certain degr de contrle exerc par
lagent sur le procs est envisageable (cest pourquoi les verbes pouvoir
et devoir nont pas dimpratif en franais ). En outre, lagentivit neutra-
lise est incompatible avec la transitivit . Il sensuit que la conjugaison
des verbes correspondants exclut toutes les formes du passif (ce dernier
ntant possible quavec des verbes transitifs , dans le cadre de contraintes
smantiques prcises).
25
Le schme faula est relativement peu frquent dans les textes, notamment
en arabe littraire moderne . On lui prfre sans doute les constructions
adjectivales correspondantes. Cela ne signife nullement que ce schme
soit sorti de lusage ou en voie de disparition, comme le montre lexemple
de daufa ci-dessus, emprunt la presse actuelle. On trouve galement
assez souvent dans la prose contemporaine des emplois de construction
impersonnelle , la faveur, pour ainsi dire, de lagentivit neutralise :
l yah sunu an . . ., il ne convient pas que . . . [Yawm., 79, 18] ; yajduru,
il convient, suivi de an ou dune forme infnitive . . .
25
Les critres qui permettent, interdisent ou restreignent lapparition de limpratif
ou du passif en arabe ont t prsents dans Ammar et Dichy (1999, 1920).
344 joseph dichy
4.2 Les valeurs smantiques de base du schme faila
Le deuxime schme du verbe simple est dalternance vocalique i/a.
Comme on la vu au 2.2.1, une variante en i/i porte sur un petit nombre
de verbes : sur 15 ou 16 verbes inventoris par les traits de morphologie
arabe mdivale , huit sont attests en arabe moderne par le dictionnaire
de Hans Wehr (cf. liste dans Ammar et Dichy 1999, 24). Cette variante,
qui afecte essentiellement des verbes de 1
re
consonne radicale w ou y est
purement formelle, et nentrane pas de difrence de valeur smantique
avec le schme faila dont elle relve. Exemples :
warit a i hriter ; waliya i suivre ou succder .
On rencontre dans les verbes du schme faila tous les degrs dagenti-
vit (pleine ou partielle ) ou labsence de cette proprit (non agentivit ,
agentivit neutralise ), ainsi que les difrentes formes de transitivit ou
de son contraire. Do la partition des sens associables ce schme qui
a t observe par les auteurs qui se sont attachs le caractriser.
26
La
valeur la plus gnrale est celle de verbes moyens ou de diathse interne
(au sens revisit prsent plus haut), ce schme comportant en outre un
nombre limit de verbes associs un adjectif dtat caractristique .
Il y a quatre grandes catgories smantiques de verbes en faila , dis-
tingues principalement au moyen des proprits dagentivit et de tran-
sitivit .
4.2.1 Les faila associs un adjectif dtat caractristique (verbes dtat
ou dacquisition dtat )
Un certain nombre de verbes dalternance vocalique i/a (schme faila )
relve des verbes associs un adjectif dtat caractristique . Ces derniers
incluent tous les traits prsents plus haut propos du schme faula , et
comportent comme eux les deux grandes valeurs smantiques troite-
ment corrles tre x et devenir x. Les verbes en faila relevant de cette
sous-catgorie sont intransitifs , dagentivit neutralise (valeur tre x)
ou non agentifs (mdio-passifs , de valeur devenir x), et ne peuvent avoir
de participe actif que dans les conditions, rarement ralises, prsentes
plus haut ( 4.1.3). Ils dcrivent pour la plupart un tat caractristique
qui correspond :
26
Outre M. Cohen (1929) et Joon (1930), voir notamment : Fleisch (1979, 220
suiv.) ; Ammar et Dichy (1999, 2324).
fa
ula, fa
ila, fa
ala 345
soit une particularit physique ou psychique (dfaut ou qualit )
27
afectant un sujet humain ou un animal , domestique ou appartenant
un troupeau,
soit une particularit, souvent mais pas ncessairement connote
ngativement, afectant une entit concrte ou abstraite.
Ladjectif assimil est le plus souvent du schme af al, fm. fal.
Une srie limite de verbes de cette sous-catgorie relve de racines 2
e
radicale w ou y. Jen ai dnombr 71, dont 18 de 2
e
consonne y dans Ilys
et Ns f (1995). Ces verbes ont pour particularit de ne pas appliquer
les rgles de transformation phonologiques habituelles.
28
Exemples :
(a) Verbes adjectif dtat caractristique dcrivant une qualit ou un
dfaut le plus souvent physique (humain ou animal ) :
jayida yajyadu avoir le cou long (signe de beaut) adj. ajyad, fm.
jayd ; awiza yawazu tre ou tomber dans le besoin (adj. awaz) ;
layisa yalyasu tre ou se montrer courageux, tenir sa position (dans
un combat) adj. alyas ; rawia yawau tre ou se montrer admira-
ble (humain ) pour son intelligence, sa beaut, son courage, etc. (adj.
arwa) ; awira yawaru tre ou devenir borgne (adj. awar) ; qawida
a avoir le dos et le cou allongs (cheval . . .).
(b) Verbes adjectif dtat caractristique dcrivant une particularit
afectant une entit non anime concrte ou abstraite :
awis a yawas u tre ou devenir difcile, incomprhensible, abscons
(afaire, explication) adj. awas et aws ; awija yawaju tre ou
devenir tordu, incurv, tortueux (adj. awaj, fm. awj) ; ayisa
yayasu tre ou devenir doux, souple (adj. ayas) ; qawira yaqwaru
tre vaste (maison) adj. aqwar.
Un nombre galement limit, mais plus important, de verbes en faila
associs un adjectif dtat caractristique relve de racines sans trans-
formation, exemples :
27
Voir notamment : Sbawayhi , Kitb IV : 17 ; az-Zamaxar , Mufas s al : 278. La syn-
thse ci-dessus provient de mon propre examen des donnes.
28
Rappelons que dans xfa, yaxfu craindre, dalternance vocalique i/a, la 2
e
consonne radicale w est lobjet de transformations en fonction de la voyelle qui la suit
et la prcde immdiatement, la difrence des verbes de 2
e
radicale w ou y dont on
trouve ici des exemples.
346 joseph dichy
(c) Verbes dcrivant une qualit ou un dfaut chez un humain ou un
animal :
baliha a devenir faible desprit ou tre stupide, idiot (adj. ablah, fm.
balh) ; ramida a devenir, puis tre chassieux (il) ou (mot--mot)
devenir ayant, puis avoir lil chassieux (humain ) adj. armad ;
darida a perdre ses dents, devenir ou tre dent (adj. adrad) ;
s awifa a (mot--mot) devenir ayant, puis avoir beaucoup de laine,
tre laineux (ovin ou caprin) adj. as waf.
(d) Verbes dcrivant une particularit afectant une entit concrte ou
abstraite :
xadira a verdir, puis tre vert (adj. axdar vert) ; sawida a noir-
cir, puis tre noir (adj. aswad noir) ; xaliqa a suser, puis tre us
(vtement) adj. axlaq ; sawisa a se gter, tre rong par des vers
(viande, nourriture) adj. aswas.
(e) Verbes de cette sous-catgorie dont ladjectif nest pas de schme
af al :
yasira yaysaru devenir puis tre facile (adj. yasir et yasr).
La valeur smantique de base des verbes prsents en (c), (d) et (e) est
celle de lacquisition dtat , la difrence des verbes des sous-catgories
(a) et (b). Dans leur sens de base, les premiers sont donc non agentifs , et
correspondent des verbes mdio-passifs .
4.2.2 Autres verbes mdio-passifs et non agentifs en faila
Dautres verbes du mme schme sont non agentifs et intransitifs . Ils
correspondent, non des verbes diathse interne ou moyens , comme
dans les deux paragraphes suivants, mais des verbes mdio-passifs .
Exemples :
verbes de sujet grammatical non-humain :
rasila a tre longue et pendre (chevelure) ; rawiya a tre irrigu,
arros (terre) ; xadiba a verdir (arbre, jardin, terre) ;
verbes de sujet grammatical humain :
ariqa a se noyer (au sens propre), mourir noy adj. assimil
arq ; marida a, tre ou tomber malade, adjectif assimil : mard.
fa
ula, fa
ila, fa
ala 347
Tous les verbes de cette catgorie ne sont pas associs, dans les diction-
naires , un adjectif assimil . Dans le cas des deux derniers exemples
ci-dessus, il est noter que le participe actif est attest, mais avec une
difrence de sens du plus grand intrt pour la distinction entre les
valeurs smantiques du participe actif et celles de ladjectif assimil (ci-
dessus, 4.1.3) :
Le Munjid al-lua l-arabiyya l-mus ira (2000) donne lexemple de
anqaz tu riqan jai sauv un noy (racine /-r-q/), et Roman : l
takul hd t -t ama fa-innaka mridun in akaltahu Ne mange pas
de cette nourriture, car tu seras malade si tu en manges (1999, 71,
daprs le Lisn al-arab, XIII
e
s., racine /m-r-d/).
Comparons ces emplois. Les deux adjectifs assimils arq et mard
correspondent des tats rsultatifs engendrs par lvnement ou con-
cidant avec la dernire tape du processus dcrit par le verbe (Descls
1994) : le processus se noyer a pour dernire tape ( !) tre mort noy,
dcrit par ladjectif arq, de mme que tomber malade entrane ltat
dcrit par ladj. mard. Par contraste, les participes actifs correspon-
dants riq et mrid ont une valeur de progressif terme spcif (Dichy
2002/2003, 34, 36, 60) :
le procs de noyade na pas atteint son terme dans lexemple du Mun-
jid (le sauvetage a pu avoir lieu) ;
le terme (ou lachvement ) du procs tomber malade est projet dans
le futur dans lexemple du Lisn cit par A. Roman . Ce dictionnaire
glose, signifcativement mridun la suite de la phrase ci-dessus par
ay tamradu cest--dire : tu seras malade.
Il est signifcatif ce quatteste notamment lexemple du Munjid al-lua
l-arabiyya l-mus ira que ces drivations restent vivantes en arabe
contemporain , mme si leur frquence est faible.
4.2.3 Les verbes de schme faila diathse interne (ou moyens ) et
dagentivit partielle
Un nombre beaucoup plus important de verbes du schme faila corres-
pond des verbes diathse interne , ou verbes moyens : le procs afecte
le sujet , que le verbe soit transitif ou non (ainsi : rire de qqun ou rire).
Ces verbes sont transitifs indirects ou intransitifs et dagentivit partielle
348 joseph dichy
(lagent , ntant pas autonome par rapport au procs , ne conserve quun
contrle limit sur celui-ci). Ils peuvent soit admettre un participe actif ,
mais non un adjectif assimil , soit tre associs dans le lexique un
adjectif assimil , et nadmettre de participe actif que dans les construc-
tions vues au paragraphe prcdent et en 4.1.3. Exemples :
(a) Verbes admettant un participe actif , mais non un adjectif assimil :
verbes intransitifs :
baqiya a rester, demeurer qq part (agent humain ) [Yawm., 12, 12] ;
nma a dormir ; dah ika a rire (en emploi absolu, difrent de : rire
de . . ., se moquer de . . .) ;
verbes transitifs directs ou indirects :
dah ika a min ou al rire de, se moquer de qqun ; nasiya a oublier
qqch ; samia a entendre qqch ; saima a sennuyer de . . . (qqch)
[Yawm., 50, 10] ; h asiba a anna . . . penser, croire, estimer que . . . ;
alima a anna . . . savoir que ou apprendre que . . . alima a bi-, avoir
connaissance de . . . .
(b) Verbes associs dans le lexique un adjectif assimil , mais nadmet-
tant de participe actif que dans les constructions lies loccurrence dun
procs (h udt) 4.1.3 :
verbes intransitifs :
at ia a avoir soif (adj. at n) ; taiba a prouver de la fatigue, tre
fatigu (adj. taib) ; xafra a prouver un fort sentiment de timidit,
tre trs timide, notamment pour une jeune flle (adj. xafr);
29
salima
a tre sain et sauf , chapper ( un mal ou un danger) adj. salm,
mais galement slim ;
30
verbes transitifs directs ou indirects :
h afya a bi- accueillir trs chaleureusement qqun (adj. h afyy) ; adiba
a al tre, se mettre en colre contre qqun (adj. adib).
Dans les exemples que lon vient de voir, la catgorie morphologique
de ladjectif assimil ne correspond pas un tat caractristique , mais
29
Il existe un second adjectif assimil , mixfar grande timide : associ ce dernier,
xafrat tre une grande timide relve des verbes valeur dtat caractristique , dans
lesquels lagentivit est neutralise.
30
Dans des cas tels que slim ladjectif assimil est dit, dans la grammaire arabe tra-
ditionnelle, prendre la forme (s a) du participe actif . Autres exemples : yis dsespr,
bis misrable, mhir adroit, habile (artisan) . . .
fa
ula, fa
ila, fa
ala 349
plutt ltat rsultatif produit pas le procs (vnement ou processus ) :
le verbe salima produit ainsi salm sain et sauf , mais galement slim
min . . . sauf de . . . (mal ou danger). Dans un verbe exprimant un pro-
cessus , comme dans adiba se mettre en colre, ladjectif adib dcrit
le dernier tat de ce processus ; si adiba se trouve, dans un contexte
donn, dcrire un vnement, ladjectif dcrit ltat rsultatif engendr
par cet vnement.
Ces verbes permettent de prciser la notion dagentivit partielle . Dans
les exemples ci-dessus, le contrle limit du procs par lagent peut sur-
tout tre glos ngativement :
31
refuser de rester quelque part, voire dtre sain et sauf dun danger
(et agir pour sortir de ce lieu ou de cette situation) ; combattre ou sur-
monter un sentiment ou une motion (rire, ennui, timidit, chaleur
humaine ou colre envers quelquun), une sensation (sommeil, fait
dentendre, soif , fatigue), un mouvement de la pense ou proces-
sus cognitif (croire, oublier, savoir), etc.
On notera enfn que la relation entre le verbe et, soit un participe actif ,
soit un adjectif assimil, est lexicalise (et non compositionnelle, comme
elle ltait sans doute aux plus anciennes poques de la langue arabe ) : on
doit recourir aux connaissances lexicales pour la construire, et non aux
seules connaissances grammaticales ou une dduction partir du sens
du verbe.
4.2.4 Les verbes du schme faila diathse interne , transitifs et
dagentivit entire
Lautre grande partie des verbes du schme faila est constitue de verbes
galement moyens (ou diathse interne ), dagentivit entire et de rgime
syntaxique transitif direct ou indirect . Dans cette confguration, le pro-
cs comporte un agent autonome contrlant efectivement (du fait de
lagentivit entire ) lvnement ou le processus dcrit par le verbe et un
patient (du fait de la transitivit ). Il englobe en outre dune manire ou
dune autre lagent ou se droule au moins partiellement en lui (diathse
31
Cf. Roman (1990, 4243), qui dcrit lagentivit partielle comme une agentivit
ractive.
350 joseph dichy
interne ). Ces verbes ont un participe actif , et nadmettent pas dadjectif
assimil . Exemples :
ariba a, boire ; radiya a bi- tre satisfait de . . . ; lah iqa, suivre, pour-
suivre qqun, lah iqa bi-, suivre qqun au sens de le rejoindre [Yawm.,
66, 21] ; rakiba a, monter, chevaucher ; wat iqa i bi-, faire confance
(qqun).
4.3 Les valeurs smantiques de faala et leur dispersion
Le troisime schme de la forme simple du verbe est faala . Il comporte
comme on la vu une alternance vocalique de base ( faala i) et deux
variantes rsultant de divers conditionnements phontiques ( faala u et
faala a).
On pourrait poser que ce schme comprend majoritairement des ver-
bes relevant de la catgorie des verbes actifs ou externes qui soppose
structurellement celle des verbes moyens , au sens donn ces termes
dans larticle de Benveniste cit plus haut (section 3.4). Ce trait serait
alors ce qui distingue faala des schmes faila et faula .
Or il existe un nombre trs lev de verbes ne rpondant pas cette
dfnition de principe. Nous avons vu que ce schme connaissait une
importante dispersion sur le plan formel : lordonnance systmatique du
tableau des schmes de base du verbe simple (fg. 1) est principalement
rompue dans le tableau des schmes du verbe simple et de leurs sous-ca-
tgories (fg. 2) par les trois alternances vocaliques possibles du schme
faala . Les exemples ci-dessous montrent en outre que cette dispersion
se vrife galement du point de vue smantique.
Considrons dabord les catgories manant des proprits dagentivit
et de transitivit :
Verbes pleinement agentifs et transitifs (directs ou indirects ) :
amana u bi- avoir foi en ou dans ; qla u dire ; saala a interroger,
questionner ; s h a i bi- crier vers, en direction de, contre qqun, le
hler ; da u il revenir .
Verbes pleinement agentifs et transitifs indirects , auxquels est associ un
adjectif exprimant un tat rsultatif , ou tat acquis :
lat afa u bi- se montrer gentil, prvenant envers qqun (adjectif assi-
mil , exprimant un tat rsultatif : lat f cf. Coran 42 : 19) ; daxala u
f entrer dans un clan ou une tribu, se placer sous sa protection (adj.
exprimant un tat rsultatif : daxl) ;
fa
ula, fa
ila, fa
ala 351
Verbes pleinement agentifs et intransitifs :
makat a u rester, demeurer (sujet humain ) ; jalasa i sasseoir ; haraa
a se prcipiter (tte baisse) ; sakata u se taire (aprs avoir parl) ;
s amata u se taire, garder le silence.
Verbes dagentivit partielle transitifs (directs ou indirects ) :
ra a voir ; fza bi- gagner, obtenir qqch ; aara u bi- sentir, res-
sentir qqch.
Verbes dagentivit partielle intransitifs :
z ahara a apparatre (intentionnellement) lorsque lagent est humain
32
[Yawm., 32, 22] ; warada i, apparatre intentionnellement, arriver
(avec un agent humain ).
Verbes non agentifs intransitifs :
mad i passer, se passer (laps de temps, vnements ou processus
dont on prend en compte la dure ) [Yawm., 32, 16] ; bad u paratre
(avec un sujet non humain) [Yawm., 32, 17] ; h na i se produire acci-
dentellement [Yawm., 61, 15] ; araba u se coucher (soleil) ; mta u
mourir.
Verbes non agentifs transitifs (directs ou indirects ) :
waqa a al tomber sur [Yawm., 32, 17] ; h aw i contenir, inclure,
comprendre.
Verbes dagentivit neutralise ou non-agentifs et intransitifs , associs
un adjectif assimil dcrivant un tat caractristique (verbes dacquisi-
tion dtat ) :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
jda u tre ou devenir excellent (adj. jayyid) ; xafa i tre ou devenir
lger (adj. xaff ) ; adda u tre ou devenir intense, dur ou violent
(adj. add) ; s amala u devenir ou tre rigide ou sec (adj. s mil et
s aml).
Reprenons cette analyse en classant les verbes de schme faala selon les
grandes catgories prsentes plus haut. On a :
32
Lagentivit est susceptible de varier selon que le sujet grammatical rfre un
humain (qil) agissant intentionnellement ou non, ou un non humain (ayr qil).
Avec un sujet grammatical non humain, ou humain, mais dpourvu, dans le contexte,
dintentionnalit, z ahara ou warada sont des verbes non agentifs . Un autre exemple (du
schme faila cette fois) est dans Yawm. (16, 15) : fa z afra n-nawmu bi-jufn, le som-
meil triompha de mes paupires. Dans cette phrase, le verbe z afra perd le trait dagenti-
vit quil aurait avec un sujet grammatical humain , et devient non agentif (le sommeil ne
disposant daucun contrle sur le procs , dont il nest dailleurs pas lagent ). En revanche,
waqaa tomber demeure non agentif, que le sujet soit humain [Yawm. 16, 12] ou non
humain [Yawm. 16, 20 et 32, 17].
352 joseph dichy
(a) Verbes valeur dtat caractristique ou dacquisition dtat caractris-
tique repris en partie des exemples ci-dessus :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
lna i tre ou devenir souple (sens physique), ou tendre (sens psy-
chologique afectant un sujet humain ) (adj. layyin) ; jda u tre
ou devenir excellent (adj. jayyid) ; sda u tre ou devenir le chef
(adj. sayyid) ; xafa i tre ou devenir lger (adj. xaff ) ; qalla i tre
ou devenir peu nombreux (adj. qall) ; adda i et u tre ou devenir
intense, dur ou violent (adj. add) ; s amala u devenir ou tre rigide
ou sec (adj. s mil et s aml) ; fasada u et i se corrompre, pourrir,
i.e. pour une viande, de la nourriture ou les murs dune personne,
etc. : devenir, puis tre corrompu (adj. fsid, fasd).
(b) Verbes diathse interne ou moyens repris des exemples ci-dessus :
Verbes associs dans le lexique un adjectif assimil exprimant
un tat rsultatif :
lat afa u bi- se montrer gentil, prvenant envers qqun (adj. lat f ) ;
daxala u f entrer dans un clan ou une tribu, se placer sous sa
protection (adj. daxl) ;
Verbes sans adjectif assimil :
amana u bi- avoir foi en ou dans ; makat a u rester, demeurer
(agent humain ) ; jalasa i sasseoir ; haraa a se prcipiter ; ra a
voir ; da u il revenir ; fza bi-, gagner, obtenir qqch ; z ahara
a apparatre, se manifester intentionnellement (agent humain )
[Yawm., 32, 22] ; warada i apparatre intentionnellement, arri-
ver (agent humain ).
(c) Verbes mdio-passifs repris des exemples ci-dessus :
mad i passer, se passer (laps de temps, vnement ou processus
dont on prend en compte la dure ) [Yawm., 32, 16] ; bad u para-
tre (avec un sujet non humain ) [Yawm., 32, 17] ; h na i se produire
accidentellement (vnement) [Yawm., 61, 15] ; araba u se cou-
cher (soleil) ; mta u mourir ; saqat a a tomber, chuter ; h aw i
contenir, inclure, comprendre.
(d) Verbes diathse externe (transitifs ) :
qatala u tuer ; daraba i frapper ; h asaba i compter (des objets) ;
madda u tendre, tendre ; sah ara a charmer, ensorceler ; sajana u
emprisonner ; xadaa a tromper, duper, leurrer ; fas ala i sparer,
diviser ; qat aa a couper ; fatah a a ouvrir ; ba i vendre ; qla u
dire ; saala a interroger, questionner ; t alaba u min an . . . de-
fa
ula, fa
ila, fa
ala 353
mander qqun de . . . ; s h a i bi- crier vers, en direction de qqun, le
hler [Yawm., 61, 11].
33
4.3.1 Les corrlats formels de la prsence de verbes adjectif dtat
caractristique dans le schme faala
Les verbes adjectif dtat caractristique (tat caractristique ou acquis)
du schme faala , dont on aurait attendu la prsence parmi les faula ,
sont associs des corrlats formels. Ils relvent :
soit des verbes concaves (ajwaf, i.e. de 2
e
consonne radicale w ou y),
pour lesquelles la ralisation de lalternance vocalique u/u ne semble
atteste que dans le cas de t la/yat lu adj. t awl (bien que lon
puisse galement voir dans cette forme un verbe en a/u) ;
soit des verbes redoubls (mudaaf, de 2
e
et 3
e
consonnes radicales
identiques), pour lesquels il nexiste quun tout petit nombre de ver-
bes en faula , qui sont signals comme des exceptions par Sbawayhi.
34
Deux verbes redoubls en u/u sont attests en arabe contemporain
(Hans Wehr-Cowan 1979) : arra (1
e
pers. arurtu) u se monter ou tre
mauvais, malfaisant (adj. arr) et labba (1
e
pers. labubtu) u se mon-
trer ou tre intelligent (adj. labb autre alternance vocalique : i/a).
Ces deux verbes sont en outre donns par le dictionnaire dal-Fayrz
bd , dans la glose de h abba (1
e
pers. h abubtu) u il devenir un
proche [qui reoit des marques damiti] de qqun (adj. h abb), comme
tant, avec ce verbe, les seuls membres de cette classe formelle.
35
33
Les verbes de parole du schme simple sont en arabe des verbes diathse externe .
Le passage une diathse interne se fait au moyen de schmes augments incluant ce que
Roman (1990 ; 1999/2005) appelle le morphme-cho t voir Ammar et Dichy (1999,
2728) ; Dichy (2002/2003, 4.2). Exemples : sala a interroger [schme simple , verbe
diathse externe ] vs sala an interroger, questionner (qqun) sur, puis tasala an sin-
terroger sur [schme augment incluant le morphme-cho, verbe diathse interne ] ;
qla u dire [schme simple , verbe de diathse externe ] vs qawwala attribuer (des paro-
les) qqun, puis taqawwala prtendre, allguer (au proft, dans lintrt, en faveur de
soi-mme) [schme augment incluant le morphme-cho, verbe de diathse interne ].
34
Sbawayhi indique quil ny a presque pas de verbes en faula de 2
e
et troisime radi-
cale identiques (ou redoubls, mina t-tadf Kitb IV : 3637). Ibn Xlawayh (Laysa . . . :
27), signale labsence de verbes redoubls en faula , lexception des deux cas de labba u
(dj cit par Sbawayhi, loc. cit.) et azza u avoir peu de lait (dit dune chvre).
35
Malgr lindication cite, on trouve dans le dictionnaire dal-Fayrz bd trois
autres verbes : azza (3
e
pers. fm. sing. azuzat) u voir (pour une chvre) son lait dimi-
nuer, avoir peu de lait (adj. azz), ainsi que damma (1
e
pers. damumtu) u devenir ou
tre trs laid, repoussant (adj. damm), et fakka (1
e
pers. fakuktu) u, se montrer ou tre
354 joseph dichy
Toutefois, il ne sagit pas dune rgle oprant dans les deux sens : tous les
verbes concaves ou redoubls de schme faala ne relvent pas, il sen
faut de beaucoup des verbes adjectif dtat caractristique . Les exem-
ples de s amala u devenir ou tre rigide ou sec et de fasada i et u voir
4.3, exemples (a) montrent quil existe galement un certain nom-
bre de faala adjectif dtat caractristique qui ne relvent daucune des
deux catgories formelles que lon vient dindiquer. Leur sens de base est
toutefois celui de lacquisition dun tat caractristique plutt que celui
de verbes dtat .
4.3.2 Un glissement smantique partiellement responsable de la
prsence de verbes diathse interne et de verbes mdio-passifs
dans le schme faala
Les verbes diathse externe sont agentifs et transitifs : le procs ver-
bal afecte un objet externe lagent . Un glissement smantique trs
rpandu et observable dans plusieurs langues peut toutefois modifer la
diathse de ces verbes, qui devient alors interne. Deux cas se prsentent,
selon que la valeur smantique rsultant de ce glissement est agentive ou
non. Dans le premier, le sens produit est celui dun verbe moyen ou a
diathse interne : le verbe est agentif et le procs afecte lagent. Dans le
second, le sens produit par le glissement est mdio-passif : le procs, qui
est non agentif, afecte le sujet grammatical du verbe (lagent existe, mais
il nest pas dsign). Comparons les donnes de plusieurs langues :
Verbes agentifs diathse interne , partir de verbes diathse
externe :
En franais , le sens de base du verbe transitif plonger est celui dun
procs diathse externe dans lequel un agent plonge un objet
dans un liquide, ex. : Pierre plongea sa chemise dans leau de la les-
sive. Ce verbe a pour pendant une forme pronominale exprimant
une diathse interne , se plonger, dont le sens est mtaphorique,
ex. : se plonger dans un livre (mais non *dans la piscine). En arabe ,
les sens de plonger et se plonger sont associs au verbe at asa i.
Joon donne un exemple similaire en hbreu : t aval tremper et se
tremper (1923, 95). La difrence est dans la prsence en franais
dune marque morphosyntaxique, celle des verbes pronominaux .
stupide ou faible, ces deux derniers verbes pouvant galement tre dalternance i/a. Cf.
aussi Ibn Xlawayh , Laysa . . . : 73 (note de lauteur).
fa
ula, fa
ila, fa
ala 355
Observons toutefois quen franais, de manire en partie similaire
t aval en hbreu et at asa en arabe , il existe, outre les deux sens
que lon vient de voir, une forme intransitive moyenne plonger, de
sens physique, sans marque morphosyntaxique associe.
36
Tous ces
exemples ont en commun dtre lexicaliss (la relation forme-sens
ne rsulte par dun phnomne de drivation compositionnel, mais
est inscrite dans le lexique).
Le verbe diathse externe daraba i frapper, battre en anglais,
to hit, to beat est cit dans tous les ouvrages, en langue arabe
ou en langues occidentales, pour illustrer sens du schme faala ,
lexemple mille fois reproduit tant : daraba zaydun amran Zayd
a frapp Amr. Il comporte toutefois un sens dans lequel le procs
exprim par le verbe ne rencontre pas dobjet externe lagent , mais
afecte ce dernier (le verbe est donc diathse interne ), comme
dans lexpression fge daraba f l-ardi parcourir la terre (ou une
contre, une rgion . . .). Un glissement smantique analogue se
produit en franais dans les expressions battre les chemins (sens
propre), ou battre la campagne (sens fgur).
Verbes mdio-passifs (et donc non agentifs ), partir de verbes dia-
thse externe :
Dans axad a qalbuhu yadribu f s adrihi son cur se mit battre
dans sa poitrine, le verbe daraba prend une valeur non agentive ; le
procs afecte le sujet grammatical . Il est signifcatif que le mme
glissement smantique est observable en franais avec battre et en
anglais avec to beat dans les expressions correspondantes.
Le verbe diathse externe aqqa u fendre, briser (en anglais to
break) prend galement le sens de clore (bouton de feur . . .),
cest--dire, fendre sa gaine, ou, pour une dent pousser. Ces
deux sens sont associs, respectivement, aux formes infnitives
(mas dar ) aqq et uqq. Comme dans tous les exemples ci-dessus,
le sens mdio-passif ou diathse interne drive du sens transitif
et diathse externe : il sagit dun phnomne de polysmie , inscrit
dans le lexique.
36
Toutefois, la difrence des exemples emprunts larabe et lhbreu , le sens
physique de plonger intransitif et moyen cit ici comporte en franais un ajout
smantique (ide de plongeon), les donnes attaches ce verbe tant relativement
complexes. On a, par exemple, soit : plonger dans la piscine de son jardin, soit plonger
les mains dans leau froide du bain, mais on ne peut plonger (= faire un plongeon) dans
son bain . . .
356 joseph dichy
4.3.3 Les faala diathse interne ou mdio-passifs ne correspondant ni
un corrlat formel ni un glissement smantique
Il existe par ailleurs au sein du schme faala des verbes dont le sens est
ou en tout cas parat tre originellement celui de procs diathse
interne ou mdio-passifs , et qui ne relvent pas formellement des verbes
concaves ou redoubls . Exemples :
faala diathse interne :
d ahaba a partir, aller, sen aller (agent humain) ; ma i marcher ;
daxala u entrer (agent humain ) ; akala u manger (alors que ariba
a boire est, comme attendu, de schme faila ) ; lamasa a toucher et
ra a voir (mais samia a entendre relve de faila) . . .
faala mdio-passifs :
zahaqa a prir, svanouir, disparatre ; waqaa a tomber ou se trou-
ver, tre situ (pour un lieu, une ville, etc.) ; d ahaba a sen aller (sujet
grammatical non humain ) ; fasada u et i se corrompre, pourrir (adj.
fsid, fasd).
4.3.4 Dispersion et rgularit dans le schme faala
Les exemples ci-dessus montrent quon trouve parmi les verbes en faala
toutes les catgories et proprits smantiques associes aux deux autres
schmes. la dispersion formelle prsente au dbut de ce travail semble
donc rpondre une dispersion smantique quivalente, dont on constate
lexamen quelle ne semble pas pouvoir tre rduite des corrlats for-
mels ou des glissements smantiques .
Une grande rgularit smantique peut toutefois tre observe : les
verbes diathse externe apparaissent exclusivement dans le schme
faala .
Cette zone de stabilit est, selon toute vraisemblance, rendue possible
par lorientation particulire du glissement smantique qui permet un
verbe diathse externe de devenir un verbe diathse interne ou un
verbe mdio-passif lorsque le procs verbal revient sur lagent ou sur
le sujet grammatical , ou, selon la formule de Benveniste , quand le sujet
devient intrieur au procs (1950/1966), exemples :
wadaa a poser, dposer est un verbe diathse externe : lagent
efectue laction dimprimer un objet le mouvement correspondant.
Le mme verbe prend le sens diathse interne de accoucher, qui
peut tre soit intransitif , soit transitif (Coran 3 : 36 et 46 : 15). Lagent
devient, dans ce cas, le lieu du procs . De mme, dans h amala i, le
fa
ula, fa
ila, fa
ala 357
sens diathse externe porter (un objet . . .) devient, avec une dia-
thse interne : porter [un enfant], devenir ou tre enceinte.
Le verbe was ala i joindre, unir une chose une autre (avec la prp.
bi-) est, dans son sens premier, diathse externe ; il est transitif un
double objet (de manire, respectivement, directe et indirecte). Au
sens de tre uni quelquun par un lien de parent, damiti . . . et de
frquenter assidment quelquun, lagent est en mme temps lun des
deux objets du procs : le verbe est alors diathse interne et transitif
( un seul objet).
Ces exemples, qui sajoutent ceux qui ont t prsents au 4.3.2,
illustrent un processus qui opre dans un sens , le sujet grammatical ou
lagent devenant le lieu du procs (comme dans wadaa et h amala) ou
lobjet de celui-ci (comme dans was ala). On peut en revanche formuler
lhypothse que le glissement smantique de la diathse interne vers la
diathse externe nest pas possible, compte-tenu des donnes de larabe :
une fois le sujet impliqu dans le procs, le verbe conserve cette pro-
prit smantique. Le passage de la diathse interne (exemple : jalasa i
sasseoir) une diathse externe sopre en construisant un autre verbe,
de mme racine, mais de schme difrent, ainsi : ajlasa faire asseoir
(schme IV, af ala, de valeur causative-factitive ).
Le passage dun verbe dtat caractristique un verbe diathse
interne se produit galement en changeant de schme, ainsi :
lat ufa u tre ou devenir gentil ou subtil est de schme faula ; lat afa
u traiter quelquun avec gentillesse, de schme faala , est diathse
interne , la gentillesse afectant lagent .
Ces donnes dobservation permettent donc de formuler une hypothse
de porte descriptive.
5. Structure de la relation entre sens et forme
dans les schmes simples
La dispersion de la relation entre sens et forme dans les schmes du
verbe simple nest donc pas irrductible. Les raisons de cette disper-
sion ne sont pas, me semble-t-il, chercher dans la dtrioration dun
schma originel ou dun proto-schma qui entrerait en cohrence avec
celui des relations formelles entre les deux paradigmes de la conjugaison
objet de la fgure 1. Sil parat raisonnable de partir dun proto-schma
358 joseph dichy
prsentant sur le mode hypothtique la relation entre sens et forme des
schmes du verbe simple dans son principe, il est essentiel de ne pas sar-
rter cette hypothse. Celle-ci est fonde sur lide que la relation sens
l forme dans les schmes serait essentiellement, voire originellement,
de nature bijective. Or je montrerai plus loin que ce nest gure le cas.
5.1 Le proto-schma, reconsidrer, des relation entre sens et forme
dans les schmes simples
Le principe de ce proto-schma associerait, de manire exclusive, la
valeur smantique des verbes adjectif dtat caractristique au schme
faula , celle des verbes diathse interne ou mdio-passifs faila , et
celle des verbes diathse externe faala :
Schme Sens grammatical
faula Verbes adjectif dtat caractristique (tat ou
acquisition dtat )
faila Verbes diathse interne ou mdio-passifs
faala Verbes diathse externe
FIGURE 3 LES RELATIONS ENTRE SENS ET FORME DANS LE
VERBE SIMPLE , CONSIDRES DE MANIRE HYPOTHTIQUE
DANS LEUR PRINCIPE
Toutes les relations ci-dessus sont bijectives (ce qui est reprsent par
une fche deux ttes l). Il est peu vraisemblable quun tel systme ait
jamais exist en ltat , car, en sus dinluctables transformations morpho-
phonologiques, des glissements smantiques dune catgorie lautre et
divers efets de fgement se sont ncessairement fait jour ds les toutes
premires poques de la langue. Un tel glissement est dj perceptible,
dans la fgure 3, dans le fait que les verbes adjectif dtat caractristi-
que peuvent se raliser comme des verbes dacquisition dtat , qui sont
mdio-passifs . Or les verbes en faila peuvent galement correspondre
des mdio-passifs . Cest donc plus une reconstruction hypothti-
que du principe qui gouverne ces relations qu une reconstitution dia-
chronique que ce schma nous convie. Ce principe est reconsidrer
de deux points de vue : il masque larbre conceptuel des notions qui le
sous-tendent ( 5.2) ; il est mis en dfaut par les grandes lignes de glisse-
ments smantiques ou de modifcations formelles observables dans les
schmes simples ( 5.3).
fa
ula, fa
ila, fa
ala 359
5.2 Larbre conceptuel des notions incluses dans cette structure
Conceptuellement, les grandes notions qui nous ont permis dtudier
les sens grammaticaux des schmes simples ne sont pas mettre sur
le mme plan : comme on la vu plus haut, les verbes diathse interne
(ou moyens ) et mdio-passifs dun ct, et les verbes diathse externe
de lautre sont dans une relation dopposition. Or, les uns et les autres
peuvent correspondre un vnement ou un processus , mais non
un tat opposition fondamentale que masquait la distinction cou-
ramment rpandue entre verbes daction et verbes dtat . Lensemble
constitu par les verbes valeur dtat caractristique soppose dune part
aux verbes mdio-passifs qui incluent les verbes dacquisition dtat
caractristique ou non , et aux verbes et diathse interne ou externe
de lautre.
Lagentivit enfn opre elle mme une partition entre les verbes
valeur dtat caractristique , dans lesquels cette proprit est neutrali-
se, et les verbes mdio-passifs , qui sont non-agentifs . Par contraste,
les verbes diathse interne ou externe sont, selon le cas, entirement
ou partiellement agentifs . Ces oppositions sont rsumes dans larbre
conceptuel suivant :
notion aspectuelle dtat notions aspectuelles dvnement ou de processus
(agentivit neutralise) non agentivit agentivit entire ou
partielle
diathse diathse
interne externe
Verbes valeur Verbes Verbes Verbes
dtat caractristique mdio-passifs diathse diathse
( y compris interne externe
dacquisition dtat )
FIGURE 4 LES VALEURS SMANTIQUES DE BASE DES VERBES
DE SCHME SIMPLE , ARBRE CONCEPTUEL
Conventions : Les proprits smantiques sont en caractres gras ; les
catgories smantiques des verbes du schme simple sont en italiques.
On notera quen raison de phnomnes de polysmie , un mme verbe
peut appartenir des catgories smantiques difrentes : les proprits
ci-dessus peuvent donc correspondre soit des catgories lexicales , soit
des occurrences en contexte.
360 joseph dichy
5.3 Glissements smantiques luvre dans la relation sens -forme dans
les schmes simples
Si lon part de laspect formel des schmes, on observe que :
(a) tous les verbes en faula correspondent un verbe adjectif dtat
caractristique (valeurs dtat caractristique ou dacquisition de
celui-ci) ;
(b) presque tous les verbes en faila relvent de lune ou lautre des
valeurs du verbe diathse interne ou du mdio-passif . Ce schme
inclut par ailleurs un sous-ensemble, rduit en nombre, de verbes
adjectif dtat caractristique ;
(c) les verbes en faala , cependant, nappartiennent que partiellement
aux verbes diathse externe .
Si lon part au contraire du sens grammatical , on observe que :
(d) tous les verbes diathse externe sont inclus dans le schme faala ;
(e) les verbes adjectif dtat caractristique sont en contrepartie repr-
sents dans les trois schmes ;
(f) les verbes moyens et mdio-passifs sont rpartis entre les schmes
faila et faala (dans le schme faula , les verbes dacquisition dtat
sont mdio-passifs ).
Les deux relations (a) et (d) sont de la forme tous les . . . sont . . . ; la rela-
tion (b) est du type presque tous les . . . sont . . .. Ce qui prcde peut tre
reprsent par la fgure suivante :
Schme Sens grammatical
faula
faila
faala
Verbes adjectif dtat
caractristique
(tat ou acquisition dtat )
Verbes diathse interne ou
mdio-passifs
Verbes diathse externe
FIGURE 5 LES RELATIONS ENTRE SENS ET FORME DANS
LE VERBE SIMPLE
fa
ula, fa
ila, fa
ala 361
Conventions :
Les fches dont le trait est doubl ( ) indiquent une rela-
tion mettant en jeu la totalit des lments de lensemble considr. La
convention se lit : pour tout schme ou sens grammatical situ au dbut
de la fche, le sens ou le schme est celui qui est indiqu par la pointe de
la fche (cf. section 4.1).
La fche interrompue de trait doubl ( ) indique une
relation mettant en jeu la quasi totalit des lments de lensemble
considr, avec toutefois un ensemble dexceptions clairement identi-
fable. La fche il ny en a quune se lit : pour presque tout schme
situ au dbut de la fche, le sens grammatical est celui qui est indiqu
par la pointe de la fche (voir section 4.2).
Les fches en pointills ( ) indiquent une relation qui ne
concerne quun nombre rduit et ferm de verbes , dans linventaire
duquel apparaissent des contraintes morphologiques , comme par
exemple dans les verbes dtat caractristique du schme faala
( 4.2.1 et 4.3.1).
La fche en trait plein ( ) signale une relation entre sens
grammatical et schme qui concerne un nombre important et en tout
tat de cause ouvert de verbes . Ces verbes sont produits par un pro-
cessus identifable de glissement smantique ( 4.3.2 et 4.3.3).
6. Le dcrochage partiel entre sens et forme :
une relation non bijective
Le dcrochage partiel entre sens et forme illustr ci-dessus est d au jeu
de deux types de contraintes : formelles (incluses dans les fches trans-
versales en pointill) et smantiques (fche transversale en trait plein).
Pour comprendre la structure gnrale laquelle on est confront, il faut
tenir compte du fait que la relation entre les schmes et leur sens gram-
matical nobit par une relation bijective. On rencontre en efet, soit des
relations schme o sens grammatical , comme dans faula et faila , soit
une relation schme m sens grammatical , comme dans faala , mais non
la relation bijective ( deux sens) schme l sens grammatical , comme le
laisserait croire une conception nave du langage.
Un tel dcrochage est frquent dans les langues : on le rencontre,
typiquement, dans la relation entre les actes de langage (Searle) et leurs
362 joseph dichy
ralisations dans une langue donne. En reconnaissance, il est gale-
ment trs frquent dans les valeurs aspectuo-temporelles associes en
contexte une forme verbale : une mme forme supportant plusieurs
interprtations, cest la mise en rapport de celle-ci avec divers indices
contextuels qui permet au processus de comprhension doprer (Des-
cls et al. 1998 ; Descls et Guentchva , paratre). Le reprage dindices
est particulirement important dans le processus de lecture en arabe ,
lcriture courante tant dpourvue des signes diacritiques secondaires
dits de vocalisation.
Dans la constitution du lexique arabe, la relation entre les sens gram-
maticaux associs un schme donn et les sens lexicaux qui sadjoignent
un nom ou un verbe est soumise lefet du principe de fgement
lexical (PFL), qui soumet le sens grammatical des schmes, ds le
niveau du mot, des carts smantiques (Dichy 2003, 204208). Dans le
verbe simple , des glissements smantiques tels que ceux qui sinstaurent
entre les valeurs de verbe dtat (tre x) et dacquisition dtat (devenir
x) ou celui qui permet certains verbes diathse externe de pren-
dre une valeur de diathse interne , soumettent les relations entre le sens
grammatical des schmes et leur forme de fortes pressions. Le proto-
schma de ces relations (fgure 3) persiste toutefois dans le schma des
relations observables (fgure 5), mais avec des connexions entre sens et
forme qui, ntant pas bijectives, sinscrivent dans la structure gnrale
des rapports entre le sens et la forme que lon observe dans le processus
de comprhension du langage humain.
7. Rfrences bibliographiques
7.1 Sources primaires
Ibn Xlawayh , al-H usayn b. Ah mad (m. en 370/981). Laysa f kalm al-arab, d. Ah mad
A. At t r. Beyrouth : Dr al-Ilm li-l-malyn, 1979.
Ibn Ya (m. en 643/1245). arh al-Mulk f t-tas rf, d. Faxr ad-Dn Qabbwa. Alep :
Al-Maktaba l-arabiyya, 1973.
. arh al-Mufas s al. Le Caire : Maktabat al-Mutanabb / Beyrouth : lam al-kutub,
10 tomes en 2 vols.
al-Mubarrid (ou al-Mubarrad), Ab l-Abbs (m. en 285/898). Al-Muqtadab, d.
Muh ammad A. Ud ayma. Le Caire : Al-Ahrm, 1399/1978, 4 vols.
Sbawayhi (m. v. 180/796). Al-Kitb, d. Abd as-Salm M. Hrn. Le Caire : Al-Haya
l-mis riyya l-mma li-l-kitb, 1977, 5 vols.
Yawm. = Tawfq al-H akm . 1937. Yawmiyyt nib f l-aryf. Le Caire : Dr Mis r li-t -
t iba.
az-Zajjj , Ab l-Qsim (m. v. 340/952). Al-Jumal f nah w, d. Al T. al-H amad. Bey-
routh : Muassasat ar-risla, 1988 (4
e
d.).
az-Zamaxar , Ab l-Qsim. Al-Mufas s al f ilm al-arabiyya. Beyrouth : Dr al-Jl, s.d.
fa
ula, fa
ila, fa
ala 363
7.2 Sources secondaires
Abbs , Ramzi. 2004. La conception et la ralisation dun concordancier lectronique pour
larabe. Tse de doctorat en sciences de linformation, Lyon, ENSSIB/INSA.
Abraham , Maryvonne. 1995. Reprsentations smantico-cognitives des verbes (chan-
tillon des verbes de mouvement) : problmes et mthode. Tse dinformatique, Paris :
EHESS.
Ammar , Sam et Joseph Dichy . 1999. Les verbes arabes. Paris : Hatier (coll. Bescherelle).
Audebert , Claude. 2002. Verbes actifs et moyens dans le parler du Caire : une suite.
Annales islamologiques 36, Le Caire, I.F.A.O.
Badawi , Elsaid, Michael G. Carter , and Adrian Gully . 2004. Modern Written Arabic. A
comprehensive Grammar. London and New York : Routledge.
Belot , J. B. 1922. Cours pratique de langue arabe. Beyrouth : Imprimerie catholique.
Benveniste , mile. 1950/1966. Actif et moyen dans le verbe, in : Problmes de linguisti-
que gnrale. Paris, Gallimard, t. 1 (1966), 1950, repris in 1966, ch. XIV, 168175.
Blachre , Rgis et Maurice Godefroy-Demombynes .1952. Grammaire de larabe classi-
que. Paris : Maisonneuve (3
e
d.).
Boormans , Maurice. 1967. Grammaire darabe littral. Feuilles de travail. Rome : Institut
pontifcal dtudes arabes (document mimographi).
Brockelmann , Carl. 1948. Arabische Grammatik. Leipzig : O. Harrasowitz.
Cantineau , Jean. 1950. La notion de schme et son altration dans diverses langues
smitiques. Semitica III. 7383.
Caspari , C.P. 1881. Grammaire arabe. Trad. E. Uricoecha. Paris : Maisonneuve.
Cohen , Marcel. 1929. Verbes dponents internes (ou verbes adhrents) en smitique.
Mmoires de la Socit de Linguistique de Paris XXIII : 4, 225248.
Comrie , Bernard. 1976. Aspect. An Introduction to the study of verbal aspect and related
problems. Cambridge University Press.
. 1985. Tense. Cambridge University Press.
Descls , Jean-Pierre. 1990. Langages applicatifs, langues naturelles et cognition. Paris :
Herms.
. 1994. Quelques concepts relatifs au temps et laspect pour lanalyse des textes.
Studia kognitywne 1, SOW, Varsovie. 5788.
, Valrie Flageul, Christiane Kekenbosch, Jean-Marc Meunier et Jean-Franois
Richard. 1998. Smantique cognitive de laction : 1. Contexte thorique. 2. tude
exprimentale de la catgorisation des verbes daction. Langages 132 (Cognition, cat-
gorisation, langage). 2847 ; 4868.
et Zlatka Guentchva . paratre. Aspectualit, temporalit : une approche cognitive
et formelle partir des langues. Document mimographi.
Dichy , Joseph. 1993. Knowledge-system simulation and the computer-aided learning
of Arabic verb-form synthesis and analysis. Processing Arabic Report, 6/7. T.C.M.O.,
Universit de Nimgue. 6784 ; 9395.
. 2002/2003. Structure de la drivation lexicale en arabe : sens et forme des verbes et
des drivs nominaux les plus immdiats, Cours de prparation au CAPES darabe,
session 2003, question de linguistique. Paris : C.N.E.D.
. 2003. Sens des schmes et sens des racines en arabe : le principe de fgement lexical
(PFL) et ses efets sur le lexique dune langue smitique. In Sylvianne Rmi-Giraud
et Louis Panier, dirs. La polysmie ou lempire des sens. Lexique, discours, reprsenta-
tions. Presses universitaires de Lyon. 189211 (www.concours-arabe.paris4.sorbonne.
fr/cours/dichy.doc)
et Mohamed Hassoun . 2005. Te DIINAR.1 Arabic Lexical Resource, an
outline of contents and methodology. Te ELRA Newsletter, vol. 10, 2, April-June 2005.
510.
, Abdelfattah Braham , Salem Ghazali et Mohamed Hassoun . 2002. La Base de
connaissances linguistique DIINAR.1, (DIctionnaire INformatis de lARabe
version 1). In A. Braham, d. Actes du colloque international sur Le Traitement auto-
matique de larabe. Tunis : Universit de La Manouba. 4556.
364 joseph dichy
Fleisch , Henri. 1957. tudes sur le verbe arabe. Mlanges Louis Massignon. Institut
Franais de Damas. 153181.
. 1968. Larabe classique. Esquisse dune structure linguistique. Beyrouth : Dr al-
Machreq.
.1979. Trait de philologie arabe vol. 2. Beyrouth : Dr al-Machreq.
alyn , Mus t af. 1912. Jmi ad-durs al-arabiyya. 19
e
d. revue par Muh ammad A.
an-Ndir. Beyrouth, al-Maktaba l-As riyya, 1994, 3 vols. en un.
Guillaume , Jean-Patrick (1984), Quelques aspects de la thorie morpho-phonologique
dIbn Jinn. propos des verbes glide mdian. In G. Bohas et J.-P. Guillaume, ds.
tude des thories des grammairiens arabes. I. Morphologie et phonologie. Institut fran-
ais de Damas. 338490.
Ilys , Jzf et Jirjis Ns f . Mujam ayn al-fl. Beyrouth : Dr al-Ilm li-l-malyn, 1995.
Joon , Paul. 1923. Grammaire de lhbreu biblique. Rome : Institut pontifcal, rd.
1965.
. 1930. Smantique des verbes statifs de la forme qatila (qatel) en arabe, hbreu et
aramen. Mlanges de lUniversit Saint Joseph, XV : 1, 332.
Larcher Pierre. 1995. O il est montr quen arabe classique la racine na pas de sens et
quil ny a pas de sens driver delle. Arabica XLII. 291314.
. 1996. Drivation lexicale et relation au passif en arabe classique. Journal asiatique
284, 2. 265290.
. 2003. Le systme verbal de larabe classique. Aix-en-Provence : Publications de
lUniversit de Provence.
Leeman-Bouix , Danielle. 1994. Grammaire du verbe franais, des formes au sens. Paris :
Nathan.
Lyons , John. 1978/1990. Smantique linguistique, trad. fran. de la 3
e
d., revue et corri-
ge de Semantics II, Cambridge University Press (1978). Paris : Larousse, 1990.
Maingueneau , Dominique. 1994. Lnonciation en linguistique franaise. Paris : Hachette.
Moscati , Sabatino, d. 1964. An Introduction to the Comparative Grammar of the Semitic
Languages. Phonology and Morphology. Wiesbaden : O. Harrassowitz (2
e
d., 1969).
Al-Mujam al-Wast . 1973. Ibrhm Ans , Abdalh alm Muntas ir , At iyya as -S awlih ,
Muh ammad X. Ah mad , eds. Le Caire : Dr al-Marif, 2
e
d. (1
e
d. 1960).
Al-Munjid f l-lua l-arabiyya l-mus ira. 2000. Subh i H amw , d. Beyrouth : Dr al-
Mariq.
Neyreneuf Michel et Ghalib Al-Hakkak . 1996. Grammaire active de larabe. Paris : Livre
de Poche.
Nr ad-Dn , Is m. 2002. Abniyat al-fl f fyat Ibn al-H jib. Beyrouth : Dr al-Fikr
al-lubnn (1
e
d. 1982).
Qabbwa , Faxr ad-Dn. 1998. Tas rf al-asm wa-l-af l. Beyrouth : Maktabat al-Ma
c
rif,
3
e
d.
Roman , Andr. 1983. tude de la phonologie et de la morphologie de la koin arabe. Mar-
seille : Jeanne Laftte, 2 vols.
. 1990. Grammaire de larabe. Paris : P.U.F. (coll. Que sais-je ?).
. 1999/2005. La cration lexicale en arabe, ressources et limites de la nomination dans
une langue humaine naturelle. Presses Universitaires de Lyon, 1999. 2
e
d. revue et
augmente, Presses Universitaires de Lyon et Universit de Kaslik, 2005.
a-artn , Rad. 1912. Mabdi al-arabiyya. Beyrouth : Imprimerie catholique (1
e
d.).
Sylvestre de Sacy , Antoine. 1831. Grammaire arabe lusage des lves de lcole spciale
des langues orientales vivantes. Paris : Imprimerie royale (2
e
d.). Rimpr. photomca-
nique. Paris : Institut du Monde arabe, 2 vols. s.d.
Versteegh , Kees. 1997/2003. Te Arabic Language. New York : Columbia University Press,
1997. Trad. arabe : Al-lua al-arabiyya, trxuh wa-mustawaytuh wa-tat ruh,
trad. Muh ammad a-arqw. Le Caire : Al-Majlis al-al li-t -t aqfa, 2003.
. 2004. Meanings of speech. Te category of sentential mood in Arabic grammar.
In Joseph Dichy et Hassan Hamz, ds. Le voyage et la langue, Mlanges en lhon-
fa
ula, fa
ila, fa
ala 365
neur dAnouar Louca et Andr Roman. Damas : Institut Franais du Proche-Orient,
269287.
Wehr , Hans, John M. Cowan (ed). 1979. A Dictionary of Modern Written Arabic (Arabic-
English), edited by J. M. Cowan, 4th edition. Wiesbaden : O. Harrassowtiz.
Wright , W. 189698. A Grammar of the Arabic Language. Rimpr. Beyrouth : Librairie
du Liban, 1974.
FEATURING AS A DISAMBIGUATION
TOOL IN ARABIC NLP
Everhard Ditters
University of Nijmegen
1. Introduction
Featuring is the afxation of, general as well as language dependent,
second-level labels (such as gender
1
or animate ) to frst-level non-
terminals (such as noun , verb or particle ) in a formal description of
the syntax and semantics of a specifc natural language . Such a formal
grammar may serve as input for the automated processing of the natural
language (NLP ) concerned: Modern Standard Arabic .
About 98% of Modern Standard Arabic texts, whether in printed or
electronic form, are represented in a non-vocalized shorthand form.
Te analysis of this kind of data, automated or otherwise, has to cope
with an exponential combinatorial ambiguity unless:
one considers the units of linguistic description beyond word level
such as constituent , sentence , paragraph or even text-level;
one combines this description with an adequate, coherent, consistent,
and as exhaustive as possible, featuring system for analysis purposes.
Terefore, our objective is: the design of a featuring system as a tool
for the disambiguation
2
of undesired analysis results in the automated
processing of Arabic text data .
3
1
Feature names and feature values are represented in the text in Italics. We use two
types of features: inherent and inherited features . Te former represents the intrinsic
semantic value of the entry concerned. Te latter follows as modifer the entry it modi-
fes according to language dependent concord and agreement rules .
2
We use the term disambiguation for attempts to obtain a single and most probable
syntactically and semantically correct analysis of input data in an automated processing
environment.
3
Language phenomena such as homonymy , polysemy and even antinomy should
also be controlled by means of adequate featuring .
368 everhard ditters
A comprehensive inventory of general or universal linguistic features
is not (yet) available for Modern Standard Arabic . Tis is mainly due to
the complexity of the description of semantic features . Usually, linguistic
theories and descriptive language models tend to become language
independent. However, we preferwhile exploiting current, language
independent linguistic theories and descriptive models to describe the
features that are specifc for the language under consideration.
If the linguistic model chosen for the description may have some
impact on the organization of the featuring system , it remains possible to
use heuristics and to refne the formal description any time we consider
it worthwhile in order to avoid undesired ambiguities. So we are free to
insert or to combine features such as agreement and concord ,
4
in order
to account for regularities , relationships and dependencies occurring
between diferent constituents , or between diferent elements within a
constituent .
Considering Modern Standard Arabic language description, we
distinguish three layers of featuring: a morphological , a syntactic and a
semantic layer. Tey describe the form, the function and the meaning
of elements used in a, agreed upon by convention and expandable by
language evolution, system of information interchange, tailored to the
needs of users of any natural language system . Tis tri-partition roughly
coincides with traditional classifcations into orthography , phonology ,
morphology , syntax , stylistics , discourse , and even semantics .
5
A description of the Arabic phonemes for analysis purposes does not
require a diferentiation in terms of their production characteristics. A
listing of phonemes (26 consonants , three glides [semi-consonants or
semi-vowels],
6
and three vowels [in a short and long variety]), together
with a number of graphemic alternatives, constitutes the basis for our
phonemic description of Modern Standard Arabic .
4
We distinguish between concord and agreement and reserve the latter to describe
regular patterns in relationship between an explicit agent and its predicate in a ver-
bal sentence (S
v
) involving gender (Ditters 1992, 169, n. 13; Kihm 2006, 1415; cf. also:
Bahloul 2006, 4348). We use the term concord for feature-value sharing within the
noun phrase (NP ) between the head and its modifers , as well as in the nominal sentence
(S
n
) for the marked relation between the topic and comment , involving, if applicable,
defniteness , gender , number , and case .
5
We are defnitely not trying to introduce a formalized dynamic description of the
Arabic concept of the world. We rather follow a static semantic approach using fnite
enumeration of pertinent (static) semantic features . See also subsection 4 below.
6
For coherency within our description, we adopt Sbawayhi s inventory of 29 Arabic
consonants (Hrn 1982, 4, 431).
featuring as a disambiguation tool in arabic nlp 369
Te introduction of a feature such as verbtype may well account for
alternative realizations of the fnite forms of so-called weak verbs . A
feature like complement structure
7
controls the number (argument-1
[arg-1], [arg-2], [arg-3] etc.) as well as the format (noun phrase (NP ),
prepositional phrase (PP ), complement clause (CCL ) etc.) of possible
complements of a lexical verbal entry . By means of predicate-subject
and predicate-object matching features we are in a position to consider-
ably refne our formal description and to distinguish, for example,
between ordinary and metaphorical language use . However, for the dis-
ambiguation of analysis results our main task at this moment remains: to
draf a general pattern for nominal and verbal featuring ; and to compose
a basic set of (semantic) features and feature-clusters embedded in an
adequate description of verbal complement structure .
A remark about the linguistic model chosen for the description may
be useful. We still opt for a Phrase Structure Grammar (psg) model
based on immediate constituency (ic ), but enlarged with a second level
of description to account for prevailing relationships and dependencies
within a Modern Standard Arabic sentence.
8
We even foresee the
introduction of a third level of description, grouping together pertinent
semantic features with information about their characteristic form of
realization at syntax level. Inspired by Saad s (1982) syntactic-semantic
study of the Classical Arabic verb in Case Grammar (cg ) terms, we
will use a number of Fillmore s (1968) cg ideas as well as later
developments for the sub-categorization and patterning of verbs in
Modern Standard Arabic .
9
Finally some words about the processing environment. We use the
agfl -system,
10
a parser generator for research and applications in
7
To be elaborated upon later at syntax level when speaking about: tri-, bi-, mono-,
and intransitive verbs or the reduction to a one-argument realization used to uniquely
emphasize the semantic load of the verbal entry ; and at the semantic level while speaking
about: semantic feature hierarchy; the minimal and the maximal realization of comple-
ment structure ; compulsory and optional complement structure realization; case roles ;
and verb sub-categorization.
8
For other applications such as information retrieval (IR ) or text-summarizing
(TS ), we use a Dependency Grammar (DG ) approach since it appears to express more
adequately semantic properties in terms of nodes and relations between nodes (Ditters
and Koster 2004).
9
Only recently we decided to include al-Safr (1979) in our research for his seman-
tico-syntactic features in Case-Grammar terms.
10
For more information about the AGFL formalism , now in its 2.4th Windows ver-
sion, see Koster (1991) and www.cs.ru.nl/agf. For more information about the formal
description of Arabic , see Ditters 1992; 2003a; 2003b; and forthcoming.
370 everhard ditters
nlp. Te acronym refers to Afx Grammars over Finite Lattices.
11
Te
parsers generated by the system are top-down recursive backup parsers
(Koster 1974, Meijer 1986), based on non-deterministic concepts and
the unifcation-principle (Nederhof 1993). Te agfl -formalism is
part of the family of two-level grammars : a context-free grammar is
augmented with set-valued features for expressing agreement between
syntactic categories . Te formalism is, in principle, suited for describing
morphological and syntactic structure, as well as fnite semantics in
terms of animate , concrete , human , volition and many others.
We discuss feature -sets at morphologic ( 2), syntactic ( 3), and
semantic level ( 4). We end with a conclusion ( 5) and a list of
references used ( 6).
12
2. Morphology
Strictly speaking, morphology relates to the formal description of the
individual Arabic parts of speech: verb , noun and particle , and, if needed
or useful, of their diferentiation into a further sub-categorization.
13
In
a broader sense, morphology also comprises feature names and values
such as: aspect , case , defniteness , derivation , gender , number , person ,
tense , voice , and many other features used at syntax level. Morphology,
fnally, has to account for punctuation marks and other textual noise,
normally more conveniently stored in the lexicon module.
Literary Arabic has been described as a predominantly root and
pattern language type.
14
Moreover, vowel-pattern variation combined
11
Here, the term afx , a variable with a fnite set of values , has to be taken in its for-
mal and not in its linguistic sense.
12
Within the morphological , syntactic and semantic sections, we use the traditional
Arabic language parts-of-speech (POS) diferentiation into noun , verb and particle as
headings for the subsections. In these subsections we only discuss the, for us, relevant
features .
13
Cf. Sbawayhi s tri-partition (Hrn 1982, 1:12) and the subdivision of POS into 7
classes by alSq (1977, 214).
14
Cohen (1970, 49 f.) has been one of the frst to describe, for automated Arabic lan-
guage processing , a frame of mostly three, sometimes four and rarely fve, consonants or
semi-consonants flled with a combination of vowels (including the absence of a vowel
at a certain slot) expressing semantic diferentiations to the global meaning of the conso-
nantal root combination. Elements of a small subset of the phoneme inventory are used
to produce other derivates of the base frame, whether of the category verb, noun or
adjective, with their own specifc variation on the global meaning of the consonantal
root combination.
featuring as a disambiguation tool in arabic nlp 371
with the insertion of a fnite set of auxiliary phonemes triggers a more
or less regular, more or less predictable change in the semantic load of
the base root meaning.
Te Arabic phoneme inventory has extensively been described else-
where.
15
As far as the consonants are concerned, the description com-
prises a cluster of features referring to their way and place of production,
as well as to their marking for velarization and degree of voicing . As far
as the vowels are concerned, they have been described by their place and
the openness of that place, set out against the position of the tongue in
their production process.
For analysis purposes, we have the Arabic grapheme inventory with
variant scriptures for alternatives occurring in initial, median, end or
independent position. In order to account for fully, partially or non-
vocalized Arabic input data, the listing of the Arabic vowel graphemes
will include an empty vowel realization.
2.1 Verbs
Assigning morphological features and feature values to verbs, we
account for: the realization of fnite verb occurrences on the basis of
number and place of the root consonants (radical , r1,
16
r2, r3, and less
frequent: r4 and r5); the realization of the vowel (a, i, u) of the second
consonant (C2) of the base (Stem I) verb realization in perfect and mood
aspect (vowperf, vowimperf ); and the realization of prefixed or infixed
vowels (vow or sukn ).
Morphological features , attached to elements of a fnite verb realization
at syntactic level (see also 3.1 below), afect values concerning: aspect ,
tense ,
17
voice , person , gender and number . For the description of a fnite
15
Cf. Fleisch mainly used three, sometimes four, features to defne the elements of
the consonantal system, which ends up in a 28 u 4 matrix (1961, 1:5665) or a 16 u 5
matrix (1968, 19); Versteegh (2001, 20) used a 9 u 6 matrix; Saad (1982, 6) used a 7 u
9 matrix for the consonants and a 4 u 2 matrix for the vowels. See also the phoneme
featuring of Bohas and Saguer (2007, 255 f.) in this volume.
16
A listing of the Arabic phonemic (consonant and vowel) system is given in the lexi-
cal module(s). For analysis purposes, a single description of the r1 sufces. For genera-
tion purposes, features describing occurrence incompatibilities between an r1, r2 and r3
(r4, r5) should be provided for.
17
As Badawi et al. (2004, 362 f.) do, we distinguish between a perfect and a non-
perfect (mood ) value of the feature aspect of the verb. In combination with modal and
temporal verbs and/or adverbials a complete range of temporal and aspectual diferen-
tiations (tenses ) can be described (see on aspect and tense also Eisele 2006, 195201;
Bahloul 2006, 506; and Reese 2006, 5053).
372 everhard ditters
verb form of the so-called weak verbs with alternative realizations in its
conjugation, a variable verbtype accounts for alternative verb realizations.
To account for the diferentiation in base and derived stems we use the
feature derivation . With that we can list, in the form of a context-free
metagrammar, the following feature names or non-terminal afxes
(in capital letters) with their fnite-set of values or terminal afxes (in
lower case ), connected to a fnite verb form:
18
ASPECT :: perfect; MOOD.
MOOD :: indicative; subjunctive; jussive; imperative.
19
TENSE :: temporal; modal.
VOICE :: active; passive.
PERSON :: frst; second; third.
GENDER :: feminine; masculine.
NUMBER :: singular; dual; plural.
VERBTYPE :: r1r2r3; r2=r3; r1=w; r1=y; r2=a; r2=w;
r2=y; r3=w; r3=y; r1=w, r3=y; r2=w, r3=y;
r1=hamza; r2=hamza; r3=hamza.
DERIVATION :: base; DERIVED.
DERIVED :: ii ; iii ; iv ; v; vi; vii; viii; ix; x.
RADICALS :: three; four; fve.
R1 :: alphabet.
R2 :: R1.
R3 :: R1.
R4 :: R1.
R5 :: R1.
VOW :: vowel; sukn.
VOWPAT :: VOW.
VOWIMPERF :: VOW.
VOWPERF :: VOW.
18
We will use the following conventions for the formal description of features and
the fnite-set of values at the second level of description which closely follow the AGFL
convention:
feature names are written in upper case;
feature values are written in lower case;
the rewrite symbol is a double colon ::;
a single lef-hand entry is rewritten in one ore more feature names
and/or feature values at the right-hand side;
alternative realizations at right-hand side are separated by a semicolon;
options in the right-hand side are separated by a vertical bar;
a rule will be closed by a period ..
19
Te archaic energetic-1 and energetic-2 fnite verb forms are not accounted for in
our formal description of Modern Standard Arabic .
featuring as a disambiguation tool in arabic nlp 373
2.2 Nouns
Features , at the morphological level assigned to elements of the category
noun , concern variables such as: person , gender and number .
20
For the
description of nominal plural building, it is worthwhile to diferentiate
between an external and internal formation . Besides person, gender and
number, the feature case is a pertinent part of the declension system,
in Modern Standard Arabic closely connected with the type of
declension.
For the description of a diferentiation in base and derived stems for
deverbal nouns (active and passive participles , infnitives , nouns of
time , place , doing an action once, or referring to instruments used in an
occurrence), we use the feature derivation . By means of morphological
base rules, we describe the subset denominatives of the category noun,
such as: individuality nouns , multitude nouns , vessel nouns , relation
nouns , abstract quality nouns and diminutives .
21
Te feature variable determination with values as defnite , indefnite
or morphologically neutral is another inherent or inherited (at syntax
level) characteristic of elements of the category noun . Elements of
the sub-classes demonstratives , personal pronouns , proper nouns , and
relative pronouns are inherently defnite. Elements of the sub-classes
indefnite and interrogative pronouns defnitely possess an inherent
indefnite value.
As far as the defnite article is concerned, we follow Wright s inter-
pretation of the Arabic grammatical tradition in describing this article
on the basis of its deictic and not of its generic value as a sub-class of the
demonstratives .
22
Furthermore, elements of the sub-classes adjectives ,
common nouns , elatives , numerals , and quantifers may, or may not,
receive a defnite or an indefnite value depending on their function
and occurrence in phrases at syntax level. So we list the following
morphological feature rules for a noun :
CASE :: nominative; NOMINATIVE.
23
DECLENSION :: invariable; diptote; triptote.
20
Since identical looking feature values are attached to distinct non-terminal names
(diferent category labels ), we can reuse these feature names without any risk for unde-
sired ambiguities.
21
Cf. Wright 1974, 1:109177.
22
Cf. Wright 1974, 1:264270. See also Fleisch 1961, 1:339347.
23
Te AGFL formalism allows for the use of logical markers such as +, -, and others
in combination with feature names and values .
374 everhard ditters
DENOMINATIVE :: individuality; multitude; vessel; relation; quality;
diminutive.
DERIVATION :: base; DERIVED.
DERIVED :: ii; iii; iv; v; vi; vii; viii; ix; x.
DETERMINATION :: defnite; indefnite.
DEVERBAL :: actpart; paspart; infnitive; time;
place; once; instrument.
GENDER :: feminine; masculine.
NOMINATIVE :: genitive; accusative.
NUMBER :: singular; dual; PLURAL.
PERSON :: frst; second; third.
PLURAL :: external; internal.
2.3 Particles
According to the traditional description
24
of particles we may distinguish
between: adverbs , conjunctions , interjections , and prepositions .
25
A formal
description should include:
PARTICLE TYPE :: ADVERB ; CONJUNCTION ;
INTERJECTION ; PREPOSITION .
2.3.1 Adverbs
A for us still useful sub-categorization of adverbs , at morphological level,
is that in invariable, as far as case marking is concerned, and bound and
free forms , as far as the orthographic representation is concerned.
ADVERB :: bound; free; invariable.
2.3.2 Conjunctions
As far as the coordinating particles are concerned, a distinction should
be made in cumulative and selective particles because of the importance
of number value in concord and agreement phenomena at syntax level.
We then divide the selective particles into: alternative, consecutive,
explicative, exclusive, inclusive, restrictive, and successive elements.
26
Tis
sub-categorization will be extensively used at the next higher level of
description.
24
Cf. Wright , 1974, 1:282296.
25
CF. Badawi et al. 2004, 174219; Cantarino , 19745, 2:253 f.; El-Ayoubi et al. 2003,
1:2, 275460.
26
Cf. Ditters 1992, 222228.
featuring as a disambiguation tool in arabic nlp 375
For the conditional particles we prepare their application at syntax
level with a diferentiation as to the nature of the condition in: possible,
real and unreal.
CONJUNCTION :: COORDINATING; CONDITIONAL.
COORDINATING :: cumulative; SELECTIVE.
SELECTIVE :: alternative; consecutive; exclusive; explicative;
inclusive; restrictive; successive.
CONDITIONAL :: possible; real; unreal.
2.3.3 Interjections
Elements of this sub-section of particles , rather close to discourse and
formalized language use, can, as far the morphological level is concerned,
be divided into vocatives and exclamations .
27
However, we also have to
decide whether or not proverbs and frozen expressions should be
lodged among the interjections and labeled as adverbs or adverbials at
syntax level. In any case we also like to mention here the category of
formulaic greetings .
28
INTERJECTION :: vocative; exclamation; f-expressions; proverbs;
greetings.
2.3.4 Prepositions
Te standard diferentiation of prepositions at this level is into a primary
and a secondary group. Te secondary group consists of an open set of
noun-derived entries , marked for their function by means of a defnite
accusative case value and, at syntax level, engaged in a construct state -like
link with the prepositional complement , itself marked with genitive case
value (NPgen). Te primary group comprises a fnite-set of non-derived
entries , some of which are, in their orthographic representation, directly
bound to the prepositional complement while others are unbound.
29
PREPOSITION :: PRIMARY; secondary.
PRIMARY :: bound; unbound.
27
A still poorly described domain of frozen or set expressions like: greetings, insults,
proverbs and similar insertions should be included here. Cf. Bergman 2007, 136137.
28
Cf. Elzeiny 2007, 202207.
29
Cf. El-Ayoubi et al., 2003, I:2:574592.
376 everhard ditters
3. Syntax
At syntax level, elements of the earlier mentioned word categories
and sub-classes are assembled into constituents , labeled: verb phrase
(vp), noun phrase (np), and particle phrase , sub-categorized according
to the element realizing the compulsory head function .
30
Tese phrases
realize specifc compulsory and optional functions within the two main
units of linguistic description at sentence level: the nominal (s
n
) and the
verbal (s
v
) sentence.
31
It may be clear, that the combinatorial behavior
of elements within a constituent , and that of constituents at sentence
level, comprising mutual relationships and dependencies , is controlled
by features and feature-condition rules.
In Modern Standard Arabic text data one distinguishes within the
nominal sentence (s
n
) the compulsory topic and comment functions
(Figure 1a). In a verbal sentence (s
v
) we are dealing with an compulsory
predicate function and several optional complements (Figure 1b).
Optional sentence adverbials can occur in both sentence types .
A np with a nominative case value for the head (npnom ) always
32
realizes the topic function.
33
Te comment function is realized by: a np
34
with nominative case value for the head (npnom ); an adjp with nominative
case value for the head (adjpnom ); a vp ; a pp ; or an adverb phrase (advp ).
A vp always flls in a s
v
the predicate slot. Sentence adverbials (s
adv
) are
realized by: a np with accusative case value of the head (npacc ); a pp ; an
advp; a cp ; or a ccl .
35
30
As mentioned in 1, a particle phrase could be subdivided into a prepositional
phrase (PP ), an adverbial phrase (ADVP ), an interjectional phrase (IP ), and (to remain
consistent) a conjunctive phrase (CP ), or a complement clause (CCL ).
31
Distinct concord and agreement phenomena mark a productive syntactic (and
semantic) distinction between a nominal and a verbal sentence type in MSA . Discourse
sensitive emphasis on the topic agent should be maintained, side by side with the pos-
sibility of emphasizing the action performance of, sometimes, the same agent in typical
VSO-oriented approaches. Cf. Ditters 2001, 3137.
32
Te optional occurrence of a particle , like inna with an emphasizing semantic load
and governing its complement by an accusative case value, only represents an alternative
realization within the base structure.
33
An alternative topic realization is a CCL (complement clause ), introduced by the
emphatic particle inna, governing the head of the following NP in the accusative case.
34
A sub-class of the nouns is constituted by diferent subsets of adjectives . We should
add the adjective phrase (ADJP ) with a nominalized adjective in head position as pos-
sible alternative for a head or modifer function in the sentence.
35
For detailed structural descriptions of phrases in MSA , see Ditters 1992, chapters
III and IV, and for a formal description of sentence structures see Ditters forthcoming.
featuring as a disambiguation tool in arabic nlp 377
In Figure 1a are listed the function slots within the s
n
, as well as the
categories able to fll these slots. One may easily see how many diferent
realizations of a nominal sentence are accounted for in this diagram
where alternatives are represented within square brackets and optional
elements within parentheses.
FIGURE 1A SLOTS AND FILLERS WITHIN THE S
N
S
n
Slots TOPIC COMMENT (S
adv
)
36
Fillers [NPnom ]
37
[ADJPnom ] [ADVP ]
[NPacc ]
38
[ADVP ] [CCL ]
39
[CCL ] [NPacc ]
[NPnom ] [PP ]
Figure [PCCL ]
2 [PP ]
[VP ]
In Figure 1b are listed the function slots within the s
v
, as well as the
categories able to fll these slots. One may easily see how many diferent
realizations of a verbal sentence are accounted for in this diagram
By means of variables as predicate-subject match (psmatch ) and topic-
comment match (tcmatch ) we describe regular patterns in concord and
agreement between the elements involved in a s
v
or s
n
, checking at the
same time the matching of a verb-argument-1 relation as well as the
compatibility of the elements involved in a topic-comment occurrence. In
the same way we control the occurrence of other argument realizations
of a verbal entry by means of a predicate-object match (pomatch ).
36
Te AGFL -formalism allows for free sequence variation at any level of description (this
means for us: at function and category level) of entries within the formal description and
processing . Terefore we do not need to list all mathematically possible realizations.
37
An alternative realization in the form of a CCL fller in a topic slot, as recorded from
Classical Arabic data (Qurn 2:184): an tas m (topic ) xayrun la-kum (comment ),
is, for its poor frequency in modern text data, not accounted for in our formal descrip-
tion of the S
n
.
38
We include this alternative realization to account for an absolute negator-head
combination in topic position (see below 3.2).
39
In our formal description of Modern Standard Arabic we also include under the
variable name CCL the protasis (the condition posed) in conditional and hypothetical
sentences.
378 everhard ditters
3.1 Verb Phrase
At an early stage in the Arabic grammatical tradition , regular patterns
in syntactic behavior and the meaning of the elements involved in
verbal constructions brought grammarians to a general description
of the verbal complement structure
40
and the grouping together of
verbs.
41
Briefy summarizing this tradition we can say, that a fnite-verb
realization (marked for perfect aspect , indicative , imperative mood or
their alternatives)
42
functions as the compulsory head of a vp . In a
minimum vp confguration, an implicit agent provides values for gender ,
number and person of the verbal head . Particles may precede the head
realizing optional pre-modifying (prem) functions such as a modal ,
aspectual or temporal modifer , or carrying a negation value (neg). In
a maximum vp confguration, an explicit agent , objects (compl ) as
well as other optional adverbial (adv pom) or peripheral (phr pom )
modifers may follow the head . Te general structure of a vp is presented
in Figure 2, where alternatives are included in square brackets and
optional elements in parentheses:
FIGURE 1B SLOTS AND FILLERS WITHIN THE S
V
S
v
Slots PREDICATE (S
adv
)
Fillers VP [ADVP ]
[CCL ]
[NPacc ]
Figure 3 [PP ]
40
Within the grammatical tradition grammarians spoke about: mutaaddin (transi-
tive) and ayr mutaaddin (intransitive), further diferentiated into and complemented
with: maf l bi-hi (direct object ), maf l mut laq (absolute object), maf l f-hi (object of
time or place ), maf l la-hu or li-ajli-hi (object of cause or reason), maf l maa-hu (con-
comitative object), h l (circumstantial object), tamyz (object of specifcation). Cf. Sba-
wayhi s al-Kitb. Hrn , 1:3454, 297310, 367384; see also Fleisch 1968, 177185.
41
We mention briefy: incomplete verbs (al-af l an-nqisa ), the verbs of hope (af l
ar-ra ), the verbs of beginning (af l a-ur ), the verbs of the heart (af l al-qulb ),
the verbs of praise and blame (af l al-madh wa-d-d amm ), the verbs of approxima-
tion (af l al-muqraba ), and the verbs of esteem (af l at-tafd l ). Cf. a.o.: Ayoub 1980;
Cuvalay 1994, 1996; see also Wright , 2:4752.
42
Te negation of the perfect aspect by means of the particle lam followed by a fnite
verb form in the jussive or the negation particle lan governing a fnite verb form in the
subjunctive.
featuring as a disambiguation tool in arabic nlp 379
FIGURE 2 GENERAL STRUCTURE OF THE VP
43
VP
Slots (PREM ) (NEG ) HEAD (COMPL ) (ADV POM ) (PHR POM )
Fillers [ASP] [NEG PART] VERB [ADJPacc ] ADVP [ADJPacc ]
[MOD] [CCL ] [NPacc ] [NPacc ]
[TMP] [NPnom ] [PP ] [PCCL ]
[NPacc ] [PP ]
[PP ] [VP ]
[VP ]
An active or passive voice value of the verbal head , in combination with
the variable transitivity
44
(with values as intransitive or transitive), afects
the realization of a verbal complement structure , not only in the number
of compulsory or optional arguments, but also in the form of these
arguments as fxed in the lexically given DNA-structure of the verbal
entry concerned. With the term form we here refer to the argument
realization by means of a np with nominative case (npnom ) for an
explicit agent , a np with accusative case (npacc ) for direct object (s) or its
equivalent,
45
and any subject or object attribute,
46
a complement clause
(ccl ), a co-referential vp for verbal extension ,
47
a pp as prepositional
object or its equivalent.
48
Each fnite verb occurrence in MSA , intransitive and transitive alike,
has in its complement structure a slot for an implicit agent or argument-
1 realization in the form of a sufxed pronoun (in past tense) or a
discontinuous pre - and sufxed pronoun combination (in present tense).
43
Ditters 1992, 286 f.
44
We prefer the term transitivity, with the values: intransitive, transitive, bi-transitive
and tri-transitive to the term valency or valence since the latter is less apt to determine
non-optional core meanings.
45
A complement clause (CCL ) introduced by the particle an or anna.
46
In de form of an NPacc or an indefnite adjective phrase marked for accusative case
value (ADJPindef,acc).
47
With the variable name verbal-extension we describe, in our formal grammar, the
occurrence of co-referential verb clusters, sometimes called auxiliaries, such as: incom-
plete verbs (kna or laysa, to be(come) or not to be); inceptive verbs (badaa, to begin
and the like); continuity verbs (istamarra, to continue and the like); anticipation verbs
(kda, to be on the point and the like). Cf. Haak 2006, Badawi et al. 2004, 422 f.
48
A prepositional complement clause (PCCL ) introduced by a preposition followed
by the particle an or anna.
380 everhard ditters
Combined with an explicit agent in the form of a npnom , the fnite verb
realization immobilizes into a 3
rd
person singular realization liable to
vary in gender value.
For monotransitive verbs with a direct object in the form of a npacc , in
arg-2 position, this argument will assume the arg-1 position in case of
a passive voice value of the verbal head . Te same holds for bitransitive
or tritransitive verbs : the frst npacc direct object will assume the arg-1
position in case of passive voice realization.
We may summarize the verbal complement structure in Modern
Standard Arabic in two tables. Te frst matches the features form ,
function , and transitivity against the number of possible completive argu-
ments. Together with the semantic load of the verbal entry concerned,
we have the actors for the predicate-subject (psmatch ) and the predicate
object matching (pomatch ).
TABLE 1 VERBAL COMPLEMENT STRUCTURE (COMPLETIVE )
arg-1 arg-2 arg-3 arg-4 arg-5 arg-6
function agent v-extension object -1 object - 2 object -3 indirect object
transitivity intrans/trans intrans/trans monotrans bitrans tritrans intrans/trans
form [NPnom ] [VP i] [NPacc ] [ADJP indef,acc ] [NPacc ] [PP prep]
[PP prep] [NPacc ] [ADJP indef,acc ]
[CCL an/anna/inna] [PPprep]
[PCCL prep, an/anna]
Te following table (Table 2) presents, in a general way, slots and
possible fller for (mainly circumstantial) elements occurring within
the complement structure of a lexically given verbal entry . As far as the
featuring is concerned, it is gradually being introduced in the form of
subscripts to the phrasal heads at syntax level.
TABLE 2 VERBAL COMPLEMENT STRUCTURE (CIRCUMSTANTIAL)
arg-7 arg-8 arg-9 arg-10 arg-11 arg-12 arg-13 arg-14
function situation absolute source goal time place instrument result
form [NPacc ] [NPacc ] [PPmin/an] [PPprep] [NPacc ,time ] [NPacc ,place] [PP prep] [NPacc ]
[CCL wa] [PP prep,time ] [PP prep,place ] [PP prep]
For the description of predicate-subject and predicate-object relations
in MSA (psmatch and pomatch ), we have so far exploited the Arabic
grammatical tradition . Coming to Modern Standard Arabic , we like to
featuring as a disambiguation tool in arabic nlp 381
adopt, in the next section ( 3.1), Saad s work (1982, 2026) in a Case
Grammar perspective for Classical Arabic .
Some of the features linked with a lexical verbal entry are listed
below:
AGREEMENT :: +human; human.
[PSMATCH :: ]
49
[POMATCH :: ]
TRANSITIVITY :: intransitive; MONOTRANS; BITRANS;
TRITRANS.
MONOTRANS :: acc; prep.
BITRANS :: acc2; accprep; prep2.
TRITRANS :: acc3; acc2prep; accprep2; prep3.
3.2 Noun Phrase
We distinguish within the np one compulsory and several optional
functions. Only a noun (or a nominalized adjective) can realize the
compulsory head function . Among the optional functions we mention:
negator (neg ), a mutually exclusive predeterminer (predet) and post-
determiner (postdet), postmodifer (pom ), and, for some head rea-
lizations, a complement (compl).
50
In the following general structure
of the np (Figure 3), optional functions are included in parentheses and
mutually exclusive functions in square brackets.
51
Relationships and dependencies within the np are specifed and
controlled by features and feature-condition rules. In a neg -head
combination it is the negative particle l that imposes the unique
occurrence of an indefnite common noun in head -position provided
with a defnite version of the accusative case value. Personal pronouns and
proper nouns , interrogative and indefnite pronouns (of the subset-type:
man and m) in head-position, are inherently marked as defnite and
indefnite respectively, thus they may only occur with an optional pom .
49
Tese rules will be discussed later.
50
For example: a masdar in head function with a subjective complement in the slot
for postdeterminer (NPgen) and an objective complement in the slot for COMPL (NPacc ),
both realizations in the form of a sufxed pronoun thus realizing a postdet and a
compl function: h ubb--h = love-mine-her my love for her.
51
In Figure 3 we have included neg among the reciprocally exclusive function rea-
lizations. However, here we underline the incompatibility of a predet and a postdet
in a np . Cfr. the examples in Cantarino 1974, I:114; 1975, II:220222), Badawi et al.,
2004, 464466. El-Ayoubi et al. (20012003, I:12) do not mention a neg ; neither in the
Vorfeld der Nominalgruppe, nor in the Nachfeld der Nominalgruppe.
382 everhard ditters
Only (de)verbal nouns (those marked with acc , or acc2 for transitivity ) as
well as some nominalized adjectives in head-position may occur with a
compl.
53
Other relationships and dependencies concerning defniteness ,
person , gender , number , and case , will be discussed below.
3.2.1 Defniteness
Elements of the noun subsets proper nouns and personal pronouns have
the value defnite for defniteness with consequences for post-modifying
elements at constituent and sentence level. Elements of the noun-subsets
indefnite pronouns and interrogative pronouns have the value indefnite.
Common nouns , (de )verbal nouns , and nominalized adjectives receive
their value for defniteness by the occurrence of a predet or a postdet
modifer.
3.2.2 Person
Only elements of the noun subset personal pronouns vary in their value
for person ranging from frst, second to third, while varying in number
and case feature values. All elements of other noun subsets bear the fxed
value third for the feature person.
3.2.3 Gender
Occurrences of the subsets proper nouns , personal pronouns , and common
nouns possess a lexically given value for the feature gender , whether this
FIGURE 3 GENERAL STRUCTURE OF THE NP
52
NP
Slots ([NEG ]) ([PREDET ]) HEAD ([POSTDET ]) (POM ) (COMPL )
Fillers [NEG PARTl] [ART] NOUN [NPgen] [NP ] [PP ]
[ADJP ] [NPacc ]
[PP ]
[REL CL]
52
Ditters 1992, 163 f.
53
For a more detailed discussion about adjectives in the context of the ellipsis of the
head of an NP and of nominalized adjectives see Ditters (forthcoming).
featuring as a disambiguation tool in arabic nlp 383
is marked morphologically by a default null-value, by one of the femi-
nine markers or not. Elements of the subsets indefnite and interrogative
pronouns bear the default gender value masculine, but may, in retro-
spective and supported by the unifcation-principle of the agfl -
formalism, receive an inherited feminine value provided by the context
in which they occur beyond sentence level. At syntax level we may have
to deal with the interference of semantic feature values such as animate ,
concrete , human , volition etc., here mainly related to as: concord and
agreement phenomena.
3.2.4 Number
Te singular and plural realization of elements of the subsets proper
nouns , personal pronouns , and common nouns are given lexically. A
dual number value realization is normally the result of the application
of a grammar-rule except for the personal pronouns . Only infnitives or
mas dars always have a singular number value. Apparent evidence to the
contrary is caused by the exploitation of mas dars for the creation of an
abstract lexicon.
3.2.5 Case
Case is a syntax dependent application feature . Its values (nom , gen , and
acc ) are related to the function the element concerned has at phrasal or
higher levels of linguistic description . A null-value for case , any other
ambiguity concerning diptotic np -head occurrences, and gender issues
for a frst person realization have to be disambiguated beyond phrasal
and sentence levels of description.
3.2.6 Trash
Earlier we mentioned feature-clusters . In some of them np s are involved
(concord , agreement , ps-match , po-match , and tc-match ). In this section
we discuss the occurrence of a noun in the head-position of a phrase at
syntax level. Feature values of such a noun should be capable to match
with elements occurring within the same constituent or with feature
values of heads of other constituents within a given context.
Summarizing some conclusions at this point, we list the following
formal rules connected with a lexical noun entry at syntax level:
384 everhard ditters
[AGREEMENT :: +human; human].
54
[CONCORD :: full; partial.]
[PSMATCH :: ]
[POMATCH :: ]
[TCMATCH :: ]
CASE :: accusative; genitive; nominative; null.
DEFINITENESS :: defnite; indefnite.
HEADREALIZATION :: demonstrative; personal; proper; relative;
indefnite; interrogative; adjective; common;
elative; numeral; quantifer; verbal.
3.3 Particle Phrase
As far as the occurrence of particles at syntax level is concerned: earlier we
distinguished between an adverb phrase (advp), a prepositional phrase
(pp ), an interjectional phrase (ip ), and a conjunctive phrase (cp ).
3.3.1 Adverb Phrase
At the morphological level, we sub-categorized the adverbs into: invari-
able, as far as case marking is concerned, and bound or free forms , as far as
the orthographic representation is concerned. One feature of the bound
forms involves the temporal aspect of a verb realization itself marked by
a non-perfect or mood tense (sa, directly linked to a fnite verb form,
then suggests a future temporal aspect).
55
Te interrogative a afects
the whole of a succeeding verbal or nominal sentence . Te afrmative
la mainly occurs, in Modern Standard Arabic , as a premodifer to the
particle qad in an advp . At syntax level, the bound and free forms are
further diferentiated into, among others, time , degree and manner
adverbials , attached to verbs and adjectives as well. Finally, we have to
distinguish between adverbs as head of an advp, on the one hand, and
other realizations of an adverbial function at phrasal or sentence level
such as pp , a npacc or a pccl at the other.
Formally, an adverb phrase (advp ) can be described as a, at syntax
level occurring, constituent that, in its minimal confguration, has an
adverb in the compulsory head function . Optional functions in an advp
can be labeled as a pre- or postmodifer of the head. A premodifer may,
for example, fgure as a discontinuous negative particle to adverbials as
54
Tese rules will be discussed later.
55
Te so-called TMA (tense , mood and aspect ) auxiliary particles have been discussed
by Kinberg (2001). See also Eisele 2006.
featuring as a disambiguation tool in arabic nlp 385
qat t u never or faqat no more,
56
or in the form of a preposition with
an head as in min qablu previously. An optional postmodifer can be
realized by a PP .
57
FIGURE 4 GENERAL STRUCTURE OF THE ADVP
58
ADVP
Slots (PREM ) HEAD (POM )
Fillers [PART] [ADV] [PART]
[PREP ] [NPacc ]
For the adverb we suggest the following combination of feature names
with corresponding values :
59
ADVERB :: ARGUMENTATION; CONFIRMATION;
DISCOURSE ; FOCUS; FREQUENCE;
interrogative; LOCAL; MEASURE;
MODALITY; SEQUENCE; TEMPORAL.
ARGUMENTATION :: evaluative; comparative; CONCLUSIVE.
CONCLUSIVE :: positive; negative.
CONFIRMATION :: certain; afrmative; limitative.
DISCOURSE :: greetings; wishes; emphasis.
FOCUS :: specifying; inclusive; exclusive; cumulative.
FREQUENCE :: number; repetition; iteration; coinciding.
LOCAL :: position; dimension; direction.
MEASURE :: degree; quantity.
MODALITY :: expression; noun; adjective; nisbe.
TEMPORAL :: past; neutral; future; SUFFIXED; relative.
SEQUENCE :: coinciding; preceding; succeeding.
SUFFIXED :: id in; d ka; pronoun.
3.3.2 Prepositional Phrase
Within a prepositional phrase we distinguish two compulsory func-
tions: a prepositional complement (pcompl ) and a linker (plink ) that
connects that complement to the next higher level of linguistic descrip-
tion . Te complement function can be realized by: an adverb phrase
56
Cf. El-Ayoubi et al. 2003, I:2, 275460, and 406. Badawi et al. 2004, 161174.
57
In the discussion about the NP we already mentioned the use of an accusative NP
(NPacc ) for adverbial purposes.
58
Ditters 1992, 210 f.
59
Cf. El-Ayoubi et al. 2004, I:2, 284 f., and Wright 1974, I:282 f.
386 everhard ditters
(advp ); a complement clause (ccl) introduced by the particle an or
anna (ccl an|anna); a noun phrase with genitive case value (npgen); or
a verb phrase with a head marked for subjunctive mood (vpsubj). Te
prepositional linker function is usually realized by an element of the
particle-subset of prepositions . However, there are also expanded con-
structions in which a preposition, followed by an infnitive or abstract
noun , realizes the header function,
60
and prepositional clusters in which
an infnitive, marked with an indefnite accusative case value (npacc ) and
followed by a preposition realizes the linker function.
61
FIGURE 5 GENERAL STRUCTURE OF THE PP
62
PP
Slots HEADER COMPL
Fillers [PREP ] [ADVP ]
[EXPAND-P] [CCL an|anna]
[P-CLUSTER] [NP gen]
[VP subj]
63
A good basis for a discussion about semantic features in the next section
certainly is the sub-categorization of prepositions into: time , place , and
ideal (or manner) features (Wright , 1974: II:129). With corpus based
evidence, forwarded by Cantarino (1974, II: 262 f.) and El-Ayoubi
et al. (2004, I, 2:466 f.) we are able to further diferentiate this sub-
categorization of prepositions into features values :
64
PREP :: IDEAL; PLACE; TIME.
IDEAL :: adversative; benefactive; causal; COINCIDING;
comparative; content; discourse; explicative; fnal;
HYPOTHETIC; instrumental; measure; modal;
partitive; possessive; specifcation; substitution.
COINCIDING :: comitative; simultaneous.
HYPOTHETIC :: concessive; conditional.
PLACE :: destination; direction; local; partitive; position.
TIME :: direction; partitive; temporal.
60
Cf. El-Ayoubi et al. 2004, I:2,574.
61
Cf. El-Ayoubi et al. 2004, I:2,584 f.
62
Ditters 1992, 213 f.
63
Not to confuse with the infectional phrase as in the Government and Binding
approach (cf. Fassi Fehri 1993, 16 f.).
64
See also Badawi et al. 2004, 167 f.
featuring as a disambiguation tool in arabic nlp 387
3.3.3 Interjectional phrase or clauses
Earlier ( 2.3.3) we diferentiated the interjectional phrase (ip ) into:
vocatives ; exclamations ;
65
frozen expressions , proverbs and greetings . Two
of them, vocatives and exclamations , lend themselves easily for a general
structure description (see Figure 6).
Other optional ip s ( f-expressions , proverbs , greetings , and insults ) may
provisionally receive, in our formal description of Modern Standard
Arabic , the label of a phrasal or sentence adverbial since the syntactic
structure of such an insertion can easily be described by means of the
basic general syntactic patterns we already suggested. However, the
semantic value of their occurrence in a given context has still to be
evaluated and described.
66
FIGURE 6 GENERAL STRUCTURE OF THE IP
IP
Slots (PREM ) I-HEAD (I-COMPL ) (TAIL)
Fillers [PART] [INTERJ ] [NPnom ] [PART]
[NPacc ] [NPacc ]
[S
v
] [PP il|bi|al]
[PPli]
[S
adv
]
A few comments may make this scheme more transparent. Te line
of slots tells us about the compulsory function of an ip head (i-head )
with an optional tail in case of a discontinuous realization like w-h,
be it for Modern Standard Arabic a bit archaic, as in wzaydh Woe
upon Zayd! (Badawi et al., 2004:37). It further mentions the optional
occurrence of a premodifer (prem ) or a complement (i-compl ) of
the i-head . In the fller-section alternative realizations are listed, also
comprising a possible s
v
-s
adv
combination as man yaqtul yuqtal Who
kills will die killed. For this reason we included the term clause in the
heading of this subsection.
65
Cf. Firanescu 2007, 7981.
66
Cf. Bergman 2007, II:136137.
388 everhard ditters
3.3.4 Conjunctive phrase (cp ) or Clause (ccl)
Following the Wright tradition, at the morphological level, we made a
distinction between a separable and an inseparable occurrence of con-
junctions , as well as between connective and conditional conjunctive par-
ticles (Wright , 1974, I:290294). At syntax level, however, we also prefer
to distinguish between coordinating and subordinating conjunctions .
3.3.4.1 Coordination
Te compulsory head-function in the cp is realized by an element of
the subset coordinators of the conjunctions . An example of an optional
premodifer would be the narrative connector wa preceeding the
adversative coordinator lkin but (Badawi et al., 2004:555). Te com-
pulsory complement function may be realized by a variety of alterna-
tives (Figure 7).
FIGURE 7 GENERAL STRUCTURE OF THE CP
CP
Slots (PREM ) C-HEAD
67
COMPL
Fillers [PART ] CONJ [NP ]
[VP ]
[ADJP ]
[ADVP ]
[S
n
]
[S
v
]
3.3.4.2 Subordination
Te compulsory head function in a complement clause is realized by
an element of the subset subordinators , which may contain a null value,
for example, in caseof the occurrence of a purposive li followed by a
verb with a subjunctive mood value. In our formal grammar of Modern
Standard Arabic , the compulsory complement function may vary
between a s
n
and a s
v
.
67
Te conjunctive head may be empty is case of asyndetic coordination (Badawi
et al. 2004, 539 f.
featuring as a disambiguation tool in arabic nlp 389
FIGURE 8 GENERAL STRUCTURE OF THE CCL
CCL
Slots (PREM ) HEAD COMPL
Fillers [PREP ]
68
SUB [S
n
]
[PART] [S
v
]
Te variables and features we like to retain for the conjunctive phrase or
clause are:
CONJUNCTION :: COORDINATING ; SUBORDINATING.
COORDINATING :: cumulative; SELECTIVE.
SELECTIVE :: adversative; alternative; consecutive;
disjunctive; exclusive; explicative; inclusive;
restrictive; successive.
SUBORDINATING :: circumstantial; consequential; SINGLE-
CLAUSE; DOUBLE-CLAUSE.
SINGLE-CLAUSE :: comparative; complemental; purposive;
locative; temporal.
DOUBLE-CLAUSE :: CONDITIONAL; temporal.
CONDITIONAL :: possible; real; unreal.
4. Semantics
We discussed so far feature variables with distinct values at the morpho-
logical and syntactic level of description, variables such as: aspect , tense ,
defniteness , gender , number , person , and case . Moreover, with feature-
clusters such as concord and agreement (involving: human , person ,
gender , and number ), we are able to describe regularities between a
predicate and an explicit agent (ps-match ), between a predicate and
possibly realized complements (po-match ) in a s
v
, as well as between
topic and comment realizations (tc-match ) in a s
n
.
As far as our semantic level of description is concerned, we earlier
insisted to remain, in this paper, within the limits of static semantics as
used in almost all other applications of NLP .
69
Table 3 summarizes our
68
In that case we can speak of a prepositional complement clause (PCCL ).
69
We refer to the general domains of machine translation and information retrieval ,
still using the basics on static semantics of Yngve (1966). See also Koster (2004).
390 everhard ditters
working space ending at level-zero (semantics
0
) of static semantics . Te
description of features at the levels listed in the squared box remains to
be done.
Tus, a comprehensive assignment of semantic features to all lexical
entries, which can occur as head element of one of the diferent phrasal
constituents, does not (yet) exist for automated analysis purposes of
Modern Standard Arabic . Tat may not be necessary when a heuristic
feature assignment for a ps-match , a po-match or a tc-match could
already meet our disambiguation objectives.
Te more or less detailed sub-categorization of features for headers
in an advp and a pp ( 3.3.12) is a good example of our hypothesis.
Tis feature sub-categorization supplies such an advp with a specifc
semantic load in the context of its occurrence. For the header of a pp , this
sub-categorization presupposes a matching with some corresponding
feature values of the head realization of the prepositional complement
for a grammatically and semantically correct analysis result. Terefore:
if the head -element of a prepositional complement shares the maximum
number of pertinent features with the prepositional header , alternative
analysis results should and could be discarded.
70
If a noun like bustn
possesses a feature hinting at an open-air outdoors location we could
easily match the local feature values of the header and the prepositional
70
Te fne-tuning of this matching occurs in the cyclic process of testing the descrip-
tion on new text data . Moreover, this matching produces optimal results when unifca-
tion provides contextual information. A sequence as min tah tihi remains ambiguous as
long as the anaphor hi is not disambiguated in context.
TABLE 3
Levels Features
morphology morphological
morpho-syntactic
syntax syntactical
syntactico-semantic
semantics
0
semantical
0
static-semantical
semantics
1
semantical
1
dynamic-semantical
semantics
2
semantical
2
etc.
featuring as a disambiguation tool in arabic nlp 391
complement and interpret the sequence f l-bustni in the garden as an
attributive or adverbial adjunct with local value. As a matter of fact, we also
have to provide individual lexical elements of verb and noun categories
with a minimum set of pertinent feature names and feature values .
4.1 Verbs
In our formal description of Modern Standard Arabic for automated
analysis purposes, individual lexical verbal elements have to provide
information about:
compulsory and optional completives and circumstantials within the
complement structure of a given verbal entry (transitivity );
the combination of a verbal entry with elements of other word
categories to denote a particular meaning of such a verbal entry
(collocation );
71
minimum qualifcations for possible implied or explicit agents and
completives (in terms of case-roles and syntactic realization).
4.1.1 Transitivity
Within the framework of the diinar -project
72
verbal entries have been
compiled according to the number and the syntactic realization of their
arguments. In this way the verbal lexicon is frst split up in intransitive and
transitive verbs . Te transitive verbs are then further divided into: mono -,
bi - or tri-transitive verbs . Finally, these verbal entries can be divided
into sub-categories governing a direct object (s) (npacc or ccl ), and/or a
prepositional object (pp or pccl). A further diferentiation concerns the
lexical realization of the preposition (ppli, ppmin, pp an etc.).
What is still missing in this approach is the semantic dimension as,
for instance, Levin (1993) did for English .
73
She compiled 49 verb classes
with sub-categories. Te members of each sub-class share the general
meaning of the class as well as a number of syntactic characteristics. So,
the members of the sub-class 10.2 of banish verbs (Levin 1993, 123):
74
71
Cf. Hoogland 1993; El-Gemei 2006, 434439.
72
Information about the DIINAR lexical database can be found at: http://catalog.
elra.info. Information about DIINAR-MBC can be found at: http://sites.unin-lyon2.fr/
langues_promodiinar/accueil.htm.
73
See also Levin & Rappaport Hovav 1996.
74
Tese members are: banish, deport, evacuate, expel, extradite, recall, and remove.
392 everhard ditters
relate to the removal of an entity, typically a person from a location. Te
location argument is expressed in a prepositional phrase headed by the
preposition from. [. . . .] Unlike the remove verbs , these verbs allow to
phrases as well as from phrases, though not simultaneously.
Te application of such an approach for the description of verbal entries
in Modern Standard Arabic is given in Table 4 where square brackets
point to mutually exclusive arguments and parentheses to optional
occurrences:
75
TABLE 4
arg-4 arg-3.2 arg-3.1 arg-2 arg-1 entry
manner destination source object agent
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos
76
)] NPacc ,+human
(NPnom ,+human )
77
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
78
(NPnom ,+human )
(PPbi) [(PPil|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
PPbi,+human
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
(PPbi) [(PP il|li,loc|pos)] [(PP an|min,loc|pos)] NPacc ,+human
(NPnom ,+human )
~ d ~ z d
~ z d
ulm since
*dulm is not an actual word. Tis distribution is problematic for the
classifcation of lexical items, a problem to be addressed briefy in dis-
cussing hybridity .
On the syntactic level, four features are chosen for analysis: relative
clause markers , demonstratives , negatives , and complementizers . Tey
all serve as functional heads of their respective clauses with clearly diver-
gent forms in the two varieties. Except for complementizers , they have
also been studied extensively in the literature of Arabic code-switching
(e.g. Bassiouney 2006, Boussofara-Omar 2006b, and Eid 1982, 1988.)
On the morphological level, two verb forms have been selected for
this study: passives and Stem IV verbs . Fus h and Egyptian passive verbs
are formed by diferent morphological processes. In fus h the passive is
marked derivationally by the vowel melody /u-i/ in the frst and sec-
ond syllable, respectively, (e.g. kataba wrote ~ kutiba was written). In
Egyptian it is marked infectionally by the prefx it- (e.g. katab wrote ~
itkatab was written). Stem IV verbs , formed on the pattern /a-CCVC/
like aqbal approach or agab like, are rare in Egyptian . Tey are some-
times replaced by verbs from diferent roots , e.g. gih to come instead of
aqbal, but more ofen they are replaced by Stem I (CVCVC) of the same
verb, e.g. agab instead of agab.
4
Following the transcription requirements for this volume, long vowels are repre-
sented as single segments / /. For reasons that have to do with syllable weight and
identification of minimal word, to be explained below, these should be understood as
sequences of two identical vowels /aa ee ii uu oo/.
5
The opposite, however, is not true. Not all words with sibilants , for example, in
Egyptian have a corresponding interdental fricative in fus h ; some words may have a
sibilant in both , e.g., fus h : sam sky and Egyptian sama.
408 mushira eid
Finally, lexical diferences between the two varieties come from dif-
ferent sources. Te examples in Table 1 illustrate two of such areas.
Some diferences are the result of word usage, the association of a mean-
ing with two diferent lexical items, one considered fus h usage and the
other Egyptian as, for example, anf and manaxr nose. Other difer-
ences are phonologically -related as, for example, the diference between
indi and andi I have. Because fus h and Egyptian are varieties of the
same language, it is expected that an overwhelming majority of lexi-
cal items would be shared. Table 1 provides examples of words that are
identical in both varieties, e.g. kitb book and fannn artist. I refer to
such words as ambiguous or unspecifed to mean non-distinct, or not
specifed, as to language variety . I use the term both for words that
include features from both varieties, i.e. hybrid forms.
1.2 Hybridity
A hybrid is something that is mixed, and hybridity is simply the mix-
ture. Te term has its origin in biology.
6
In linguistics it refers to a word
parts of which come from two diferent languages. Te term has become
central to major theoretical discussions among the discourses of race,
post-colonialism, identity , multiculturalism, and globalization. In theo-
ries of cultural studies, it is understood as recognition of two identities
or a mixture of identities but also as a refutation of assimilation into
a dominant culture. Bakhtin , for example, defnes hybridity [hybridiza-
tion] as a mixture of two social languages within the limits of a single
utterance, an encounter, within the arena of an utterance, between two
diferent linguistic consciousnesses, separated from one another by an
epoch, by social diferentiation , or by some other factor (1981, 358). Te
term has also been used in relation to mixed language varieties that
result from code-switching among bilinguals (Hinnenkamp 2003).
In Arabic hybrid or intermediate forms, as they are sometimes called,
include features from both varieties fus h and dialect and, therefore,
they cannot be clearly identifed as belonging to one or the other.
7
Table 2 illustrates Arabic hybrid forms with examples selected from the
6
It comes from the Latin: hybrida, a term used to classify the ofspring of a tame sow
and a wild boar.
7
See Schulz (1981, 8789) for a discussion of hybridity versus mistakes.
arabic on the media: hybridity and styles 409
interviews . In the following section, I also illustrate this with sample
texts from the interviews .
Here and in all other examples italics indicates Egyptian Arabic ,
bold indicates fus h , and unmarked typescript indicates material that
is unspecifed as to Arabic variety, hence ambiguous or shared material.
Te frst example in Table 2 is a hybrid , because it includes the Egyp-
tian present tense marker , the prefx ba-, attached to a fus h verb form
marked by the vowel melody /a-a/ of the Stem. Te second example
includes the fus h person marker prefx /yu-/, marked by the /u/ vowel
rather than the Egyptian /i/, and the fus h mood marker , the fnal /-u/,
but it also includes the sibilant /s/ of Egyptian , instead of the fus h /t /.
TABLE 2 HYBRID FORMS
Hybrid Fus h Egyptian Impermissible
hybrids
Gloss
1. bataammal a-taammal b-[]a-tammil *bataammil I contemplate
2. yu-massil-uyu-mat t il-u yi-massil, *yimattil *yi-massil-u he represents
3.a. salt ta t alt a talta *talsa/salta three
3.b. t alsa *t alta/talt a
3.c. ? salsa ? salsa
Te variation in correspondences between fus h /t / and Egyptian stops
and sibilants creates lexical possibilities for hybridity and ambiguity. Te
last example in Table 2, with the diferent pronunciations of the word for
three, illustrates this point. Te hybrid forms (3ab) allow the mixture,
or co-occurrence, of fus h /t / only with sibilants (e.g. t alsa). Combi-
nations with Egyptian stops are impermissible hybrids (*t alta). Tis
suggests that the sibilants themselves can be analyzed as in-between
spaces or ambiguous segments: at times they correspond to fus h /s/ as
in the many shared words with sibilants (e.g. sitta six), at other times
they correspond to fus h /t / as in the example (2) in Table 2. More-
over, at times they correspond to Egyptian /t/ as in the last examples in
(3). Te impermissibility of hybrids involving either the sibilants or the
interdental fricatives with the stops , as in /*talsa/ or /*t alta/, suggests
that sibilants and interdental fricatives may be considered members of
the same category or group, since they can substitute for each other.
Tis distribution also raises a question as to the identifcation of sibilant
pronunciation , particularly in forms where the only Egyptian corre-
spondence to the interdental fricatives is the sibilant . Are the examples
410 mushira eid
(3ab) in Table 2 hybrids , as I classifed them, or are they simply alterna-
tive pronunciations for the interdental fricative ? It is an open question
with arguments for both alternatives.
Te modus operandi adopted in earlier research contextualized the
decision (Eid 1982, 1988). In lexical items where speakers have all three
choices (t , t, and s), the sibilant pronunciation is counted as fus h since
the speaker has an alternative, namely /t/ pronunciation , should she or
he have opted for Egyptian . Likewise, in lexical items where Egyptian /t/
pronunciation is not available, sibilant pronunciation of fus h interden-
tal fricatives counts as Egyptian for similar reasons.
8
Te above discussion of hybrid forms illustrates the concept through
lexical items. In this article I consider hybridity to be a global feature of
text or discourse. It applies to forms of language that look more like a
collage, created out of two or more linguistic codes where the borders
between language varieties are blurred or no longer distinct.
2. Styles
Speech styles are forms of verbal self-expression and interaction . A style
is based on a speakers choice of linguistic form, discourse strategies ,
and modes of interaction . Styles vary depending on degree of formality ,
medium (spoken or written), and context or situation (conversation in
a cofee shop, for example, versus a lecture, speech, or an academic pre-
sentation). For Arabic, styles may also vary depending on language vari-
ety as well as the degree and nature of the mix between fus h and dialect
features.
Tree styles are identifed in each interview : reading, formal , and
casual styles . While they correspond to Labov s classifcation of speech
styles , they are not identical to it. Te diference lies in the formal and
casual styles , where for Labov casual is conversational and emerges in
relaxed situations when the participants become more comfortable with
each other and forget the context of the interview .
9
For Arabic the styles
8
Alternatively, sibilant pronunciation may be considered ambiguous, or shared by
both varieties. The high frequency of sibilant pronunciation in fus h contexts , e.g. read-
ing, speeches, and recitation of literary texts, supports this view. This pronunciation is
not permitted in recitations of the Quran , however.
9
Labov has a fourth style D: word lists, which is relevant as an elicitation technique
but not applicable to natural conversations as represented in these programs.
arabic on the media: hybridity and styles 411
also correspond to the relative distribution of linguistic features from
fus h and mmiyya . In this section I illustrate the three styles and pro-
vide a characterization of each based on a quantitative analysis of fus h
and Egyptian features in them, as described in Table 1.
Eight segments are selected to illustrate the three styles, four segments
from each of the two interview programs. Te interviewees in these pro-
grams are Yehia Haqqi and Anis Mansour. Both are represented by three
segments illustrating formal and casual styles . Te reading style is repre-
sented by two other speakers, one from each interview .
2.1 Styles illustrated
Te reading style can be easily identifed. It occurs when an individual
is seen reading from a script, as in the Haqqi interview , or heard nar-
rating, for example through the use of voice-over, as in the Mansour
interview (see section 3). Segment 1 and 2 from the Mansour and Haqqi
programs, respectively, illustrate the reading style .
Segment 1 with the voice-over is pre-recorded in the studio. No evi-
dence can be found of any Egyptian features, with the exception of the
substitution of Egyptian /g/ for fus h /j/, as in the word az-zawgat-u
for az-zawjat-u the wife. Tis substitution has become so common in
Egypt that it is considered an accepted fus h pronunciation in almost all
readings except those of the Koran and other religious texts . Otherwise,
segment 1 can be considered pure fus h , or as pure as it can get. Natu-
rally, this ignores the issue of lexical items and particles, or function
words, that are ambiguous or unspecifed for variety: galla majesty,
s ah fa journalism, f in, and, in if. Case markers are pronounced, for
example, and vowel patterns are those of fus h , e.g. al- for the defnite
article not il- as in as -s ah fa.
SEGMENT 1: San Mansour (interviewer)
as-sahfa hiya z-zawgat-u at -t niya f h ayt-i l-ktib-i l-kabr ans
mansrin gz-a ha t-ta
'
br. bada
"
at
'
alqat-u ans mansr bi shibat-i
l-galla as-sahfa f
'
m-i alf-in wa tisimiat-in wa sabat-in wa arban.
wa knat al-bidya f gardat-i l-ass.
Journalism is the second wife in the life of the distinguished writer Anis
Mansour, if such an expression is permitted. Anis Mansours relation-
ship with her royal highness [journalism] began in 1947. Te beginning
was in the newspaper al-ass.
412 mushira eid
Segment 2 read by Fouad Duwara in the Haqqi program has a few more
ambiguous lexical items; more importantly, however, is the Egyptian
pronunciation of the years 1962 and 1970. Tis is common in readings,
for example, of the news and on-the-air commentaries as is the case of
this reader, who is also reading on-the-air. Otherwise, fus h is main-
tained throughout in vowel patterns , e.g. h wal not Egyptian h wil, and
other pronunciation diferences.
SEGMENT 2: Narrator (Fouad Duwara in Haqqi interview )
wa f ibrl
'
m alf w tusumiyya itnn wi sittn
'
uyyin-a yehya h aqqi ra
"
s-an
li-tahrr-i magallat al-magalla wa zalla yatawalla mas
"
liyataha h atta
December alf w tusumiyya w sabn wa hwala tawl tilka s-sanawt an
yuhfz -a li l-magalla ala i
'
raha alladi t-taxadathu li nafsih mund u
inih wa huwa sigil it -t aqfat-i r-rafa.
And in April 1962 Yehya Haqqi was appointed editor-in-chief of al-
Magalla journal. He continued to be responsible for it until December
1970. He tried during all those years to preserve for the journal the
motto it had adopted for itself since it was established, and that is [to
be] a record of high culture.
Segments 35 are samples from Haqqis speech. Te frst, segment 3, has
all the characteristics of a reading style , although Haqqi was not read-
ing. Te segment has all the features characteristic of fus h including
case markers , vowel patterns , etc. and no features of Egyptian . It also
includes only a few words (6 total) unspecifed for variety ( fann art, sir
struggle, maa with, zaman time, bir transient, and xlid immortal).
SEGMENT 3: Haqqi (on art)
al-fann
'
ind mawqif-un drmiyy.
"
aql-u
"
anna l-fann s ir maa z-zaman,
"
annahu yurd-u
"
an yatasayyada l-bir li yuqayyidah, s ir
'
-un maa
l-mawt. wa ma
'
a dlik hnam [indam] taqif
"
amma lawha ta
'
ur
"
anna hda l-bir hda l-xlid allad xalladtahu innam huwa bir.
Art for me [is] a dramatic moment. I say that art [is] a struggle with
time, that it wants to capture the transient to preserve it, a struggle with
death. And despite this when [when] you stand in front of a painting
you feel that this transient, this immortal that you immortalized, is only
transient.
Segment 4 difers from segment 3 in that it has some evidence of Egyp-
tian Arabic features from syntax , morphology , lexicon , and pronuncia-
tion patterns: the relative marker illi, the tense/aspect prefx ba-, the
arabic on the media: hybridity and styles 413
negative ma- in malh, the word wayyit, and the pattern in baladna,
not baladina the expected fus h pronunciation .
Te segment also includes a relatively large number of words unspec-
ifed for variety (27), almost equal to those specifed for fus h (26), and
a minimal (6) specifed for Egyptian . Tis segment also includes the
hybrid form /bataammal/, refecting and perhaps also constructing the
Egyptian and day-to-day context of the discourse.
10
SEGMENT 4. Haqqi (his hobby)
min asman il-hadya allati niltuha hdihi s -s ra wa hiya min rasm
akbar musawwir s -s n f dlika l-waqt. il-huwya l-wah da illi xaragt
bha huwyat gami wayyit is yn ma-lh- ayy qma kbra lkin li kull
as ya makna f qalbi. wi bataammal kayfa axtr ihd hdihi l-usa. wa
hun aknu f h la min al-istihz qallan wa urdu an atamid ala ay
fa axt r asan min agr-i mis r min agr il-fallah n f baladna.
One of the most precious gifs that I received is this picture and it is a
painting of one the best painters of China at that time. Te only hobby
that I came out with is the hobby of collecting some canes that dont have
any value but each cane has a place in my heart. And I contemplate how
to choose one of these canes. And here I am being a little sarcastic and I
want to depend on something so I choose a cane from the trees of Egypt
from the trees of the peasants in our country.
Finally, segment 5 of the Haqqi sample includes more instances of lexi-
cal items marked for Egyptian (14) by comparison to previous segments.
Te number is still the lowest by comparison to that of lexical items
marked for the predominant fus h (36 instances) and those unspeci-
fed for variety (25 instances). Segment 5 is interesting from another
perspective: it is the only segment of the three from Haqqi that actually
has one or more sentences predominantly in Egyptian as, for example,
the last sentence. Finally, this segment also illustrates variation in the
use of fus h interdental fricatives versus Egyptian sibilants through the
alternation muwaddaf ~ muwaz z af employee. I have counted the form
with the sibilant /z / as Egyptian since the alternative with the stop pro-
nunciation (*muwaddaf ) does not exist.
10
The hybrid form bataammal in this segment and yumassilu in segment 5 are
excluded from the count in each case for the obvious reason: each has features of both
varieties.
414 mushira eid
SEGMENT 5: Haqqi (on government employees)
f hdihi l-fatra istat ana an nungiz bada l-aml. rubbam l yahiqq-u
l an adkurha lkin l bas min an aql . . . fa knat tagriba arba gid-
dan xussan f l-muwazzafn bi n-nisba li hdihi l-wazrt at -t aqfyya.
anta bayna namat ayn min al-muwazzafn: muwaz z af idri lkin s ifr f
l-fann wa muwazzaf fannn lkin s ifr f l-idra. timsiku min hina yigri min
hina wa il xirih. fa kna l-amal asr giddan. al-h us l ala muwaz z af
yumassilu xayra at -t arafayn an yakna fannnan wa f l-waqt nafsu
yistah mil innu yuud ala maktab wi yira waraa wi yiraf yiktib gawb
kwayyis.
During this period we were able to accomplish some goals. Maybe I
dont have the right to mention them but its ok that I say . . . so it was a
very strange experience especially in the employees of these ministries
of culture. You are between two types of employees: an employee [who
is] administrator but zero in art and an employee [who is] an artist but
zero in administration. You catch him from here, he runs from there and
so on. So work was very difcult: the fnding of an employee who is good
at both, to be an artist and at the same time tolerates to sit at the desk and
read a paper and know how to write a letter well.
Te speech samples from Haqqi demonstrate that in this speech style
features from Egyptian are consistently much lower in frequency than
those from fus h as well as those unspecifed for variety. One sample
(segment 3) has no features of Egyptian and includes almost all the
characteristics of a reading style .
What, one might ask, would explain such diferences? Topic is a very
likely explanation. In segment 4 with the most fus h , Haqqi speaks
about art and what it means to him: a topic that lends itself to fus h but
more importantly, a topic he has written and lectured about through-
out his long career.
11
In segments 5 and 6 the topic shifs to relatively
more personal topics and situations: his hobbies and his experiences as
a government ofcial , respectively. Te style shifs accordingly. Te last
segment includes the most dialectal features, perhaps because it is also
the one most related to everyday events and behaviors as a result of his
comments on government employees. Tis is not to say that it is neces-
sary that Egyptian be used in this case. Other segments in the program
show Haqqi not resorting to Egyptian under similar circumstances. In
fact, these two segments are the only ones in his speech sample that
include any features of Egyptian .
11
In an earlier segment Haqqi talks about the short story and the essay, stressing that
the purpose of the former is to entertain the reader (imt giving pleasure to) and the
latter is to inform (ilm) the reader.
arabic on the media: hybridity and styles 415
Haqqis style can be said to conform to the discourse on diglossia by
relating Arabic variety to topic : fus h for the abstract, informative, and
literary, but Egyptian for everyday situations. Haqqis speech sample
shows that fus h can be used successfully and appropriately in descrip-
tions and narrations about everyday situations, although it would be
odd to use it in actual day-to-day interactions.
A diferent style is represented in segments 68 from the Anis Mansour
interview . Segment 6 consists of a question from the interviewer, Sanaa
Mansour, followed by A. Mansours answer. Te question addresses
the issue of opposition to a mainstream position: What happens when
one goes against the tide by taking an unpopular or oppositional stand.
Mansours answer includes more instances of lexical items unspecifed
for variety (28) than instances of either Egyptian (14) or fus h (15). Tis
segment also shows an almost equal distribution of Egyptian and fus h
lexical items.
SEGMENT 6: A. Mansour (Swimming against the tide)
Question 1. SM:
bi-ylu l yaksab kat ran man yasbah didd it-tayyr. h adritak kisibt
il-kitr walla dayman bi-tisbah didd it-tayyr.
Tey say: He does not gain much who swims against the tide. Have you
[sir] gained much or were you always swimming against the tide?
Answer 1. AM:
da yatawaqqaf ala anni tayyr. ah ynan min al-mas lah a l-ma aw
il-wat aniyya annaki tasbahi didd it-tayyr iza kn it-tayyr inh ill aw
fasd, tafha, adam axz il-ay bi giddiyya. hun yusbih min al-wgib
ala l-ktib innu yasbah didd it-tayyr, innu yimil tayyr mudi liannu
wagbu innu yus lih aw yartaf bi n-ns. fa da yatawaqqaf ala anhi tayyr
illi fh xf aw mah azr innu yusbah didd-u.
Tis depends on which tide. Sometimes it is for the public or national wel-
fare that you swim against the tide, if the tide immorality or corruption,
triviality, not taking matters seriously. Here it becomes the duty of the
writer that he swim against the tide, that he create a counter tide because
it is his duty that he reform or uplif the people up. So this depends on
which tide is fearful to swim against.
In segment 7, which is a follow-up on the question above, there is a clear
shif in style for both interviewer and interviewee. Sanaa Mansours
question here is an excellent example of what I call collage. It consists
of 9 words of which 4 are Egyptian , 3 are unspecifed for variety, and 2
416 mushira eid
are specifed for fus h .
12
She receives a one-word answer in Egyptian ,
which she follows up with a question in clearly marked Egyptian syntax ,
indicated by the WH-question word h what, and unmarked lexi-
cal items: kn and in-natga. Her question generates another collaged
response from A. Mansour, a response that is predominantly Egyptian
(16), minimally fus h (2), with unmarked items in between (7).
SEGMENT 7: Mansour (Swimming against the tide, continued)
Q2. SM: gatlak fatart kat ra f h aytak innak sabah t didd it-tayyr.
Did you have many times in your life that you swam against the
tide?
A2AM: kitr
A lot.
Q3SM: kn in-natga h
What was the result?
A3AM: wala h ga. w-adni id uddmik ahu. w-ixtalaf kasran w
ittafat iktr lkin madm inti muqtania aw amna f daawtik
aw f bi wighit naz arik mayhimmik h ga aw mayhimmin
h ga.
Nothing, and here I am sitting in front of you. I have disagreed
ofen and agreed ofen but as long as you are convinced and hon-
est in your mission or in your point of view, you dont care for
anything or I dont care for anything.
Te last segment 8 from Mansour is also predominantly Egyptian (35
instances), minimally fus h (4), with unmarked items (20) in between.
Had it not been for the last sentence which includes a sentence, or part
thereof, in fus h , this whole segment could have been considered Egyp-
tian . Te unmarked lexical items, although a majority, have little impact
on the reader in terms of identifcation of variety. Te syntax , phonology ,
and lexicon of Egyptian predominate, and the identifcation as Egyptian
goes unchallenged.
SEGMENT 8. Anis Mansour (More about Kamel El Shenawi, ad h kan
z arf How nice he was!)
giddan yani maraf aullik nukatuh. kan yil ala Sawla H igzi allh
yirh amha innaha min adabha -idd ablima tifah id-durg tixabbat alh
il-awwil. masalan kan andi arabiyya z uayyara kida baby Ford w-ana mi
12
The fus h in her first question (segment 6) is the result of a quoted proverbial
statement.
arabic on the media: hybridity and styles 417
saww kwayyis yani fa kn yil h in il-arabiyyt il-karro timi gamb
minni timil voo [li d-darag di] yani fa kull adt Kamel El Shenawi
marah wi xift damm w malib wi tayr f k-kalm lkinnu abb w axx w
umm w nima s -s adq wa laysa lahu nazr f h ayatna s -s ah afyya.
Very much [meaning] I dont know. I tell you his jokes. He used to say about
Salwa Higazi, God have mercy on her, that because of her politeness she
would knock on the drawer frst before opening it up. For example, I used to
have a small car, a Baby Ford, and Im not a good driver [meaning] so he
used to say that karro carts would go beside me and sound voom. [laugh-
ter] to this extent [meaning]. So all his get-togethers were fun, humor, tricks,
and play with words but he was a father, a brother, a mother and the best of
friends and there is no one like him in our journalistic life.
Unlike the two previous segments, fus h here comes at the end as a
sequence, a sentence or part thereof, and is not intermingled with the
earlier narrative . It serves as a conclusion to Mansours narrative on
Kamel El Shenawi and his recollection of his sense of humor. Te con-
clusion, that there is no one like him, is what Mansour asserts and wants
the audience to remember. Te code-switch , or style shif , to fus h helps
him accomplish this goal. Mansour relies on his knowledge of code dif-
ferences to set the conclusion apart from previous narration , thereby
signaling the conclusion and highlighting its content. Te switch, or
shif , is therefore meaningful at this particular moment and within the
context of the text. Tis, one might argue, represents the basic difer-
ence between hybridity and code-switching . Hybridity creates a speech
sample that resembles a collage of elements from one code or another.
In that sense its a global feature of a text. Code-switching as a conversa-
tional and stylistic strategy is meaningful in a specifc context within a
text; in this sense its local within a text.
Te three segments 6 through 8 from Mansour, like those from Haqqi,
show an increasing shif towards Egyptian , which may also be related
to topic . Te shif in topic here is not as strong as it was in the Haqqi
samples. Te diference between segment 6 and 7, for example, is not
so much in the topic as it is in locus . Segment 6 addresses the issue of
swimming against the tide on a general, more conceptual level whereas
the question in 7 shifs the locus to the person, to Mansour himself and
his experiences in swimming against the tide. Te shif to the personal
tends to be accompanied by an increasing shif towards Egyptian . Like-
wise, the last segment (8) is located in the personal : reminiscing about
a friend and sharing jokes he used to tell with the audience to demon-
strate his sense of humor. Both are discourse functions most appropri-
ately conducted in Egyptian .
418 mushira eid
On the basis of these samples, two conversational styles are identifed:
one is fus h -based, the other Egyptian -based. Te two styles incorporate
features from both varieties and include a signifcantly large number of
lexical items that are ambiguous or unspecifed for language variety .
13
Te texts produced are ofen like a collage in which the borders sepa-
rating one variety from the other are ofen blurred making the prod-
uct look more like a hybrid text or discourse rather than one where the
boundaries are maintained and the switch from one to the other variety
is clear. Te speech samples from Mansour fall into the frst category,
those from Haqqi into the second.
Te analysis in this section identifed three styles and illustrates some
of their linguistic and discourse features. In the next section I present
a frequency-based quantitative analysis of speech samples from Haqqi
and Mansour to identify overall characteristics of the two styles they
represent.
2.2 Quantitative Analysis
Te quantitative analysis is based on speech samples of approximately
the frst 10-minutes from each speaker, Haqqi and Mansour. Te results
are presented in Table 3.
Te overall totals at the bottom of the table show clear diferences
between the speakers styles. Haqqis is clearly fus h -based (80%) and
Mansours Egyptian -based (74%). Although the two speakers have clear
preferences for diferent varieties of Arabic , they prefer one to the other
by about the same degree: a ratio for Haqqi of 80% fus h to 20% Egyp-
tian and for Mansour 74% Egyptian to 26% fus h .
Table 3 also provides a breakdown of the distribution of fus h and
Egyptian features according to the selected phonological , syntactic , and
morphological features. Te fgures from Haqqis speech show a bias
towards fus h in all areas, but particularly in the syntactic and morpho-
logical features with 96% and 100%, respectively. Te highest percentage
of Egyptian features in his speech sample comes from phonology (28%).
Of the four phonological variables used in the analysis, the alternation
between fus h interdental fricatives and Egyptian stops and sibilants is
the most signifcant; it is responsible for 85% of the Egyptian phono-
13
Since Schmidt (1974) very few studies, if any, have addressed the role of the lexicon
in Arabic mixed varieties and problems in the identification of lexical items by variety.
arabic on the media: hybridity and styles 419
logical features in Haqqis speech. In addition there is a strong difer-
ence in the distribution of the voiced and voiceless interdental fricatives
by comparison to their corresponding Egyptian pronunciation (stops
and sibilants ). Compare the 29 instances of the fus h voiced fricatives to
the 21 instances of Egyptian pronunciation and the 4 instances of fus h
voiceless fricatives to the 20 instances of their corresponding Egyptian
pronunciation .
A very diferent picture emerges from the analysis of Mansours style .
While Egyptian is clearly the dominant variety for him, based on the
distribution of features in all three components (phonology , syntax , and
morphology ), the contribution from fus h is not as minimal as was the
distribution of features from Egyptian , the less dominant variety , in the
Haqqi sample. Tis is most striking in the morphology , for example,
where the distribution of fus h and Egyptian features in Mansours
sample is almost evenly split (52% to 48%, respectively); in Haqqis
sample morphological features of the less dominant (Egyptian ) variety
Haqqi Mansour
Fus h Egyptian Fus h Egyptian
Phonological
q ~ 70 4 20 61
t ~ t ~ s 4 20 3 16
d
/ d ~ d / d ~ z/z 29 21 2 30
diphthong ~ long vowel 20 3 2 10
TOTALS 123 48 27 117
72% 28% 19% 81%
Syntactic
Relative markers 9 1 0 10
Demonstratives 12 0 4 15
Negatives 6 1 10 20
Complementizer 20 0 5 13
TOTALS 47 2 19 58
96% 4% 25% 75%
Morphological
Passive (u-i, u-a ~ it-) 13 0 9 6
Stem IV : a-f al pattern 13 0 7 9
TOTALS 26 0 16 15
100% 0% 52% 48%
OVERALL TOTALS 196 50 62 181
80% 20% 26% 74%
TABLE 3 STYLESHAQQI AND MANSOUR
420 mushira eid
are nonexistent. Likewise, the distribution of syntactic features in Man-
sours style shows relatively more contribution form the less dominant
( fus h ) with 25%; in Haqqis the less dominant (Egyptian ) contributes
only 4% of the syntactic features. Te phonology in Mansours style is
consistently Egyptian -dominant at 81%, stronger than the phonology of
Haqqis dominant variety at 72%. Te strongest contribution from the
non-dominant fus h in Mansours sample comes from the /q ~ / alter-
nation, with fus h /q/ representing 25% of all instances of this variable.
In Haqqis style the Egyptian sibilant pronunciation of the interdental
fricative accounted for 85% of all Egyptian pronunciation features. A
graphic representation of these distributions is provided in Figure 1.
Te focus of the discussion so far has been on linguistic features (of
phonology, morphology, and syntax) that are diferent in the two variet-
ies, hence serve to distinguish one from the other allowing speakers to
identify a word, a speech segment, or part thereof, as belonging to one
or the other variety. Tere is naturally a signifcant amount of overlap
in the three components. Tis overlap, or shared features, is typically
ignored in discussions of Arabic mixed varieties since similarity can be
assumed in the analysis of same language varieties.
Te lexicon is an area perhaps ignored most in analyses of mixed Ara-
bic varieties, partly because of assumed similarity and partly because
FIGURE 1 STYLESPERCENTAGES OF MIX HAQQI
MANSOUR
100
80
60
40
20
0
Pronology
Syntax
Morphology
Egyptian
28
4
0
Fus h
72
96
100
Fus h
81
75
48
Egyptian
19
25
52
Haqqi Mansour
arabic on the media: hybridity and styles 421
of difculties speakers encounter in categorizing lexical items accord-
ing to variety. Schmidt, for example, argues that speakers of Arabic are
ofen unable to classify lexical items as belonging to one or the other
variety, fus h or Egyptian, and that the difculty may be explained on
the basis of topic or semantic felds (1974, 6076). When speakers were
asked to provide fus h and Egyptian equivalents in four semantic felds
(body parts, foods, arts, and politics), they had more difculty with the
last two than they did with the frst two. Results of this type can have
more than one interpretation. Tey can be taken to show a complex,
integrated, interconnected system whereby some areas of the lexicon
such as arts and politics are shared and others distinctevidence for
complementarity typical of subparts of a whole. Tey can also be inter-
preted as pointing to areas of defciency, or incompleteness, in one or
the other language system and used to argue for the superiority of one
or the other variety. Regardless of interpretation, these subsystems of
Arabic are interconnected in ways that are yet to be discovered and sat-
isfactorily explained.
To get a sense of the distribution of lexical items by variety and the
impact the lexicon may have on identifcation of a base variety, I have
analyzed the lexical items in the 10-minute speech samples of Haqqi
and Mansour based on the presence or absence of clearly marked fea-
tures of either variety. Te absence of such features makes the lexical
item unspecifed for variety, or ambiguous. As explained earlier, hybrid
words would have features from both.
14
Table 4 provides the overall dis-
tribution of lexical items from the 10-minutes speech samples of Haqqi
and Mansour according to the three categories: fus h , Egyptian, and
unspecifed.
Te category unspecifed refers to words (content morphemes ) and
particles (grammatical morphemes )
15
that cannot be distinguished as to
variety and can therefore be considered as ambiguous or shared items.
Te results support the idea of a dominant variety underlying diglossic -
based stylistic variation. Haqqis lexicon is predominantly fus h (74%)
with a smaller 23% of his lexical items unspecifed, leaving a minimal 3%
marked as Egyptian . Mansours lexicon is unexpectedly only 45% marked
as Egyptian ; but he has an almost equal amount (41%) unspecifed and
14
The analysis of hybrid words in the speech samples of Haqqi and Mansour has not
been completed.
15
I adopt McCarthy s and Prince s (1990) definition of a minimal word in Arabic as
consisting of at least two moras.
422 mushira eid
smaller but sizeable percentage (14%) marked as fus h . Te major dif-
ference between the two styles is the proportion of the lexicon allot-
ted to the predominant variety by comparison to the unspecifed lexical
items and to those marked for the non-dominant variety . In Mansours
style the discrepancy between the dominant and non-dominant variety
is not as marked, partly because the contribution from the unspecifed
component of the lexicon is stronger than it is in Haqqis style (41% to
23%, respectively). And while the proportion of lexical items from the
dominant and non-dominant is higher in both styles, the diference is
much larger in Haqqis sample than it is in Mansours, 74% to 3% in the
former and 45% to 14% in the latter. A graphic representation of these
results is provided in Figure 2.
Te graphic view in Figure 2 suggests a relatively more balanced
distribution in Mansours style than in Haqqis. Te graph represent-
ing Haqqis style looks as though one component is suppressed (Egyp-
tian) , another is allowed some presence (the unspecifed), but the third
is asserted as an overpowering presence ( fus h ) . In Mansours there is
not one single category that is overpowering, but there is one ( fus h )
less represented than the othersan unequal participant perhaps.
Te graphic view also suggests that the linguistic boundaries are more
clearly defned in Haqqis style than they are in Mansours. For Haqqi
the domain is no doubt fus h ; Egyptian may creep in only slightly, if
necessary. For Mansour, the domain is more inclusive.
Te ambiguous, or unspecifed, lexicon serves to mediate the difer-
ence between the two varieties. It creates a shared, or an in-between
space, consistent with the two distinct codes. In doing so, it contributes
to this linguistic collage by allowing smooth transitions from one code
to the other, thereby blurring the distinction between the two. Hybrid
forms, I would add, serve a similar purpose: they can be heard and
interpreted as one, the other, or both varieties. Te result is a style that
does not sound too colloquial (dialectal) or too literate ( fus h ) a
balancing act that allows each speaker to accommodate the situation
Haqqi Mansour
Egyptian 25 (3%) 391 (45%)
Fus h 683 (74%) 119 (14%)
Unspecifed 214 (23%) 354 (41%)
TOTAL 922 (100%) 867 (100%)
TABLE 4 THE LEXICON PERCENTAGE OF MIX
arabic on the media: hybridity and styles 423
and create personas and identities that are sufciently separate yet simi-
lar enough to be still viewed as one.
Te question remains as to why these two speakers, both highly edu-
cated, intellectuals and writers, would choose styles so diferent from
each other in what appears to be very similar context: one conform-
ing to a great extent to the predominant (ofcial) discourse of separa-
tion ( fus h for public, formal contexts) and the other nonconforming
through its extensive use of Egyptian. Te answer to this question
requires a more detailed analysis of the interviews themselves includ-
ing contexts and topics of conversation as well as speakers purpose in
communication.
3. Te Interviews
An interview is a conversation or discussion between two participants
with the purpose of gathering information by the interviewer about an
individual, an institution, a topic, or some other issue. Since the inter-
viewee agrees to participate, s/he is expected by the interviewer to be
forthcoming and cooperative in this conversation. Te interviewer,
having initiated the event, is likely to have his/her agenda as well. Both
have something to gain from participation in the event, thus a motive
for a successful performance.
An interview may be private or public and formal or informal,
depending on the purpose of the interview. Participants are expected to
FIGURE 2 THE LEXICON PERCENTAGE OF MIX
80
60
40
20
0
Egyptian
Fus h a
Both
Haqqi
3
74
23
Mansour
45
14
41
424 mushira eid
follow certain rules, usually set ahead of time, regarding procedure and
at times topic as well. An interview on radio or television is in addition a
public performance intended to create the semblance of a conversation
whose purpose is to convey information to the audiences and to enter-
tain them as well. Te ultimate success or failure of such an interview
depends on how the performance is staged and how participants interact.
In this section I describe the two interviews selected for this study
in terms of three components that defne an interview: set-up, infor-
mation, and interaction. I relate these aspects of the interviews to the
styles identifed above and to speakers linguistic choices and purpose in
communication.
3.1 Te interviews compared
Te two interviews are similar in set-up and type of information solic-
ited and conveyed. Tey difer primarily in forms of interaction and
interviewees purpose in conversation, more specifcally in the identi-
ties, or persona, they negotiate and attempt to construct through the
interview. Te interviewees, Yehya Haqqi and Anis Mansour, are both
famous men of letters, very much involved in the literary and cultural
scenes of their time. Yehya Haqqi (19051992) is a literary critic, essay-
ist and short story writer. He had tremendous infuence on several gen-
erations of modern Arab writers and is considered one of the fathers
of modern Arab culture. Many of his works have been translated into
several foreign languages and are being taught in many academic insti-
tutions, especially his masterpiece Qindl Umm Hashim Te Lantern
of Umm Hashim. Anis Mansour is a journalist, an essayist, and a story
writer. He is known for his daily column in al-Ahram mawqif Posi-
tions/Opinions and has written numerous books including his famous
h awla l-lam f 200 yawm Around the world in 200 days. Both inter-
views are incomplete, with the early part missing from each. Te amount
of recorded time available from them, however, is comparable: approxi-
mately the last 3035 minutes of each program.
3.1.1 Set-up
Both interviews take place on-location outside the studio in the homes
of the interviewees. In the Mansour program, the interview is conducted
outdoors by the swimming pool with the interviewer and interviewee
seated across a small table. In the Haqqi program, the interview takes
arabic on the media: hybridity and styles 425
place inside his home. Some scenes are shot in his study, others in his
living room, around the dining room table, or in the entry hall.
3.1.2 Format
Te Mansour interview adopts a question-answer format and is hosted
by Sanaa Mansour, an experienced interviewer and host of numer-
ous programs on the ESC including Good Morning, Egypt.
16
In the
Mansour program she performs the interviewer role, asking the ques-
tions and directing the fow of conversation through topic introduction,
follow-up, commentary, and interruption where necessary. From this
perspective, the format of this interview is the conventional interview-
er-interviewee. But it actually is far from being so, as will be shown
below.
In the Haqqi program there is no interviewer on the screen. Te
questions are not aired, only responses and comments. Te program
also includes other participants: Haqqis daughter Nuha, his French wife,
and two of his colleagues, Fouad Duwara and Mohammad Rumeish; the
former is a critic and the latter a writer, as per information projected on
the screen. Both are friends and colleagues who had worked with Haqqi
during his term as editor of al-Magalla.
Despite the diference in format (one-to-one vs. multiple participa-
tion), one participant in each program performs the role of commenta-
tor and topic introducer. In the Mansour program this role is performed
by the interviewer, Sanaa Mansour, and in the Haqqi program by his
colleague, Fouad Duwara. In performing these roles they read short seg-
ments, thus illustrating the non-conversational reading style.
3.1.3 Textuality
Both programs rely on the visual, for entertainment and for reinforce-
ment of information. Images related to the topics under discussion are
ofen projected on the screen during the conversations. In the Haqqi
program, for example, close ups of paintings appear as Haqqi talks
about gifs he received during his visit to China as director of an Egyp-
tian flm festival and about the signifcance of animals in Chinese art as
illustrated in the paintings. Likewise, pictures of the movie stars Fatin
16
To my knowledge, interviewer and interviewee are not related despite the same
last name.
426 mushira eid
Hamama and Omar Sharif are projected on the screen as Haqqi talks
about Fatin Hamamas popularity among Muslims in China. Pictures of
Yehya Haqqi himself, sometimes a younger Haqqi, also appear on the
screen showing him standing by a car in Paris, for example, as Rumeish
narrates a story about Haqqis trip to Paris to undergo a surgery.
In the Mansour interview, pictures of his books are projected on
the screen when a specifc book is mentioned as happened with h awla
l-lam f 200 yawm (Around the World in 200 Days); likewise images
of newspaper articles or quotes appear on the screen when mentioned
in the conversation. Each segment begins with scenes illustrating the
topic to be discussed with the interviewers voice-over introducing the
segment. To introduce the last segment on the intellectuals and the cafs
they frequented, for example, scenes of Cairo streets with various shops
were shown during the introduction. Later in this segment, scenes of
the Brazilian cofee shop, A. Mansours favorite, were shown when the
interviewer mentioned it in her questions. Pictures of Kamel El Shenawi,
Taha Hussein, Tawfq al-Hakim, and Abbas al-Aqqad also appear at dif-
ferent points during the program as their names are mentioned.
Tis interweaving of materials from diferent text typesvisual, musi-
cal, writtenmakes both interviews multi-textual. It serves to create a
collage bringing into the main text of the interview other sub-texts and
sub-efects that contribute to the voice or voices represented in each
interview and ultimately infuence the style speakers choose to adopt
when speaking in their voice and the voices of others.
3.1.4 Topics
Both interviews can be described as reminiscences, refections on the
lives and times of these two professionals. Consequently the topics in
both interviews revolve around the personal and the professional.
In the 35 minutes available from the Mansour interview the discus-
sion is divided into four distinct segments. Te frst segment is personal
dealing with his views on such matters as having children and the nega-
tive stance he has taken towards women. Te second revolves around
his life experiences as a journalist/writer, including refections on his
relationship with leading intellectuals of his time (Kamel El Shenawi,
Taha Hussein, Abbas al-Aqqad, and Tawfq al-Hakim). Te third is
focused on his hectic (perhaps antagonistic) relationship with Abbas
al-Aqqad, and the fourth on cafs frequented by various intellectuals
including himself, a phenomenon the interviewer refers to as t aqfat
a-ri (Street Culture).
arabic on the media: hybridity and styles 427
Likewise, in the 32 minutes available from the Haqqi program, the
topics are focused on three areas: Haqqis artistic interests and his views
on literature and art (which for him transcend the literary and include
fne arts and music), his work as editor in chief of al-Magalla and as
government ofcial in the Ministry of Culture, and his daily routines
and personal life as father and husband. Tese areas and the specifc
topics they include are not as clearly bounded in the Haqqi program as
they are in the Mansour interview. Segments dealing with the personal
and private are interspersed with those dealing with the professional
and public. Te sequencing in the frst part of the program, for example,
illustrates this point. It starts with Haqqi talking about the short story
and the essay, the former being an art form very much like a painting.
Fouad Duawar follows commenting on Haqqis contribution to the per-
forming and fne arts. Haqqi follows to talk about what art means to
him. Te topic then shifs to the personal through Nuha, Haqqis daugh-
ter, who talks about him as a speaker and a writer. Haqqi follows to talk
about subject matter and description in his work with comments on the
richness of the Arabic language, meaning fus h . A switch to the per-
sonal follows again from his daughter and then his wife, and so on. Te
segments fow one into the other and are not typically introduced with a
reading segment, for example, as is the case in the Mansour program.
Tis diference in topic organization and shif can be attributed to
production decisions on program set-up and format. But it has implica-
tions for the interviews and the speech styles chosen by the speakers. It
may, for example, have reduced, even eliminated, the on-screen interac-
tion among the participants allowing, for example, Haqqi and Duwara
to maintain their predominant fus h style.
17
Te one-to-one set up in the
Mansour program allows for interaction and forms of discourse, such as
interruptions and corrections, that generate spontaneous speech, which
in turn triggers the more casual, conversational style which for Arabic
is dialect- based.
3.1.5 Voice
A major diference between the two interviews involves the issue of
voice: who speaks about what and in whose voice. Despite the difer-
ence in set up noted above, the two interviews are multi-voiced, but in
17
Duwara is included here although his speech style , other than reading, is not dis-
cussed in this article.
428 mushira eid
diferent ways. We hear Haqqi and Mansour speak in their voice, both
physically and metaphorically. Tey speak primarily in the I, not the
we, since neither is there to represent a group or organizational identity.
Mansour tends to represent others more so than Haqqi. When he speaks
in the name of the young generation of writers of his time, for example,
he speaks in the we. Haqqi brings in other fus h voices by remembering
lines of classical Arabic poetry, for example, and reciting them for the
audience on the screen.
Te one-to-one set up in the Mansour interview makes him the only
speaker, the interviewer being the other. Trough these two speakers, we
hear other voices as well. Te interviewer Sanaa Mansour speaks in her
own voice when, for example, she expresses an opinion, makes a com-
ment, or asks a question. But she also brings into the interview Man-
sours voice from the past by quoting his public pronouncements and
published views, at times challenging and questioning them, at other
times allowing him to elaborate and narrate. In so doing, she merges
the past and the present thereby adding to the multi-textual collage of
this program a time-based dimension. Although Anis Mansour speaks
in his voice, he brings in the voices of others through his narration. In
the story-telling segments of the interview, for example, he quotes and
reports the speech of others; we hear them but only through Mansours
voice and his perspective.
Trough the multi-person set up in the Haqqi program, we hear and
see four other people, in addition to Haqqi, speak about his life and his
accomplishments. Haqqis comments, however, are limited to Haqqi the
professional: his views on the arts and his track record as an editor and
government ofcial. Tis public persona is reinforced by his two col-
leagues, who provide their perspectives on Haqqi as a public fgure, a
colleague, and a friend in the profession. Projections of Haqqis private
persona are lef to his wife and daughter. As a result, the program as a
whole maintains the boundaries between the two personas: they remain
mostly separate as they come in diferent voices. Perhaps because of this
separation, segments dealing with the personal/ private in this interview
are interspersed with those dealing with the professional/ public allowing
the two to merge into one unifed whole through program production.
3.2 Projections of Identities
Although the identities projected in these interviews fall into the gen-
eral categories of the personal and the professional, the persona created
arabic on the media: hybridity and styles 429
by the main fgure in each of these programs are very diferent. Tis dif-
ference is communicated through the speakers voice, i.e., the language
that expresses the persona. Te style diferentiation described above can
be viewed as representations of speakers voices. In this section I discuss
in relatively more detail the persona being created in the interviews,
relating it to topic, speakers voices, and the interview as a whole.
Te Haqqi interview maintains the boundaries between the per-
sonal and the professional and, perhaps as a result, between fus h and
Egyptian . Te set-up and production of the program helps maintain
the boundaries by assigning participants in the program diferent roles
in the construction of Haqqis overall Persona . Haqqi himself addresses
only the professional , either as his own choice or as that of the producer;
his fus h -based speech style refects this choice. Others interviewed in
the program speak to the personal as, for example, his daughter Nuha,
or to both the personal and professional as does his colleague Meleish;
their styles vary accordingly and appear to be closer to Mansours than
they are to Haqqis.
Te focus on the professional with Haqqi allows him to express
his views on literature, art, and literary criticism and to refect on his
accomplishments in the public sphere as an intellectual, a government
ofcial , and editor of a literary journal. He expresses his views on the
short story, for example, as an art form whose primary purpose is to
entertain, or give pleasure to, the reader (imt al-qri) as opposed to
the essay whose purpose is to inform (ilm). Te voice we hear is Haqqis
throughout: his views on art, broadly defned to include literature, fne
arts, and performance arts, recollections of his three years administra-
tive experiences as Director of the Arts Department in the Ministry of
Culture, and his role as editor of al-Magalla. His voice is complemented,
and at times reinforced, by other participants . Haqqis refections on his
role as editor of al-Magalla (that he made lots of friends through this
job) are picked up in both Rumeish and Duwaras comments. Tey talk,
for example, about Haqqi as being dedicated to his work and supportive
of other writers. Rumeish recalls his frst meeting with Haqqi to submit
for publication his frst short story, while Duwara comments on Haqqis
dedication as a journal editor and his travels to solicit manuscripts for
the journal.
On the personal , or private , domain Haqqi briefy mentions his
hobby of collecting canes from all over the world. He describes, and
shows the audience, a few of his favorite canes and explains how his
choice is ofen determined by his mood. His conclusion to this segment
430 mushira eid
relates the practice of using canes to the public domain, explaining how
it used to be a practice among ministers and politicians but is now a
thing of the past. All other aspects of his private persona are lef for
others to project. His daugther, Nuha, speaks of Haqqi as a father and a
professional , merging the two identities a little. She gives the audience a
glimpse of Haqqi the writer and the public speaker through her eyes as
his daugther. She describes his eloquence as a speaker and the ease with
which words come out of his mouth when he speaks; but apparently
they dont come as easy when he writes. Nuha describes how more dif-
cult and time-consuming writing is for him, comparing the writing pro-
cess to a complicated child birth (wilda mutaassira). She projects him
as a kind and nurturing father, describing their promenades in Maadi,
where he would urge her to observe the beauty in nature and to listen
to the birds, and their walks in Paris, where they would frequent art
museums. She contrasts his perspective of himself versus her perspec-
tive of him. While Haqqi says he is old at 80, she sees him as a young
man (b) still, more knowledgeable than anyone in her generation. A
more detailed description of his private life is lef to his French wife to
construct for the audience. She speaks in French with a voice over in
Arabic, thus adding one more voice to the many voices in the program.
His wife describes their life together, their walks in Maadi and in Paris,
their visits to museums, and his eating habits (eats very little). As these
life events are narrated, we see the two of them at times sitting at the
dinner table, other times in their living room or in his study, and at
times joined by their daughter. His colleague Rumeish provides a com-
bined personal -professional account, for example, in his narration of
how Haqqi cancelled plans for surgery in Paris to return via Libya to be
with the Egyptian people during the 1967 war.
In the Haqqi program, then, the boundaries between personal
and professional identities are maintained as a result of participants
(assigned) roles in this overall construction of Haqqis identities . Tis
allows Haqqi to speak in his fus h -based style and maintain the public
image he has created for himself throughout his career, a supporter of
the fus h and its heritage. From this perspective the Mansour interview
is very diferent.
Te Mansour program, unlike Haqqis, does not separate the profes-
sional and the personal . Te boundaries between them are fuzzy, or
blurred, making the two identities appear merged into one entity, a col-
lage of the personal and the professional created primarily by Mansour
himself with the help of the interviewer. Since this program is set up
arabic on the media: hybridity and styles 431
with Anis Mansour as the only speaker, it becomes his responsibility to
create this hybrid persona , the merger of the personal and the profes-
sional . Tis Mansour does through a speech style in which the boundar-
ies between the two varieties of Arabic are no longer maintained.
Te interviewer Sanaa Mansour brings up for discussion positions
Anis Mansour has taken in public on certain issues. While these top-
ics all belong to the public and therefore professional domain, almost
all are handled on a personal level as narrative rather than as intellec-
tual debates on a more abstract level. Te segment discussed earlier on
swimming against the tide is one example. Most of the questions are
handled on a concrete, personal experience level allowing Mansour to
engage in the narration of events in his voice but ofen reporting the
voices of other participants in past events.
In the Mansour interview , the personal is also public . In the frst
segment, for example, the interviewer brings up Mansours views on
children, marriage, and women. Most, if not all, these views come as
quotations from, or paraphrases of, Mansours published articles, books,
or other public pronouncements. In the discussion regarding his nega-
tive views on women, the conversation at some point centers around
his wife. Despite his opposition to marriage and his negative views on
women, Mansour is in fact married and his wife is a public fgure. Te
interviewer wonders if his wife is then diferent from all other women?
Likewise in the discussion related to children, Mansour explains that
although he is opposed to having children (and gives his reasons based
partly on the professional ), he does love children: all children in his
family have pictures with Mansour carrying them on his shoulder. Here
the professional justifes the personal and vice versa. Te justifcation he
strongly voices in concluding this segment is based on the professional
and is expressed with a switch to clearly marked fus h syntax and mor-
phology of the embedded that-clause: yihimmini
"
an
"
akna mufak-
kiran aw ktiban aw
"
adban giddan Its very important for me that I
be a thinker or a writer or a man of letters. Tis is the public persona
he projects of himself, but always together with the personal .
Diferences in the interview set-up and the specifc topics covered in
the interviews are partly responsible for the overall persona created of
the main fgures. Te bulk of the responsibility however lies with the
individuals being interviewed. Te persona projected by Mansour is
clearly multi-voiced. He plays many roles and speaks in many voices . He
plays his own person as a writer, an intellectual, and a family person. But
he also plays the roles of others in his narratives : Salwa Hegazi, Kamel
432 mushira eid
Elshenawi, El-Aqqad, etc. In his narratives he assumes their roles by
quoting them, for example, directly or indirectly, by telling their jokes
to illustrate their sense of humor, and by reporting on their accomplish-
ments and modes of interaction as well. Troughout the interview his
voice is concrete, personalized, with a clear sense of humor. As a result,
he engages in diferent forms of discourse, e.g. telling jokes, narration ,
and argumentation/persuasion, and takes on diferent personas , e.g. the
story-teller, the writer, the friend. He projects himself as a person who
values a sense of humor, friendship, nurturing, and being there for oth-
ers; these are the values he likes in others and praises them for having
such values. Haqqi, on the other hand, plays one role: the professional ,
hence public , role of the intellectual, writer-artist, and administrator.
He speaks in one voice , that of this public persona . Tis single voice is
clearly refected in his fus h -based choice and the discourse of explica-
tion and narration . Te personal he leaves to others, to his daughter and
his wife who complement this professional persona with the personal,
Haqqi the man: father, partner, and friend.
On some level, then, the two interviews turn out to be similar in topic ,
voice , textuality , and overall informal setting. But they are actually very
diferent in terms of the roles played by the major fgures. Te overall
purpose in both programs is the creation of a public and private per-
sona for each interviewee. In so doing, the programs inform the audi-
ences about these individuals and entertain them as well. Entertainment
comes partly through the collage efect created during the interview a
collage that extends to the language itself as refected in the diversity of
speech styles and the many voices , images, and persona presented in the
programs, but not necessarily created by the two major fgures. While
Haqqi plays one role and speaks in one voice , for example, the program
itself is multi-voiced. Likewise, while the Mansour program creates
boundaries separating its segments and topics , the speakers manage
to cross these boundaries merging the personal and the professional
through their conversational interaction and stylistic choices.
4. Conclusion: Hybrid people and hybrid contexts produce
hybrid language
Te analysis presented in this article suggests that the media provides
a hybrid context whose purpose is to inform and entertain. Using the
arabic on the media: hybridity and styles 433
interview as a genre that lends itself to this type of hybridity , I have
argued that the language used in this context is also a hybrid produced
by hybrid speakers, the degree of hybridity (mixture) being dependent
on a number of variables including purpose in communication, pro-
jections of identity , and the extent to which a speaker is comfortable
with this hybridity . Te language variety produced in such contexts is
like a collage of features drawn from three linguistic spaces in Arabic :
one unspecifed for variety, one specifed for fus h , and one specifed
for Egyptian . Tis form of the language, like hybrids in general, can be
viewed as being neither fus h nor Egyptian , since it does not conform to
either, or as both, since it includes features of both. I leave this question
open, for who is to say what, for example, a culturally-hybrid , bilingual
person would be? Is an Arab-American, for example, neither Arab nor
American, or is she/he both Arab and American? In concluding this
article I return instead to elaborate on the view of hybridity as a global
feature of text and code-switching as a local feature (meaningful on a
local level).
It is hard, if not impossible, to predict when or why a person would
switch from one language, or variety of a language, to another. A quick
glance at the sample texts included in section 2 provides enough evi-
dence to support this point. Te discussion of these texts also shows that
sometimes the best one can do is identify areas (content, topics , criteria)
where people tend to move from one code to another more extensively
than other times. Te end result is a performance where language variety
matches program goal as set by the satellite channel, speakers purpose,
topic , and intended audiences . At other times, however, it was possible
to identify a reason, stylistic or otherwise, for the movement form one
variety to another (e.g. Mansours segment 8).
I have used the terms hybridity and code-switching to capture this
diference. Others have used code-mixing instead of hybridity leaving
the latter to be inclusive of both (Hinnenkamp 2003). Either way two
levels or types of analysis are acknowledged: local analysis that focuses
on the specifc motives for alternations, providing a rationale of local rel-
evance, and a global analysis that focuses on motives that can be found
in categories such as identity and group membership, providing a ratio-
nale of global relevance to the speakers involved. I have used hybrid-
ity in relation to global analysis to capture this cultural connection and
relevance to speaker. I hope to show in future research that for Ara-
bic , and possibly other alternations across codes of the same language,
434 mushira eid
the product may be better analyzed in terms of hybridity understood,
as suggested in this article, as a collage with fuzzy boundaries across
subspaces.
5. References
Awad , Luws. 1965. Mud akkirt t lib bit a (Memoirs of a Student on a Study Abroad
Mission). Cairo: Muassasat Rz al-Ysuf.
Bakhtin , M.M. 1981. Te Dialogic Imagination. Texas: University of Texas Press.
Bassiouney , Reem. 2006. Functions of Code Switching in Egypt. Evidence from mono-
logues. Leiden and Boston: Brill Academic Publishers.
Boussofara-Omar , Naima. 2006a. Diglossia. In Versteegh, Kees et al., eds. Encyclopedia
of Arabic Language and Linguistics, vol. 1, 629637.
. 2006b. Neither Tird Language nor Middle Varieties but Arabic Diglossic Switch-
ing. Zeitschrif fr Arabische Linguistik/Journal of Arabic Linguistics 45:5580.
. 2004. Diglossia as Zones of Contact in the Media. Al-Arabiyya 37:101130.
Cachia , Pierre. 1990. An Overview of Modern Arabic Literature. Edinburgh: Edinburgh
University Press.
Eid , Mushira. 1982. Te Non-Randomness of Diglossic Variation. Glossa. 16:1.
5484.
. 1988. Principles for Switching Between Standard and Egyptian Arabic. Al-Ara-
biyya 21, 5180.
. 1992. Directionality in Arabic-English Code-Switching. In Aleya Rouchdy, ed.,
Te Arabic Language in America: A Sociolinguistic Study of a Growing Bilingual Com-
munity. Detroit: Wayne State University Press, 5071.
. 2002. Language is a Choice: Variation in Egyptian Womens Written Discourse.
In Aleya Rouchdy, ed., Language Contact and Language Confict in Arabic. New York
and London: Routledge Curzon, 203232.
. 2004. Media Performances as Discourse Events. Arabic Media and Public Appear-
ance Forum, Center for the Advanced Study of Language (CASL), University of Mary-
land, College Park, June 810, 2004.
Haeri , Niloofar. 2003. Sacred Language. Ordinary People. New York: Palgrave Mac-
millan.
Hinnenkamp , Volker. 2003. Mixed Language Varieties of Migrant Adolescents and
the Discourse of Hybridity. Journal of Multilingual and Multicultural Development
24:1&2.
Mejdell , Gunvor. 2006. Code-Switching. In Versteegh, Kees et al., eds. Encyclopedia of
Arabic Language and Linguistics, vol. 1, 414421.
Muarrafa , Mus t af. n.d. Qantara allad kafar. Cairo: Muassasat T ibat al-alwn
al-muttah ida.
Schmidt , Richard. 1974. Sociostylistic Variation in Spoken Egyptian Arabic: A re-exami-
nation of the concept of diglossia. Ph.D. dissertation, Brown University.
Schulz , David. 1981. Diglossia and Variation in Formal Spoken Arabic in Egypt. Ph.D.
dissertation, University of Wisconsin-Madison.
Versteegh , Kees et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics, vol. 1.
Leiden and Boston: Brill Academic Publishers.
THE USE OF MORPHOLOGICAL PATTERNS IN
ARABIC GRAMMARS OF TURKIC
Robert Ermers
1
s-Hertogenbosch
1. Introduction
Te use of meaningful (mor)phonological patterns
2
(wazn , bunya
t
, pl.
awzn, abniya
t
) is very common in Arabic linguistic thinking. In this
contribution we examine the way these patterns were used in Arabic
grammars of Turkic , especially in the works of Ab H ayyn al-Andalus
(d. 745/1345) and Mah md al-Kar (11th century).
2. Patterns in the Arabic linguistic tradition
In the Arabic grammatical tradition, each word is analyzed in terms of a
root that contains a certain number of base radicals . Most roots contain
three radicals , but there are also roots with two-, four or even fve. For exam-
ple the verb karuma he was kind is built upon the three radicals k-r-m.
In lexicographical works k-r-m is ofen placed in the same three-radical
cluster as, say, r-k-m and m-k-r. A much later development is the alpha-
betical order in which k-r-m follows, e.g., k-r-, which starts with Ibn Fris
(d. 390/1000) (cf. Wild 1965, 66). Within a given entry the radical pat-
terns are further ordered according to the vowels they contain.
In order to make the vowels stand out more, the Arab scholars use the
paradigmatical root /f--l/ (cf. Wright 1986), which in our example kar-
uma yields faula . In this case faula is considered the wazn . In the lexi-
cographical sequence the pattern faula follows faala . For Arabic words
this system based on abstract patterns makes much sense. Verbs that
have the same pattern usually possess similar qualities, e.g. transitivity ;
verbs of the type faula are among other things usually intransitive and
1
r.ermers@hccnet.nl, P.O. Box 2176, 5202 CD, s-Hertogenbosch, Netherlands.
2
For the sake of brevity here referred to as morphological patterns .
436 robert ermers
most of them describe a state of mind or a quality (cf. e.g. Wright 1986,
I, 30, see also Versteegh 1992). Verbs with the same verbal pattern have
similar predictable morphological derivations, such as the formation of
the verbal noun mas dar . For verbs of the type faula one possible form
of the mas dar is fala
t
e.g. karma
t
kindness. In regard to nouns of
the same pattern the forming of the plural is predictable in the same
way, the pattern fila
t
may have the plural form fawil, e.g. qida
t
pl.
qawid base. Frb (d. 350/961) writes in his Dwn al-Adab that plu-
ral forms, derived from a pattern which usually yields such plurals , are
not included as entries (I:87). Also more or less predictable is the fact
that the pattern fala
t
may refer to abstract nouns which are not usually
pluralized (similar examples in Irtif I 7397). In this respect the gram-
marians also indicate augmented radicals in Arabic patterns , e.g. af ala,
ifaala, istaf ala, in which -, -t-, and -st- are augmented radicals which
add a specifc meaning to the root .
Another obvious motive for collecting words in the same pattern , is
that this is handy when writing poetry . Words with the same pattern can
easily be used in rhyme schemes (cf. also Wild 1965, 66).
Te use of morphological patterns thus appears to have two basic
objectives. Te frst is to determine which consonants in a given word are
basic and which are augmented , or more precisely: to determine which
ones are the basic radicals in a given word. Tis is for instance important
in establishing a words etymology . In the second place the pattern is
used for illustrating paradigmatic patterns , such as verbal conjugation ,
the building of regular plural forms and some types of declension , in
which consonants (e.g. /w/ in /-uwna/ [m pl] and alif /"/ in /-a"t/ t [f
pl] are assigned special meanings (cf. Versteegh 1985). Te scope of this
article is limited to the frst objective.
For Turkic the general advantage of bringing together words with
similar patterns is evident too, but the morph(ono)logical arguments
do not apply. In Turkic languages meaningful elements usually have the
form of sufxes to a given stem , never infxes or prefxes , e.g. kas cut!,
kas-d he cut (3sg pt), kas-i-d they cut together, and qul slave
qul-lar slaves, qul-juq little slave, qul-juq-lar little slaves, etc. From a
given vowel sequence or consonant pattern no predictions as to plural
and verbal conjugations can be made and the same is true for words that
bear certain sufxes .
3
In Arabic grammars of Turkic languages, some
authors make reference to morphological patterns too.
3
If we disregard the consonant assimilation and vowel harmony, that is.
morphological patterns in arabic grammars of turkic 437
3. Two Arabic treatises on Turkic
In the Arabic linguistic tradition the meaningful patterns are the basis
for the arrangement of lexicographical dictionaries . One important
work in this respect is Frb s (d. 350/961) lexicographical work Dwn
al-Adab (cf. Wild 1965, Haywood 1965). Te pattern is also used in
Arabic grammars of Turkic . Especially, the Dwn Lut at-Turk (hence-
forth Dwn ) compiled by the 11th century scholar Mah md al-Kar
is set up in the same way as Frb s work (cf. Ermers 1999a, 19).
On top of the division in patterns , Kar , like Frb , uses a pri-
mary categorization of eight chapters (each of which is named Kitb
book), the fnal two (i.e. 7 and 8) being additions to Frb s division.
Tis superdivision refects a primary interest in a morphological and
perhaps an orthographical arrangement . One item that is refected in
the headings is the position of the glides , the socalled weak consonants ,
and other non-regular consonants , such as hamza ().
Te division of Dwn is as follows:
1. Kitb al-Hamz (29160): words with initial hamza ,
2. Kitb as-Slim (160406): words having all sound consonants ,
3. Kitb al-Mudaf (406445): words containing a geminate consonant
or two identical consonants ,
4. Kitb al-Mit l (445493): words having an initial weak consonant ,
i.e. y or w,
5. Kitb D awt at -t alt a (493535): words having a medial weak conso-
nant , i.e. y, w or alif,
6. Kitb D awt al-arbaa (535599): words having a fnal weak con-
sonant ,
7. Kitb al-unna (599622): words containing [] or [n],
8. Kitb al-Jam bayn as-skinayn (622638): words containing clusters
of consonantal sounds that do not exist in Arabic , e.g. sirtl- climb
(635), td- turn down (633:15).
In this last chapter of his book Kar must have had in mind the cluster
/-yd t-/ in /tiyd tiy/ he turned down, which includes the sufx -t for 3sg
past tense , which is the way Turkic verbs are given throughout Dwn .
Each chapter is further subdivided in separate sections on nouns and
verbs respectively. Te more than 6700 entries are further collected in
minor paragraphs, ofen under a heading that contains a given morpho-
logical pattern (cf. Dankof and Kelly 19825, Auezova 2005). Kar in
438 robert ermers
total lists some 109 main patterns , some of which contain subdivisions,
in total 143 patterns (see scheme 4 in appendix). Te number of patterns
in Frb s Dwn al-Adab is much higher, about 1677.
Another scholar who deserves mention in this respect is Ab H ayyn
al-Andalus (d. 1345), who in his Kitb al-Idrk li-Lisn al-Atrk lists a
large number of meaningful morphological patterns according to which
in his view Turkic words are being construed. He mentions 91 nominal
patterns and 44 verbal ones (Idrk 101:10104:16), a total of 135. One
reason which makes it interesting to have a close look at Ab H ayyns
views on Turkic is that they can easily be compared to those on Arabic
through his oeuvre on Arabic grammar and Quranic exegesis . One of
his most prominent works on grammar is Irtif ad-D arab min Lisn
al-Arab , whose setup bears a strong resemblance with Idrk . In Irtif
(I:2072) there is a similar, but much more lengthy and elaborate sec-
tion on morphological patterns .
4. Te application of morphological patterns
In his classifcation of verbal and nominal roots (as l ) in Idrk , Ab
H ayyn diferentiates between theoretically possible patterns (qisma
aqliyya) and those that are actually used (masm).
4
In Idrk (101:13) Ab H ayyn writes that for biradical nouns there
are 12 theoretically possible forms, all of which are used. Tese are the
following: (1) fa like san you, (2) f like kim who, (3) fu like yuz face,
(4) fau, (5) faa, (6) fai, (7) fuu, (8) fi, (9) fua, (10) fa, (11) fu, (12)
fui. Obviously these can be divided into four groups of three sequences
of vowel patterns each, while afer some reshufing a more transparent
sequence results:
CVC (1) fa (2) fi (3) fu
CaCV (5) faa (6) fai (4) fau
CuCV (9) fua (12) fui (7) fuu
CiCV (10) fia (8) fii (11) fiu
SCHEME 1 PATTERNS OF BI-RADICAL NOUNS IN IDRK
4
Wild (1965, 37) mentions a similar distinction in Kitb al-Ayn between mustamal
used and muhmal not used, litt. neglected.
morphological patterns in arabic grammars of turkic 439
It is difcult to tell whether Ab H ayyn has a preference for a certain
sequence of the vowels. Te regular sequence, though is a, u, i, since in
Arabic grammar a is considered lighter than u and i. Te frst three
items are listed in the sequence a-i-u, but Ab H ayyn refrains from
holding on to this from the second set onwards.
For nouns
5
the following patterns are given:
CVC (1) fal,
"
ard back
(3) ful,
kurt tree
(2) fil,
s irt back
CVCVC
V (in Arabic
analysis CVCVCVC)
(7) faal,
(9) faal,
(8) faal
(16) fuul,
(17) fual
(18) fial
CVCVC (19) faal,
(21) faul,
(20) fail
(22) fual,
(23) fuul
(25) fial,
(26) fiul,
(24) fiil
CVCC
V (in Arabic
analysis: CVCCVC)
(4) fal,
(6) fal,
(5) fal
(13) ful,
(14) ful,
(15) ful
(10) fil,
(12) fil,
(11) fil
SCHEME 2 PATTERNS OF TRI-RADICAL NOUNS IN IDRK
In scheme 1 and 2 Ab H ayyn describes nouns , the verbs being dealt
with at the end of the section (103:13). Note that in the second listing (cf.
scheme 2) Ab H ayyn is more consequent in holding to the sequence
a-i-u, which seems to evolve, than in the frst. Further, it seems that
some of the patterns are incomplete, for example, the nominal pattern
fu-: there is no form fuil* which completes the set (16, 17). Likewise,
there is only one nominal triradical pattern that starts with f . . . (18),
where one would expect three, the other two being ful* and fal*. In
the third place, there is no noun that is formed according the pattern
fuil* (22,23). In sum, only four patterns appear to be lacking from this
overview. Nevertheless, it is easy to invent more patterns which would
ft in this triradical scheme, e.g. fual*, fual* and fual*.
5
Between parentheses the original sequence in Idrk .
440 robert ermers
For verbs Ab H ayyn gives the following patterns :
Number of radicals fa- fu- fi-
Uni-radical
CV (2) fa (3) fu (1) fi
Bi-radical
CVC (5) fa (4) fu (6) fi
CVCV (8) faa
(13) fai
(12) fua
(7) fuu,
(11) fui
(10) fia
(9) fii
Tri-radical
CVCC (14) fal (16) ful (15) fil
CVCVC (17) faal
(19) fail
(18) faul
(22) fual
(20) fuul
(23) fial
(21) fiil
CVCCV (24) fala
(27) fali
(26) fula
(28) fuli
(25) fila
CVCVCV (32) faala (31) fuila (29) fiala
(30) fiila
Tetra-radical
CVCCVC (35) falal
(34) falil
(33) falul
(37) fulal
(36) fulul
(38) filil
CVCCVCV (39) falula
(40) falala
(43) fulila
CVCVCCV (41) faalla
(42) failla
(44) fualla
(45) fuulla
(46) fiilla
Penta-radical
CVCVCCVC (47) faallal
CVCCVCCV (48) falalla (49) fililla
CVCCVCVC (50) falulal
SCHEME 3 PATTERNS OF TURKIC VERBS IN IDRK
Note that there are no long vowels in the verbs .
5. Primary and augmented radicals
Te rules are, at least so it seems, not applied systematically. In the frst
place augmented radicals are not refected in the pattern and conversely,
radicals are presented in such a way that one thinks that they are con-
sidered augments . Ab H ayyn dedicates a large section to augmented
radicals (Idrk 111:17116:13), in which he incidentally investigates
the etymology of some compounds (Idrk 103:12), but the augmenta-
morphological patterns in arabic grammars of turkic 441
tion is not refected in patterns in the way this is regularly done in the
Arabic linguistic tradition. Te nouns and verbs he cites as belonging
to a given pattern are not analysed further in terms of basic radicals .
For example,
"
us urmaq fart could be analysed as fuul-maq*, since it is
an obvious derivative of the verbal stem
"
usur-. Te same holds for the
noun t aarjuq sack, for which the pattern faal-juq* could be posited
(Idrk 103:2,5). Instead the former is attributed the pattern fuullal, and
the latter faallul, the regular way of dealing with hexa - and heptaradical
words, but neither of which gives any clues as to the status the radicals
of the respective sufxes possess.
Ab H ayyn s formulates his goal in arranging words according to
patterns as follows: It is necessary to study the structures of each of
them, so that the primary radical (al-h arf al-as l) may be distinguished
from the augmented (zid). Only then can the primary radical be com-
pared with the primary radical and the augmented radical with the
augmented radical, (Idrk 104:15, translation Ermers 1999b). In our
introduction we already briefy touched upon the importance of being
able to diferentiate between original radicals and augmented ones, e.g.
in af ala, ifaala, istaf ala, maf ala
t
. Te augmented radicals are called
h urf al-man particles of meaning (cf. Versteegh 1995, 120). One
would expect the Turkic augmented radicals in a similar way to be indi-
cated in the pattern , e.g. faal-dir* causative , faal-il* passive , faal-in*
refexive form.
Kar takes a similar position in his Dwn . Much in the same way
Ab H ayyn does in Idrk he discusses in a separate section the aug-
mented radicals for nouns and verbs (Dwn 1316). In the nouns he
deals with the glides and hamza only, e.g. tar bag,
"
adir stallion,
along with -n, as in bazn /baza"n/ hammer, in which the added
meaning of the consonant alif /"/ is not obvious. Te augments in the
verbs are dealt with in more detail. Kar mentions as many as eleven
augmented consonants, i.e. 1. alif, 2. t, 3. r, 4. s, 5. , 6. q, 7. k, 8. l, 9. n, 10.
l (lm-alif), 11. y, all of which are added because of [certain] mean-
ings ( fa-kull wh idin minh tuzd li-man, Dwn 14:2). Here we
sufce with a few examples, e.g. t in its causative meaning for the tran-
sitivity of the verb (li-t-tadiya
t
al-fl, 14:5) in
"
ari-t-t he dried, which
basic form is (as luhu)
"
ar-d it dried up. Another example is s, which,
according to Kar , expresses the meaning of wishing to carry out
that action/verb (man at-tamman li-iqma
t
d lik al-fl, 14:10 ), e.g.
suv
"
ij-a-sa-d he wished to drink (
"
ij-) water (suv). Te -s- is also used
when the action of the verb is not actual (wa-lam yakun wuq d lika
442 robert ermers
al-fl minhu h aqqa
tan
), e.g.
"
ul mank kul-um-sin-d ([uriya annahu]
yadh aku ilayya) [it seemed as if] he smiled at me (in which kul- means
laugh and kul-um-sin smile).
Even though patterns play a crucial role in Dwn , only very occa-
sionally the augmented radicals are made visible, to the extent that even
the ones dealt with in this section of the work are never referred to.
Te most obvious example of a sufx that is visible, is -d, the 3sg past
tense sufx that Kar adds to verbs , e.g. faal-d (Dwn 305). Te
use of this form is most likely a calque from the Arabic grammatical
tradition, where it is the form that is most unmarked (like karuma, lit.
he was kind), and one could therefore argue that /-diy/ is not really
considered an augmentin any case it is not mentioned in his listing
(Dwn 135).
6
A more interesting example is the pattern finl (see no.
13 in scheme 4, in appendix, Dwn 82), which contains nouns like, e.g.
itind what is being pushed, and aqind (suv) running (water). Based
on this pattern if we frst disregard a in aq-f- stands for , and -- for
t/q, the stems are reconstructed as it- push and aq- run (of water)
respectively, both of which are common in Turkic , with a sufx -nd. But
in regard to the augments itself more can be said.
pattern f i i n l i y
Turkic 1 i t i n d i y
Turkic 2 a q i n d i y
All examples Kar supplies in this subsection end in -ind, which in
spite of the categorization as triradical , suggests that in his view /-l-/,
which then on a metalevel stands for -d-, is part of a triradical root and
that only -n- and - are augments . If /-l-/ is nevertheless considered a
literal part of the augment , he still has to explain under which phono-
logical circumstances /-nl-/ is realised as -nd. In Arabic grammar , the
phonological changes in consonants are typically explained by means of
qalb, the exchange of one consonant for another, based on a set of rules,
rather than insertion of meaningful consonants . In any case, a pattern of
the type fnd* would have been a more logical choice.
Te patterns falal /fal
1
al
2
/ and fanal (Dwn 611) occur in one head-
ing under a section of tetraradical nouns in the book of Nasal Words.
Tis heading introduces nouns like tauj /tankuj/ something which
6
Perhaps for this reason Ab H ayyn does not include -d in his references to Tur-
kic verbs ; he applies the patterns to the stem only, e.g. fu for
"
ub kiss!, fal for t art for
weigh!.
morphological patterns in arabic grammars of turkic 443
rises in the air (EDT
7
[520] te), and qujr /qujnka"r/ ram. In
these nouns , which in both words is represented by /-nk-/, counts for
two consonants, e.g.:
8
pattern 1 f a l
1
a l
2
pattern 2 f a n a l
Turkic t a n k u j
Other arguments for this reasoning are in the frst place the fact that
according to the morphological pattern the augment is /-n-//-k-/
being perhaps more basiceven though Kar does not specify to
which one of both patterns the nouns in fact belong. In the second place
nouns like taik /tankik/ air and saak /sank
1
ak
2
/ cup and suuk /
sunk
1
uk
2
/ bone (Dwn 604) are considered to have a doubled con-
sonant if we take into consideration the directly preceding heading.
On a phonological level this suggests that /-k-/ is considered a separate
consonant , whereas from a morphological point of view it is probably
regarded as more basic than /-n-/. All this points to the assumption that
n and k are regarded as separate consonants .
In spite of this, nouns that are quite comparable to those above, like
taut (Dwn 603) /tankut/ name of a Turkic tribe and suqur /sunk-
qur/ falcon, kaa /kanka/ advice are assigned a triradical pattern of
the type faal, faul, fail.
pattern f a u l
Turkic t a nk u t
In those instances /--/ corresponds with the cluster /-nk-/; this never-
theless appears to be a more general morphology for words with .
In Dwn there are a few instances in which the problem is the other
way around: augments are made explicit when there is no apparent need
for it. Te preceding case of fanal may represent such an instance, since
there is no apparent augmented meaning. A speculation along these
lines is that, perhaps Kar in fact was in doubt as to which pattern
to use here, so he wrote down both. Other examples are the pattern
faland that is applied to such nouns as udrund selected and afdind
7
EDT , here and henceforth, refers to: Etymological Dictionary of Pre-Thirteenth-
Century Turkish (Clauson 1972).
8
Clauson in EDT apparently supposes that, in view of the number of radicals , /-n-/
alone must stand for , whilein his viewa superfluous /-k-/ (for g) is added, which
results in the erroneous form teg.
444 robert ermers
collected (Dwn 84). If we for both nouns take /-l-/ in fal-and as the
fnal consonant of the stem , which seems to be the case, it follows that
the paradigmal sufx or augment must have the form -nd. Te pro-
jected stems then must be udr-* and afd-* respectively.
9
Here we have
a similar problem as in regard to the pattern finl, which we discussed
earlier. If instead Kar considered /-l-/ as the frst consonant of the
sufx , one would have expected him to propose the pattern fa-dand*,
because otherwise the occurrence of -d- and -r- on the position of /-l-/
is not accounted for.
Kar s discussion of the few internal augments in nouns he gives
is interesting too. Here the augments are placed between the radicals of
the stem , e.g. fawal for yuwlij lambswool (Dwn 456), fayal for nouns
like qaymaj lambswool (Dwn 522), both of which are considered
triradical . In these cases /w/ and /y/ respectively again, are presented
as augments , quite comparable to /y/ in fuayl, a regular Arabic pattern
for diminutives , e.g. fulays small coin (< fls). But the meaning of the
inserted glides here is quite unclear. If the comparison to fuayl would
hold, this suggests that these would be etymologically related to forms
like yulij* and qamaj*, which could not be found. In view of the fact
that the augment in these instances does not seem to carry a particu-
lar meaning, one could argue that the pattern faalal, or perhaps faalil
would have served as well, and the question remains why Kar did
not apply them here instead of proposing the insertion of augments .
Tere seems to have been a general degree of confusion as to the
applicability of morphological patterns in other works as well. Talmon
(1997, 172) in this respect notes that an authoritative source like al-Xall
(d. 791/175?) in his Kitb al-Ayn is not very consequent in the use of
patterns either: muat f
t
, which according to one view takes the pat-
tern (wazn ) of mufal
t
, and according to another mufala
t
. A similar
example, still according to Talmon , is andawa
t
, which belongs to two
patterns : fanala
t
and faallwa
t
. Some consonants, especially hamza , alif,
ww, y and nn are sometimes referred to as basic (as l ) and others
as non-as l . In Irtif we came across a similar discrepancy, sanbita
t
a
period of time is assigned the pattern falata
t
, but Ab H ayyn acknowl-
edges that others apply fanala
t
instead: it is said that its pattern is . . .
9
Only in the first case this agrees with EDT [70] d r-, whereas for the second Clau-
son suggests evdin- pluck [7].
morphological patterns in arabic grammars of turkic 445
(wa-qla waznuh), obviously depending on how /-n-/ is interpreted. In
another instance here too the arbitrariness in assigning the patterns is
obvious (Irtif I:33). For example, frsan is attributed the pattern flan
(Irtif I:33), even though /-n/ cannot be considered an augmented con-
sonant . In this way, the use of the pattern does not give any clues for
determining which consonant is augmented . In Frb s Dwn al-Adab
similar examples are found, e.g. anbajn be great which follows the
pattern afaln (Dwn al-Adab I:280). In this example the form with
-n obviously is not a derivative from based on a form (anbaj), without
the sufx . Based on this brief survey we may conclude that Kar s
approach is not very exceptional.
6. Other lacunae in the system
Earlier we mentioned that Kar uses some 109 headings in which
he mentions patterns , sometimes more than one. Te total numbers of
patterns he refers to is 146. Not surprisingly, in Dwn the same pattern
recurs several times through the work. Within the diferent chapters
the same patterns almost inevitably recur. To mention only a few, af l
(see no. 4 in scheme 4 in the appendix) is the same as 14; 2 is identical
with 29 and 51; 5 with 32 and 60, more examples of similar patterns in
scheme 5 in the appendix).
Let us examine two occurrences of the same pattern . Te pattern af l
(see no. 4) refers to
"
arqr mountain sheep (Dwn 71), but it also refers
to a noun like
"
ar fshing hook (see no. 14, Dwn 83).
In book I, Kitb al-Hamz, there is no apparent reason for giving any
pattern with initial hamza /-/. Te chapter heading itself suggests that
the initial hamza /-/ is basic and thus should be represented by /f-/. Te
only reason for quoting hamza could be that it is augmented, i.e. when
it is not a basic consonant . Yet seven (i.e. the items 3, 4, 11, 12, 14, 17
and 23see appendix scheme 4) out of the 24 morphological patterns
begin with -, e.g. (3) af al, (4) af l, uf l, if l, (11) afl, etc. where
one would have expected f- instead. Let us study two examples in more
detail:
A.
"
arqr 1. Book I, words with initial hamza , 2. nouns 3. vocalization
of middle radical (i.e. q), 4. Section on af l (and other patterns )
with a vowel on the middle radical , 5. R.
446 robert ermers
B.
"
ar 1. Book I, words with initial hamza , 2. tetraradical nouns , 3.
pattern af l, 4. a double sequence of the same radical within one
root (i.e. ).
B is explicitly placed under the heading of tetraradical words; it then
follows that A is considered triradical . Apart from being explicitly
quoted in the pattern , the initial hamza in A is also the only reasonable
candidate for being considered an augment , which is in contradiction
with the main chapter heading. It is difcult to speculate on the reason-
ing Kar follows here that would account for these two instances of
analysis.
Tere is also a small number of words that have a double entry in the
work in diferent places. One is kuk light shadow (EDT 753). Appar-
ently Kar has interpreted the pattern of this word in diferent ways:
A. 1. Book II regular words, 2. Nouns , 3. Words with added glide
between the second () and third radical (l), 4. Te pattern fal and
the like, 5. (nouns with) K (Dwn 225:8; Auezova 2005, no. 2606)
B. 1. Book V words with a middle weak consonant, 2. [Nouns ], 3. Sec-
tions on difcult (?) words, 4. Te pattern fal and the like (871),
5. (nouns with) K (Dwn 521:12; Auezova 2005, no. 5500)
If we apply the patterns fal /faa"la"/ (1) and fal /faa"luw/ (2) to
kuk /kuiyka"/, in which /f-/ stands for /k/, /--/ for /--/, etc., the fol-
lowing scheme emerges:
10
pattern 1 f a a '' l a ''
pattern 2 f a a '' l u w
Turkic k u i y k a ''
Te application of either pattern suggests that -y- is not basic, but
inserted as an augment between -- and -k-. Te goal here is of course
obvious: lengthening of /i/, but -y- is still not accounted for as a mean-
ingful augment . B is placed in the main division, the Book of words with
a glide as a middle radical out of these.
11
If -y- is not considered a mean-
ingful augment , it must be an instance of prosodic lenghtening . Tis
same Book starts, as expected, with nouns like th th /ta"h/, but they
10
We will discuss below the fact that the vowels of the schemes do not correspond at
all with those in the Turkic , and that /-"-/ is apparently realized as y, and /-w/ as ".
11
If we correctly interpret d awt at -t alt a
t
, cf. Dwn al-Adab I 76,80.
morphological patterns in arabic grammars of turkic 447
are to our surprise considered biradical (t uniyya). In other words:
the alif is, in spite of the heading, not considered a basic radical , but
apparently a mere instrument for indicating lengthening . Tis suspicion
is reinforced by Kar s following remark elsewhere the lenghtening
consonants may be elided when they are pronounced quickly; in that
case one says tah tah (wa-h urf al-ln suqit a min inda sura
t
an-nut q
bih, fa-yuqlu tah tah, Dwn 439:12). Tus the status of the glide still
remains, at least in these words, uncertain.
Finally there is the status of the vowels within the pattern . Ab H ayyn
is both in Idrk and Irtif quite accurate in his selection of matching
verbs and nouns to a given pattern , and the same holds for Frb in
his Dwn al-Adab . Whenever Kar presents a given pattern in a
heading, he nevertheless lists almost everywhere words which follow
an entirely diferent vowel pattern . To give but a few examples: fal is
applied to yij tree which in fact follows fl, yipr musk in fact fl,
yulr rein in fact ful (see item 60 in the scheme below; Dwn 456:3).
Tese instances are ofen covered with subtitles like f h araktihi literally
in the vowels (also 84:11) afer the given pattern . In this way, patterns
containing /a"/ (indicating ) are also ofen applied to words with /
uw/ or /iy/, e.g. fal for yant response, which in fact has the pattern
fal (Dwn 456:3). It seems that Kar throughout his work uses the
patterns for indicating the mere position of the vowels and glides rather
than as a precise pattern .
In some instances Kar s presentation is in contradiction with the
regular practice. When Kar introduces the pattern faal (cf. Scheme
4, no. 76), explicitly described as with the frst and second consonant
vocalized (muh arraka al-f wa-l-ayn), and further specifed as from
the weak words which contain a w (d awt al-ww) (Dwn 507:6).
However, w is not mentioned at all; Kar exclusively lists nouns of the
type (CC), such as qb /qa"b/ pot, tz /ta"z/ bald (Dwn 509:6). It is
unclear how Kar applies this pattern here, since alif ("), which must
be the second or middle consonant, cannot carry a vowel ; it can only be
preceded by /a/.
7. Te glides
Te status of the glides is an important issue in morphological pat-
terns . Te patterns were a great aid to the Arab grammarians for determin-
ing the status of the glides . In Arabic grammatical theory the glides , i.e.
448 robert ermers
/"/ (alif), /w/ and /y/, are considered consonants . For example, bb
un
door is derived from a projected form /ba
1
wa
2
b
un
/* of which the under-
lying pattern is /fa
1
a
2
l
un
/, by replacement of /w/ by //, which results
in /b
1
aa
2
b
un
/*, and a subsequent elision of /a
2
/: /ba
1
b
un
/. Even though
glides are used as instruments for indicating lengthening of the preced-
ing vowels , they are not considered vowels themselves. In this way, alif
(/"/), whose only function is expressing lengthening of /a/, is a con-
sonant as well. Te glides , like a number of other consonants (e.g. /"/,
/n/, /t/, etc.), can be inserted or prefxed as augments . For example, the
verbal form qtala he battled is, according to the Arabic grammatical
tradition, derived from qatala he killed by inserting an alif /"/ accord-
ing to the pattern /fa"ala/, rather than by lengthening /a/ (cf. Bohas
1982, 168).
Tis does not mean that the concept of lengthening does not exist. In
a practical sense the concept of long vowel does exist in Arabic gram-
matical theory, but in the analysis a sharp distinction is made between
meaningful augments on the one hand, i.e. the insertion of phonemes,
and prosodic lengthening , i.e. lengthening for non-phonemical reasons
(ib) on the other. Or, in other words, the analysis of a given word is
based on the question whether a given long vowel is the result of either
ziyda
t
(meaningful addition, or insertion) or of ib (prosodic length-
ening ). If we think further along the lines of Arabic linguistic reasoning,
there is hardly a plausible reason to give for refecting prosodic length-
ening in orthography . Tis is especially true for lexicography . Prosodic
lengthening typically occurs when the word is put in a context, not in
isolation and it is made explicit in poetry , not in regular prose.
Te question of how to interpret lengthened vowels is also an issue in
our source material on Turkic languages. Ab H ayyn starts his expos
on this problem with the uniradical nouns s water, which he writes as
/s uw/, and y /ya"/ bow, j /jiy/, moist (?). His almost casual remark
that the semiconsonants (h urf al-liyn wa-l-madd) [w], ["] and [y] form
no part of the root (as l , 101, 12), but rather arise from lengthening
of the vowels (nawi an ib al-h arakt) now gains importance. In
the case of s , y and j, Ab H ayyn says, not f /fuw/ or y /fa"/ is
intended, but rather fu+ and fa+, respectively (in which we use + for
prosodic lengthening ). In this respect qis (qis a") short too must be
ranged under the pattern fa+. In spite of all this, the author still pro-
duces the patterns fal for words like barj all, faal for t araz scales
(see the patterns 418 in the listing given above). In these patterns the
glides , which, according to the author, here serve to indicate prosodic
morphological patterns in arabic grammars of turkic 449
lengthening , might as well be mistakenly regarded as a meaningful part
of the pattern . It seems that Ab H ayyn clearly was in doubt as to how
to refect the morphological patterns .
Te same ambivalence towards long vowels and their morphological
classifcation is found in Kar s Dwn . Kar says that in general it
is possible to delete the glides (Dwn , 516:6) as in Arabic , in the way
zin decorate! is formed by eliding /y/ from the form /ziyn/*. Te reason
he says so, no doubt is the fact that in Uygur script vowels are indicated
by means of the glides , without the possibility, as in the Arabic script, to
distinguish between long and short vowels (see Ermers 1999a, 130).
In regard of the previous one is inclined to think that vowel length is
unlikely to be considered a phoneme in Kar s lexicon . Tis is not the
case. Kar takes either the long form or the short as an entry . In some
instances he gives the other as a dialectal exception to the general rule.
In other words, he deals with the short and long forms in separate para-
digmal forms . A nice illustration is the pair
"
aj- open (Dwn 92:4) and
"
aj- be hungry (Dwn 95:17). Each of them has a separate entry in the
chapter on the biradical verbs where // counts as the frst consonant
but only
"
aj- be hungry is dealt with in the subdivision (lit.) Insufcient
(al-manqs ). A similar example is
"
z, which has a number of meanings,
e.g. fat; deaf; self; valley; heart (Dwn 35:1),
"
z fat (Dwn 30:6) and
the verbs
"
z- cut (Dwn 93:1), and
"
z- pass frst (Dwn 96:11), which
is also listed under the subheading al-manqs . In the case of
"
it/
"
t push/
bless (Dwn 95:10), which occur under the same pattern , lengthening
may have been used to indicate a diferent pronunciation altogether, e.g.
t- (cf. EDT 36). Even though of one and the same entry more than one
version does occur throughout the work, from the separate treatment it
is evident that for Kar length is a phoneme. One could even make
the assumption that the fact that length can be made explicit in Arabic
script gave Kar the opportunity to arrange his lexicon in a diferent
way than he could have in the Uygur script.
8. Summary
In this article we made a brief inventory of the way Arabic grammar-
ians applied morphological patterns to Turkic words. In Arabic linguis-
tic thinking, the patterns are a convenient instrument for indicating
which consonants in a word are basic and which are not. Non-basic
450 robert ermers
consonants are usually attributed a special meaning, such as, e.g., causa-
tivity or diminutive , which is added to the root . Furthermore, based
on the pattern words that contain glides , such as bb
un
can easily be
reconstructed in terms of regular patterns (*/bawab
un
/). An additional
motivation for lexicographical interests appears to have been bringing
together words with the same or a similar pattern . Tis comes in handy
for writing poetry , since words with the same pattern can be used in the
same rhyme scheme . Surprisingly, the two sources we examined, Diwn
Lut at-Turk (11th century) and Kitb al-Idrk li-Lisn al-Atrk (14th
century) hardly use Turkic meaningful consonants and sufxes in com-
bination with the patterns , even though the authors are clearly aware of
them. Tis appears to be a general feature of lexicographical works: the
patterns seem to be applied more or less arbitrarily. In Dwn Kar
goes one step further, in that he roughly indicates the position of the
vowels in a given pattern by means of /a/, while at the same time in those
instances he deals with words with u, and i, .
9. Appendix
9.1 An overview of the morphological patterns in Dwn Lut at-Turk
Scheme 4 below contains the patterns mentioned in the chapter and sec-
tion headings in Dwn . Sometimes more than one morphological pat-
tern is mentioned (e.g. in 1, 2, 4, 9, 28, etc.). Te roman numerals refer
to the number of radicals mentioned in the closest preceding heading.
morphological patterns in arabic grammars of turkic 451
Book I, Kitb al-Hamz (29160): words with initial hamza
Nouns (Dwn 33) III (1) fal, ful, fil, (2) faal, faul, fail, (3) afal, (4) afl, ufl,
ifl, (5) fal, fal, (6) fal, fal, (7) fal, (8) fuul, (9)
fainl, (10) fal, ful, fil, (11) afl, (12) af aln,
(13) fiinl
IV (14) af l, (15) fal, (16) faall, (17) aflil, (18) fall,
(19) faland, (20) fuln,
V (21) faalal, (22) faaln, (23) afalal, (24) fulil, (25) falil
(Dwn 91)
Verbs (26) faal (Dwn 97), (27) falal-d
Book II, Kitb as-Slim (160406): words having all sound consonants
Nouns (Dwn 172) (28) fal, ful, fil, (29) faal, faul, fail, (30) fil, (31) fl,
(32) fal, fal, fal, (33) fal, fal, fal, (34) faln, fuln,
filn, (35) fal, (36) faln, faln, (37) faanl, (38)
fulul, (39) falal, fall, fulal,
V? (40) falil, V? (41) faalal, faall, (42) fulul, (43) fulund,
(44) faalal, (45) faalln,
VI (46) falall, (47) faalalal
Verbs (Dwn 305) III (48) faal-d
IV (49) falal-d
V (50) faalal-d
Book III, Kitb al-Mudaf (406445): words containing a geminate or two identical
consonants
Nouns III (51) faal
IV (52) fall
V (53) faalal
Verbs III (54) faal-d (Dwn 415)
IV (55) falan-d
Book IV, Kitb al-Mit l (445493): words with an initial weak consonant , y or w
Nouns III? (56) fal, ful, fil (Dwn 447), (57) faal, faul, fail, (58)
falal, (59) fawal, (60) fal, (61) fal, (62) fal, (63) faln,
(64) faln, (65) faanl
IV (66) falal, fall, (67) falil, (68) faall
V (69) faalal, (70) faaln, (71) falalin (Dwn 468)
Verbs (72) faal-d (Dwn 473)
IV (73) falal-d
V (74) faalal-d (Dwn 491)
Book V, Kitb D awt at -Talt a (493535): words with a medial weak consonant , i.e. y,
w or alif
Nouns (75) fal, ful, fil (Dwn 494), (76) faal, (77) fil, (78) fal,
(79)
fal, (80) fal, (81) faln, (82) fayal, (83) falil
V (84) faalal (Dwn 523)
Verbs (85) faal-d (Dwn 526), (86) falal-d (Dwn 529)
Book VI, Kitb D awt al-Arbaa (535599): words having a final weak consonant
Nouns (87) faal, faul, fail (Dwn 540), (88) fil, (89) fal, (90) fal,
(91) fall
VI (92) falal, (93) faalll (Dwn 552)
Verbs III (94) faal-d (Dwn 565)
IV (95) falal-d, (96) fal-d, fal-d, fal-d
V (97) faal-d, V (98) falal-d (Dwn 597)
452 robert ermers
Book VII, Kitb al-unna (599622): words containing [] or [n]
Nouns (99) faal, faul, fail (Dwn 602) (100) fal; (101) fal,
(102) fal
IV (103) falal, fanal, (104) falal,
IV (105) faanll (Dwn 613)
Verbs (106) faal-d (Dwn 615)
IV (107) falal-d, (108) fall-d, (109) fal-d (Dwn 619)
Book VIII, Kitb al-Jam bayn as-skinayn (622638): words containing clusters of
consonantal sounds, does not refer to morphological patterns .
SCHEME 4 MORPHOLOGICAL PATTERNS EXPLICITLY
MENTIONED IN DWN
Te patterns mentioned above recur in diferent places throughout the
work. Te scheme below indicates which patterns are repeated:
pattern occurs in items no. in scheme 4
1 af l 4, 14
2 faal 2, 29, 51, 57, 76, 87, 99
3 fal 5, 32, 60, 78, 89, 100
4 fal 35, 96 ww
5 falil 40, 67, 83
6 fal 10, 80, 102
7 faalal 21, 41, 44, 53, 69, 84
8 faalal-d 50, 74
9 faaln 22, 70
10 faal-d 48, 54, 72, 85, 94, 106
11 faall 41, 68
12 fil 30, 77, 88
13 fal, ful, fil 1, 28, 56, 75
18 fal 6, 33, 61, 79, 90, 101
16 faln 34, 63, 81
17 faland 19 (noun ), 55 (verb )
14 falal 35, 58, 66, 103
19 fall 39, 52, 66, 91, 108 (verb )
15 falald 27, 49, 73, 86, 95, 107
20 fulul 38, 42
SCHEME 5 DOUBLE OCCURRENCES OF MORPHOLOGICAL
PATTERNS IN IN DWN
morphological patterns in arabic grammars of turkic 453
10. References
10.1 Primary sources
al-Andalus , Ab H ayyn, Muh ammad b. Ysuf. Irtif ad-D arab min Lisn al-Arab. (3
vol.) Mus t af Ah mad an-Namms, ed. 198419871989. (I: Cairo: Mat baat an-Nar
ad-D ahab; II/III: Cairo: Mat baa al-Madan).
Kitb al-Idrk li-Lisn al-Atrk. Amet Caferolu, ed. 1931. Istanbul: Evkaf
Matbaas.
al-Frb , Ab Ibrhm Ish q b. Ibrhm. Dwn al-Adab. Ah mad Muxtr Umar, ed.
197478 (4 vol.) Cairo: al-Haya al-mma.
al-Kar , Mah md b. al-H usayn b. Muh ammad. Diwn Lut at-Turk. Facsimile edi-
tion of the MS; Ankara: Turkish Ministry of Culture. 1990.
al-Xall = Ab Abd ar-Rah mn al-Xall b. Ah mad al-Farhid. Kitb al-Ayn. al-Maxzm
and as-Smarr, ed. Beirut: al-Alam.
10.2 Secondary sources
Auezova , Zifa. M. 2005. Diwan Lughat at-Turk. Translated (into Russian), with intro-
duction by Z.-A. Auezova; indices by R. Ermers. Almaty: Daik Press.
Bohas , Georges. 1982. Contribution ltude de la mthode des grammairiens arabes en
morphologie et en phonologie daprs grammairiens tardifs. Tse Universit de Paris
III (1979).
Clauson , Gerard. 1972. EDT = An Etymological Dictionary of Pre-Tirteenth-Century
Turkish. Oxford.
Dankof , Robert and Kelly , James. 19825. Compendium of the Turkic Dialects. Trans-
lation of Mah mud al-Kars Dwn Lut at-Turk. 3 Vol.: I (1982); II (1984); III
(1985) indices to Vol. I and II. Harvard: University Press.
Ermers, Robert. 1999a. Arabic Grammars of Turkic. Te Arabic Linguistic Model Applied
to Foreign Languages. Leiden: E.J. Brill.
. 1999b. Translation of Ab H ayyn al-Andaluss Kitb al-Idrk li-Lisn al-Atrk.
Leiden: E.J. Brill.
Haywood, John 1965. Arabic Lexicography. Leiden: E.J. Brill.
Owens , Jonathan. 1988. Te Foundations of Grammar: An introduction to medieval Ara-
bic grammatical theory. Amsterdam: Benjamins
Talmon , Rafael. 1997. Arabic Grammar in its Formative Age. Kitb al-Ayn and its attri-
bution to H all b. Ah mad. Leiden, New York, Kln: Brill.
Versteegh , Kees. 1985. Te development of argumentation in Arabic grammar: Te
declension of the dual and the plural. Studies in the History of Arabic Grammar, ed. by
Hartmut Bobzin and Kees Versteegh, 152173. Wiesbaden: O. Harrassowitz.
. 1992. Grammar and rhetoric: Jurjn on the verbs of admiration. Studies in Semitic
Linguistics in honor of Joshua Blau, M. Bar-Asher et al., eds. Jerusalem: Hebrew Uni-
versity of Jerusalem. 113133.
. 1995. Te explanation of Linguistic Causes. Az-Zajjjs theory of grammar. Intro-
duction, translation, commentary. Amsterdam: Benjamins.
Wild , Stefan. 1965. Das Kitb al-Ain und die arabische Lexikographie. Wiesbaden: Har-
rassowitz.
Wright , William. 1986. A Grammar of the Arabic Language. Translated from the German
of Caspari. Cambridge: University Press. (2 vols) (First edited in 1859/1862).
LEXICAL GAPS IN ARABIC LEXICOGRAPHY WITH
EVIDENCE FROM ARABIC DICTIONARIES
Jan Hoogland
Nijmegen University
1. Introduction
Between 1997 and 2003 we compiled with a dedicated group of specialists
at the Arabic department of the University of Nijmegen a twin set of
Arabic-Dutch and Dutch-Arabic dictionaries . Kees Versteegh , Manfred
Woidich , and myself were the responsible editors.
1
It may be clear, that the compilation of these dictionaries not only
resulted in their publication, but also in the gathering of a lot of valuable
experience leading to some striking conclusions. One of these was the
observation that, in the Dutch-Arabic volume, many source (Dutch )
language units could not be paired with an equivalent in Arabic , since
they appeared to be non-existent in Arabic . As a matter of fact, almost
25% of all Dutch entries (words or expressions ) could only be translated
by means of paraphrases . Paraphrased descriptions are, in most cases,
explanations and not direct equivalents of the source language word or
expression .
From this observation the question did arise: do these untranslatable
units represent lexical gaps in Arabic ? In other words: do they represent
concepts without any lexicographically acceptable one-to-one translation
in Arabic ?
It goes without saying that all dictionary compilers also beneft from
the work of others. Also the authors of the Nijmegen dictionary referred
to a large number of existing monolingual, bilingual and multilingual
dictionaries in order to single out adequate equivalents for the Dutch
entries selected to occur in the dictionary . Too ofen we concluded
that the other dictionaries also contained paraphrased descriptions or
defnitions instead of one-to-one translations .
1
Te whole process of compiling has been described on the project site (www.let.
ru.nl/wba).
456 jan hoogland
Upon this observation we came to the following hypothesis: when
a certain entry is translated into Arabic by means of a paraphrased
description or defnition while all, or almost all, bilingual dictionaries ,
included in this enquiry, translate the same entry in the form of a
description or defnition , it is justifed to conclude, that the entry
concerned does not have a one-to-one equivalent in Arabic , and that we,
therefore, are dealing with a lexical gap in Arabic . In order to verify this
hypothesis, a large sample of entries from the Dutch-Arabic dictionary ,
with one-to-one source equivalents in various other dictionaries , was
gathered to compare their translations into Arabic .
Te source language units can be divided in simple words (lexical units ),
compound words and expressions . Compounds in Dutch are simply
written together as one word and consequently have to be entered as
independent lemmas in the dictionary .
Tis article limits a comparison of translations for simple words
from various bilingual dictionaries , since it turned out difcult to fnd
unambiguous translations in English , French or German for Dutch
compounds . Te lack of those translations as a starting point for a
comparison of Arabic translations would result in data that would be
difcult to compare with each other.
2. Defnition of description
As stated in the introduction, descriptions are, in most cases, not direct
equivalents of the word or expression in the source language , but
explanations, i.e. expressions of more than one word which provide
the dictionary user with an explanation in the target language . Since a
description is a combination that is not lexicalized, it is not entered in
the database as an expression in the target language .
Descriptions represent a unidirectional translation relation, i.e. they
are not included in the reversion process to produce the reverse part
of the dictionary . Assuming the description describes a concept that
represents a lexical gap in the target language , it is obvious this concept
could not be entered in the reverse part of the dictionary as an entry .
First of all, a sample of 25 Dutch simple words translated with
descriptions was gathered. Tese words were chosen randomly from the
underlying database. Tese words can be found in Table 1 below.
lexical gaps in arabic: evidence from dictionaries 457
In order to be able to look up these words in English-Arabic, French-
Arabic and German-Arabic dictionaries it was necessary to obtain
translations of the Dutch words from the sample in these languages.
Te table below shows these 25 Dutch words with translations in
three western languages. Some Dutch words which were candidates
to be included in the sample, turned out to be untranslatable in one
or more western languages. Te absence of an obvious translational
equivalent in one of these languages of course would cause problems
in using that language as source language for the comparison of Arabic
translations . For this reason, some randomly selected Dutch words,
translated with descriptions in the Nijmegen Dutch-Arabic dictionary
were not included in the sample.
TABLE 1 ENGLISH, FRENCH AND GERMAN TRANSLATIONS
OF THE SAMPLE
Dutch English French German
bestek cutlery couvert Besteck
francofel francophile francophile frankophil
frankeren to frank afranchir frankieren
freak freak fana Freak
fresco fresco fresque Fresco
gletsjer glacier glacier Gletscher
glibberen slither glisser glitschen,
schlittern, rutschen
gnifelen snigger rire sous cape, rire
tout bas, rire dans
sa barbe
schmunzeln, in
sich hinein lcheln
log guest hte Gast, Logiergast,
Logierbesuch
logement lodging (house) auberge Gasthaus
lommerd pawnshop mont-de-pit Leihaus, Leihamt
loods pilot pilote, lamaneur Lotse
loops in heat/season en chaleur, en rut brnstig
lotgenoot partner (in
misfortune /
adversity)
compagnon,
compagne
dinfortune
Leidensgenosse,
Schicksalsgenosse
ouderwets old-fashioned vieux jeu, dmod,
pass de mode,
dsuet, surann,
archaque, prim
altmodisch
ouvreuse usherette ouvreuse Platzanweiserin
ouwel wafer pain ( m. ) azyme,
dautel, hostie non
consacre
Oblate
458 jan hoogland
ribbel rib, cord cte Rippe
rif reef rcif Rif
riposteren riposte riposter ripostieren
roe rod verge Rute
rog ray raie Rochen
trampoline trampoline trampoline Trampoline
transcriberen transcribe transcrire transkribieren
traumatisch traumatic traumatique traumatisch
If the assumption that the concepts behind these words represent
lexical gaps in Arabic is correct, at least a part of these words have to
be translated with descriptions in other dictionaries too, if they were
included in these dictionaries at all. So in order to investigate this, the
translations of these words were checked in a number of dictionaries .
For this comparison the following dictionaries were consulted:
Al Mawrid English-Arabic
Oxford, Doniach English-Arabic
El Mounged English-Arabic
Al-Kamel Al-Kabir French-Arabic
Al Manhal French-Arabic
Schregle German-Arabic
Van Mol Dutch-Arabic
However, before presenting the results of this comparison of the seven
dictionaries mentioned, it seems useful to present the Arabic translations
from the Nijmegen dictionaries here frst. Obviously, these should not
be part of the comparison, since the lack of translational equivalents for
these words in the Nijmegen Dutch-Arabic dictionary were the starting
point for the comparison. Inclusion of these translations would distort
the results of the comparison.
Tables 3a and 3b below contain the Arabic translations of the Dutch
words from the seven dictionaries introduced above.
Dutch English French German
TABLE 1 (CONT.)
lexical gaps in arabic: evidence from dictionaries 459
DU_LU Eng_LU DA_Nijmegen
bestek cutlery
francofel francophile
frankeren to frank
freak freak
fresco fresco
gletsjer glacier (
glibberen slither
gnifelen snigger
log guest (
loods pilot (nautic)
loops in heat/season ,
)
lotgenoot partner (in
misfortune/adversity
ouderwets old-fashioned
ouvreuse usherette (
ouwel wafer
rif reef
riposteren riposte
roe rod
rog ray
trampoline trampoline
transcriberen transcribe
traumatisch traumatic
TABLE 2 ARABIC TRANSLATIONS FROM THE NIJMEGEN
DUTCH-ARABIC DICTIONARY
460 jan hoogland
TABLE 3A ARABIC TRANSLATIONS FROM 3 OF THE
7 OTHER DICTIONARIES
DU_LU Eng_LU EA_Mawrid EA_Oxford EA_Mounged
bestek cutlery :
francofel francophile
()
frankeren to frank
( )
() )
freak freak
fresco fresco )
(
gletsjer glacier
glibberen slither
gnifelen snigger )
(
-)
(
log guest ( )
logement lodging
(house)
: :
)
(
)
(
lommerd pawn shop :
)
(
lexical gaps in arabic: evidence from dictionaries 461
TABLE 3A (CONT.)
DU_LU Eng_LU EA_Mawrid EA_Oxford EA_Mounged
loods pilot
(nautic)
)
(
( )
loops in heat/
season
:
()
()
lotgenoot partner (in
misfortune/
adversity)
ouderwets old-
fashioned
:2
()
()
)
(
) )
ouvreuse usherette :
( )
ouwel wafer
( )
roe rod
rog ray
trampoline trampoline
)
(
)
(
462 jan hoogland
TABLE 3B ARABIC TRANSLATIONS FROM THE REMAINING 4
OTHER DICTIONARIES
TABLE 3A (CONT.)
DU_LU Eng_LU EA_Mawrid EA_Oxford EA_Mounged
transcriberen transcribe
)
(
traumatisch traumatic
DU_LU Eng_LU FA_Kamil FA_Manhal GA_Schregle DA_Mol
bestek cutlery )
(
francofel francophile
frankeren to frank
)
(
freak freak
fresco fresco
gletsjer glacier )
(
)
(
glibberen slither
lexical gaps in arabic: evidence from dictionaries 463
TABLE 3B (CONT.)
DU_LU Eng_LU FA_Kamil FA_Manhal GA_Schregle DA_Mol
gnifelen snigger
log guest
logement lodging
(house)
loods pilot
(nautic)
loops in heat/
season
)
(
()
ouderwets old-
fashioned
)
(
ouvreuse usherette :
) *
(
ouwel wafer
()
)
(
464 jan hoogland
TABLE 3B (CONT.)
DU_LU Eng_LU FA_Kamil FA_Manhal GA_Schregle DA_Mol
riposteren riposte
()
(
)
roe rod (
rog ray
)
(. . .
trampoline trampoline
transcriberen transcribe
traumatisch traumatic
So, in order to verify if the concepts behind these Dutch words do
represent lexical gaps in Arabic , it is useful to compare and to evaluate
the translations presented in these other dictionaries that provided the
same units with a description , if the word in question is included in the
other dictionary at all.
For all translations obtained from the 7 dictionaries mentioned above,
information was entered if the dictionary in question has included the
Dutch word or its English / French / German equivalent as an entry , and
if so, if this translation is a description . For this an evaluation system was
devised. All translations are evaluated with a code from the following
code system:
0 = entry not included
1 = entry is included, translation is an equivalent (not a description)
2 = entry is included, translation is description
3 = entry is included, translation is a neologism
4 = entry is included, translation is a hyperonym
5 = entry is included, translation is inaccurate or translation belongs to
a diferent or more specifc meaning
lexical gaps in arabic: evidence from dictionaries 465
6 = entry is included, translation consists of a neologism and a
description
Some remarks need to be made in relation to these values. For the sake
of clarity examples for each of these 6 categories will be presented. Tese
examples are taken from Table 3 a and 3b above.
Value 0 is obvious. If a word is not included in a dictionary , the value for
this word is 0. For example: glibberen (to slither) is not included as an
entry in D-A Van Mol , nor in G-A Schregle .
All other values (16) imply that the word occurs as an entry in the
source language of the dictionary heading the column.
Value 1 implies the translation is a translation equivalent . For example:
glibberen / to slither is included and its translation seems to be a real
equivalent in E-A Oxford ( ), E-A Mounged ( ),
and F-A Kamil ( ).
Value 2 refers to a description . For example: glibberen / to slither
is included, and its translation is a description in E-A Mawrid (
).
Tis applies, frst of all, to the real explanations which can be found,
for example, in Oxford English-Arabic at bestek / cutlery (
) or logement / lodging ( )
( ). However, one might as well defne as descriptions
those combinations of two words that are not frequent. If we look at
the translations for francofel / francophile for example, we fnd the
translation in both F-A Al Kamel and F-A Al Manhal . Tis is
a combination of words that is not lexicalized, and therefore should not
be entered in the database as an expression .
Value 3 refers to a neologism . For example: lommerd / pawnshop is
translated with a neologism in F-A Kamel ( ). Some dictionaries
ofer the user made up words in case of a lexical gap in Arabic . In Al
Manhal French-Arabic these neologisms are marked with an asterisk
466 jan hoogland
and an explanation comes with the neologism . See the examples of value
6 below.
Value 4 represents a hyperonym translation, i.e. the word in the target
language is actually the translation of a hyperonym of the source language
word. For example: transcriberen / to transcribe is translated as
in several dictionaries . Tese translations do not express the
fact that to transcribe means decoding from one system and encoding
in another system. Terefore these translations express a more general
meaning and not the specifc meaning of the source language word.
Value 5 as an evaluation means that the translation can be inaccurate or
wrong, or that it refers to a diferent, more specifc meaning of the source
language word. For example: rif / reef is included, but its translation is
inaccurate in G-A Schregle ( ).
Value 6, fnally, is a combination of 2 and 3, since it consists of a
neologism , with a description (mostly between brackets) added to it.
For example: gletsjer / glacier is included, but its translation consists
of a neologism and a description in F-A Manhal (
) and F-A Kamil ( )
).
Before presenting all evaluations in a table, it is useful to introduce
the concept translation profle . Tis profle is a compilation of al the
evaluations, indicating the number of dictionaries in which a translation
was found, followed by an enumeration of all the evaluations. Tis profle
gives a quick indication of the number and types of translations found
in the various dictionaries .
For example 3:223 means a word was found in three dictionaries , and
in two of these dictionaries the word was translated with a description
(code 2), and in one with a neologism (code 3).
Below is a table containing columns with the Dutch words, the English
translations and 7 columns for all 7 sample dictionaries , and a fnal
evaluating column containing the translation profle of the concept.
3. Interpretation of Table 4
lexical gaps in arabic: evidence from dictionaries 467
Te translation profles of the words should indicate if these words
represent concepts that can be considered lexical gaps in Arabic . Te
higher the number of dictionaries in which these words were included,
and the higher the number of description translations or neologism
translations , the higher is the probability that the concept represents a
lexical gap in Arabic. So, profles containing value 2 (description ) and
value 6 (neologism + description ) are the clearest indications.
Seven Dutch words have been included in all 7 dictionaries . Tese
words are: bestek / cutlery, gletsjer / glacier, logement / lodging,
loods / pilot, ouderwets / old-fashioned, rif / reef , and transcriberen
/ transcribe. A more thorough look at the translation profles of these
words is useful in order to determine if the coding system seems reliable.
1. Te word bestek / cutlery has the following profle 7: 2222222. It
seems justifed to conclude this word/concept indeed represents a
lexical gap in Arabic since there is unanimity among all dictionaries
in this, and probably no other dictionary or other reference work
will mention an Arabic word for this concept. Obviously, we do not
have to go very deep into Arab or Islamic culture to conclude tat this
lexical gap can be explained by the fact that in the Arab world people
were and are still used to eat with their hands.
2. Te word ouderwets / old-fashioned has the same profle :
7: 2222222. So, obviously this concept also represents a lexical gap
in Arabic .
3. Te word loods / pilot has the following profle : 7: 1111222. Te
Dutch word loods, as a person, is monoseme and can only mean
a pilot in shipping, and certainly not airplane pilot or any other
type of guide. Te translation is correct but requires a specifc
context in order to distinguish this specifc meaning from other more
general meanings of the word .
4. Te word transcriberen / transcribe has the following profle :
7: 2222444. As indicated above, there are presented description
translations and hyperonym translations .
5. Te word rif / reef has the following profle 7: 2255666. Tree
dictionaries have the word as a neologism , in all cases followed
or preceded by a description . However, this word was not found with
this meaning in the text corpus or on the internet.
6. Te word gletsjer / glacier has the following profle : 7: 1116666.
468 jan hoogland
Te equivalent which is mentioned in three dictionaries
seems to have become a lexicalized expression during the last years.
As for the neologisms presented by the other four dictionaries , the
fact that four dictionary compilers have added descriptions to these
neologisms illustrates that they are not completely at ease about the
understandability of the neologisms they present.
7. Te word logement / lodging has the following profle 7: 2224445.
Tis profle seems to justify the conclusion that this word is a
Dutch _LU Engl_Tr E_A_
Mawr
E_A_
Oxf
E_A_
Moun
F_A_
Kamel
F_A_
Manh
G_A_
Schreg
D_A_
Mol
Transl.
Profle
bestek cutlery 2 2 2 2 2 2 2 7: 2222222
francofel francophile 2 2 2 2 2 2 0 6: 222222
frankeren to frank 2 2 2 2 2 2 0 6: 222222
freak freak 2 0 0 4 4 0 0 3: 244
fresco fresco 2 2 2 2 2 2 0 6: 222222
gletsjer glacier 6 1 1 6 6 6 1 7: 1116666
glibberen slither 2 1 1 1 1 0 0 5: 11112
gnifelen snigger 2 2 2 2 2 2 0 6: 222222
loge guest 4 0 0 4 4 2 4 5: 24444
logement lodging
(house)
2 2 2 1 1 4 1 7: 2224445
lommerd pawnshop 6 2 6 3 2 2 0 6: 222366
loods pilot
(nautic)
2 2 2 1 1 1 1 7: 1111222
loops in heat/
season
2 2 2 0 2 2 0 5: 22222
lotgenoot partner (in
misfortune
/adversity)
0 0 0 0 2 2 4 3: 224
ouderwets old-
fashioned
2 2 2 2 2 2 2 7: 2222222
ouvreuse usherette 2 2 2 6 6 0 0 5: 22266
ouwel wafer 2 2 2 5 2 4 0 6: 222245
ribbel rib, cord 4 2 0 4 2 0 0 4: 2244
rif reef 6 2 2 6 6 5 5 7: 2255666
riposteren riposte 2 2 2 2 2 0 0 5: 22222
roe rod 4 1 1 0 1 4 4 6: 111444
rog ray 1 1 1 1 1 0 0 5: 11111
trampoline trampo-
line
2 6 6 3 0 3 0 5: 23366
transcri-
beren
transcribe 2 2 2 4 4 2 4 7: 2222444
traumatisch traumatic 1 0 1 1 1 1 1 6:111111
TABLE 4 EVALUATION OF THE TRANSLATIONS IN
7 DICTIONARIES
lexical gaps in arabic: evidence from dictionaries 469
problem word. Indeed, it can easily be translated with a hyperonym ,
since it is a certain type of accommodation for staying overnight.
Te Dutch Van Dale dictionary defnes a logement as a place to stay,
a home where one can obtain a temporary stay (and food) against
payment (presently less in rank or appreciation than a hotel). Neither
is lodging the same as a hotel. So translations with or should
be evaluated as hyperonym translations .
Another way of looking at the profles is the following:
Tere are nine Dutch words that were translated with a description in
all dictionaries in which they were included:
bestek / cutlery 7: 2222222
ouderwets / old fashioned 7: 2222222
francofel/ francophile 6: 222222
frankeren/ to frank 6: 222222
fresco/ fresco 6: 222222
gnifelen / to snigger 6: 222222
loops / in heat 5: 22222
ouvreuse / usherette 5: 22266
riposteren / riposte 5: 22266
Tese are the words of which the translation profle consists of code 2
and 6 only. It seems reasonable to sustain that words found in four or
more dictionaries (out of seven) and being translated with descriptions
(possibly in combination with a neologism ) in all those dictionaries
indeed do represent a lexical gap in Arabic .
So the concepts expressed by the words bestek / cutlery, ouderwets
/ old-fashioned, francofel / francophile, frankeren / to frank, fresco
/ fresco, gnifelen / to snigger, loops / in heat, ouvreuse / usherette
and riposteren / to riposte can be considered lexical gaps in Arabic .
Tere are fve Dutch words that were translated with a description in
most dictionaries in which they were included:
lommerd / pawn shop 6: 222366
lotgenoot / partner in misfortune 3: 224
ouwel / wafer 6: 222245
rif / reef 7: 2255666
470 jan hoogland
trampoline / trampoline 5: 23366
Since most of the translations are descriptions , and the others should
be qualifed as inadequate translations (neologisms , hyperonyms or
inaccurate), we may assume that these concepts also represent lexical
gaps in Arabic .
As a matter of fact, this system for encoding types of translations and
combining them in a translation profle seems to be a useful method for
comparing dictionaries and identifying lexical gaps in Arabic .
When we take a closer look at the Dutch words meeting with lexical
gaps in Arabic , it turns out, not surprisingly, that they can be categorized
as follows:
the source unit expresses a typical Dutch or non-Arabic concept
causing a lexical gap in Arabic : fresco / fresco, ouvreuse /
usherette, transcriberen / to transcribe;
the source unit expresses a technical concept , too rarely used
inside the Arab world in order to enter a dictionary and to receive
a general translation , thus this concept represents a lexical gap in
Arabic : frankeren / to frank, trampoline / trampoline.
the source unit is related to a non-Islamic religious concept ,
for example: ouwel / wafer, coming from Catholicism within
Christianity.
4. Conclusion
4.1 Conclusion concerning the identifcation of lexical gaps in Arabic
By comparing the Arabic translations of twenty-fve Dutch words,
we concluded that nine were translated with descriptions in all the
dictionaries in which they were included. Tis justifes the conclusion
that these words can be considered to represent a lexical gap in Arabic.
Five words were translated with descriptions in most of the dictionaries
in which they were included, which also indicates they should be
considered to represent lexical gaps in Arabic .
Terefore, the hypothesis formulated at the beginning of this
contribution may be considered to be confrmed by this conclusion.
lexical gaps in arabic: evidence from dictionaries 471
4.2 Conclusions concerning lexicography in general
As a result of the comparison described above, some conclusions
concerning the lexicography of Arabic can be drawn as well.
Te frst conclusion concerns the use of Arabic neologisms in
dictionaries . During the stage of comparing the descriptions in various
dictionaries , it was noticed that in some dictionaries newly coined words,
which should be considered neologisms in Arabic, were presented as
translations .
However, when neologisms are introduced in a bilingual dictionary as
a translation equivalent for an unknown foreign language word, there is
no context in the users mother tongue. As a matter of fact the dictionary
user will only run into such a neologism by looking up a foreign word
in a foreign language-Arabic dictionary. When the user is confronted
with a neologism without any further explanation, he/she will probably
remain in confusion about the meaning of the foreign language word
as well as to the meaning of the Arabic neologism . Some dictionaries
provide neologisms with an additional explanation or description .
Tis will help the user both to understand the meanings of the foreign
language word and the neologism in his own language.
If the same dictionary is used by foreigners who use the dictionary
to produce Arabic , and these foreigners use the neologisms presented,
there is again the risk of the Arab receiver of the message who will not
understand the meaning of the neologism .
To conclude, the following question presents itself: why do dictionary
compilers decide to invent new words as translation equivalents for
foreign words / concepts? Is this related to a certain way of language
purism in order to prevent the use of loan words ? Do they realize that
the dictionary users still will be lef in confusion? Te result of this policy
will be the occurrence of ghost words , i.e. words that do only occur in
dictionaries and not in authentic texts.
A second conclusion concerns the occurrence of inaccurate trans-
lations in the dictionaries . As indicated in Table 4, some inaccurate
translations have been met during the process of comparing
the translations . Inaccurate translations are represented by value 5 in
the table. Fortunately, the total number of translations , qualifed as
inaccurate, is rather limited, since value 5 occurred only three times on
a total of 142 translations in Table 4.
No dictionary is free of inaccuracies , and no one knows this better
than the present author who was involved in compiling the Dutch-
472 jan hoogland
Arabic dictionaries. Some descriptions in the Nijmegen Dutch-Arabic
dictionary could have been replaced with real translational equivalents .
As stated before, some Dutch words in the sample turned out to be
translatable with equivalents in at least some of the dictionaries . Tis
implies frst of all that the concepts behind them certainly do not represent
lexical gaps in Arabic . And secondly, these words should not have been
translated with descriptions in the Nijmegen Dutch-Arabic dictionary.
Tis, in a certain way, can also be considered a series of inaccuracies
in the said dictionary . Would the authors involved in translating these
specifc words have consulted all the dictionaries available, this might
have led to the insertion of a translational equivalent in stead of a
description . Not only would this have enriched the Dutch-Arabic part,
but also the reverse part, since translational equivalent relations are
bidirectional and consequently would have been entered in the Arabic-
Dutch part as well.
Some dictionary makers rely heavily on their colleagues or
predecessors. Some similarities between certain pairs of dictionaries
in the comparison above, however, are striking. If we, for instance,
look at the translations for logement / lodging, lommerd / pawn
shop, loods / pilot, loops / in heat, ouvreuse / usherette, ouwel /
wafer, riposteren / to riposte, in the dictionaries E-A Oxford and E-A
Mounged , we fnd a number of literally identical translations . Tis is so
obvious, that one may conclude, that the compiler of the most recent one
of the two dictionaries (Mounged English-Arabic ) depended heavily on
his predecessor (Oxford English-Arabic ), without mentioning this in
the introduction of his dictionary, and even without entering the Oxford
dictionary in the list of sources consulted. Te possibility that both of
the dictionaries were based on a common third ancestor requires further
research.
5. References
Balabakk , Munr. 1981. Al Mawrid: A modern English-Arabic dictionary. Beirut: Dr
al-Ilm li-l-Malyn.
Doniach , Nakdimon. 1972. Te Oxford English-Arabic Dictionary of Current Usage.
Oxford: Oxford University Press.
Hoogland , Jan, Kees Versteegh, and Manfred Woidich, eds. 2003. Woordenboek Arabisch-
Nederlands Nederlands-Arabisch. 2 Vols. Amsterdam: Bulaaq.
Idriss , Souheil. 1996. Al Manhal Dictionnaire Franais-Arabe. Beirut: Dar Al Adab.
Mol, Mark van . 2001. Leerwoordenboek Nederlands-Arabisch. Amsterdam: Bulaaq.
lexical gaps in arabic: evidence from dictionaries 473
Reda , Youssof M. 1996. Al-Kamel Al-Kabir plus, dictionnaire du franais classique et
contemporain, franais-arabe. Beirut: Librairie du Liban.
Schregle , Gtz. 1977. Deutsch-Arabisches Wrterbuch. Beirut: Librairie du Liban.
Teodory , Constantin. 1996. Al-Mounged English-Arabic. Beirut: Librairie Orientale.
MAS DAR FORMATION
Joost Kremers
University of Cologne
1. Introduction
Arabic morphology includes a nominal form, traditionally called
mas dar , whose meaning and form is related to that of a verb . Te form
relation, although clearly visible, is complex, the meaning relation is
straightforward: the mas dar is a complex event nominal , in terms of
Grimshaw (1990), and names the action expressed by the verb , retaining
the verbs argument and event structure.
1
Te mas dar has been analyzed among others by Fassi Fehri (1993)
and Kremers (2003), who both base themselves on Abney s (1987) anal-
ysis of English gerunds . Tese analyses are purely syntactic , however.
Neither discusses how the morphology of these forms functions, the
tacit assumption being that the mas dar form is derived by some post-
lexical process that does not play a role in syntax .
Ackema and Neeleman (2004) discuss so-called mixed categories ,
such as English gerunds , in the framework of their theory on word for-
mation and morphological structure. Tey argue that the peculiar syn-
tactic properties of such structures follow from the way in which they
are formed morphologically.
Arabic mas dar s share the peculiar syntactic properties of other mixed
categories , but Ackema and Neeleman s analysis does not provide any
direct insight into the reason why this should be so. Teir analysis cru-
cially depends on the concatenative nature of morphology in the lan-
guages that they discuss, and since mas dar formation in Arabic uses
non-concatenative morphology , the analysis at frst sight does not carry
1
The same nominal forms can also have non-event meanings, making them result
or simplex event nominals , in Grimshaw s terms. These lack the argument and event
structure of the corresponding verb , and have no systematic meaning relation to it. See
Fassi Fehri (1993) and Kremers (2003) for some discussion. Traditionally, these nouns
are not called mas dars.
476 joost kremers
over straightforwardly. As I show in this paper, however, nothing needs
to be changed about Ackema and Neeleman s analysis if we adopt a pro-
sodic model for Arabic morphology , along the lines of McCarthy and
Prince (1990b, 1996).
2. Abney s analysis of English gerunds
In his infuential PhD dissertation, Abney (1987) presents an analysis of
the English gerund that tries to account for the fact that the gerund can
be used in at least three diferent constructions:
(1) a. Johns singing of the Marseillaise
b. Johns singing the Marseillaise
c. John singing the Marseillaise
In all three constructions in (1), the gerund singing is an event nominal ,
in the sense that it retains the argument and event structure of the under-
lying verb to sing (Grimshaw 1990). In other words, all three phrases
describe an event , and the subject and object of the event are obligatorily
expressed.
2
Abney observes (as others have done before him) that there
is a decreasing degree of nominality in the structures (1ac): (1a), the
so-called Ing-of construction , is the most nominal, in that it expresses
both the subject and the object with nominal means: the Saxon geni-
tive s and the dummy preposition /case marker of, respectively. (1b),
the Poss-ing construction , is more nominal, in that the object is marked
with accusative case , rather than with of. Yet, the subject is still marked
with the Saxon genitive . Lastly, (1c), the so-called Acc-ing construction ,
marks the object with accusative , and the subject as well. Tis is there-
fore the most verbal of the three constructions, in that there is no nomi-
nal marking present anymore.
Crucially, the fourth logical possibility, the subject being marked with
accusative and the object with of, does not occur:
(2) *John singing of the Marseillaise
2
If one or both of the arguments are not expressed, e.g. in Johns singing, the gerund is
no longer a complex event nominal , but rather a simplex event or result nominal , which,
as Grimshaw (1990) shows, have markedly different properties.
mas dar formation 477
Tis fact can be accounted for if we assume, as Abney does, that gerunds
start out as V projection s, and change into an N projection somewhere
along the way. Until the point where the change takes place, arguments
can be licensed through case, but afer the change, only nominal licens-
ing mechanisms (s and of ) are available. Once the change has taken
place, it cannot be undone, which accounts for the impossibility of (2).
Te analysis is supported by the observation that (1bc) allow adverbs
but no adjectives , while (1a), the purely nominal gerund type, allows
adjectives but no adverbs :
(3) a. Johns constant/*constantly singing of the Marseillaise
b. Johns *constant/constantly singing the Marseillaise
c. John *constant/constantly singing the Marseillaise
Te analysis that Abney proposes assumes that there is an afx -ing,
which attaches to a verbal category , changing it into its corresponding
nominal category . It can attach at three levels: at V, creating an N, at VP ,
creating an NP , and at IP, creating its corresponding nominal projec-
tion DP. Crucially, -ing is not a head, i.e., it does not project a syntactic
phrase of its own. It just attaches to a projection, changing its category.
Te tree structures that Abney proposes are the following:
(4) Ing-of :
DP
Johns D
D NP
N PP
-ing V
sing
of the Marsaillaise
478 joost kremers
Tere is of course one unattractive aspect to these structures: it is not
clear how the verb stem sing and the sufx -ing combine to form the
gerund singing. In fact, it is not at all clear what it really means for an
afx to attach to an XP .
3
3. Ackema and Neeleman s treatment of mixed categories
Ackema and Neeleman (2004) retain the idea that what is diferent about
the diferent gerund structures is the level at which the nominalizing
3
Abney notices this problem himself. In order to solve it, he argues for an elaboration
of X theory, which basically makes a distinction between below X
0
and above X
0
syntax
which is not altogether convincing.
DP
Johns D
D NP
-ing VP
V
PP
sing the Marsaillaise
DP
-ing IP
John I
I VP
V DP
sing
the Marsaillaise
(5) Poss-ing :
(6) Acc-ing :
mas dar formation 479
afx attaches in syntax . However, they place this idea in the context of a
much broader theory on word formation , which enables them to show
what exactly it means for an afx to attach to diferent levels of projec-
tion. Before we look at how they deal with gerunds , it is necessary to
discuss some of the aspects of their theory.
3.1 Short outline of the theory
Ackema and Neeleman argue for a view of the language faculty that is
inspired by Jackendof (1997, 2002). In this view, the language faculty
contains three generative systems , one for syntax , one for semantics and
one for phonology . Te output of these generative systems are linked to
each other by mapping rules .
In this model of the language faculty , a lexical item is not just a con-
glomerate of the semantic , syntactic and phonological properties of a
word. Rather, the three types of features are essentially separate, func-
tioning in separate components of the language faculty , linked through
mapping rules . A word such as tree has the syntactic representation
N[+count,sg], which is used in the syntactic module, and it has the
phonological representation /t
i:/
Afxes function the same way: they have, apart from a semantic rep-
resentation , which does not concern us here, a morphosyntactic one,
which Ackema and Neeleman represent in small caps, as affix , and a
morphophonological one, which they represent with slashes, as /afx/.
Te point is that what is traditionally seen as a single afx actually con-
sists of three separate elements, linked by mapping principles .
Tere are diferent types of mapping principles . Ackema and Neele-
man argue that there are (at least) three general mapping principles , of
which two concern us here.
4
Te frst is Linear Correspondence :
4
The third mapping principle is Quantitative Correspondence which states that no
element in the morphosyntax is spelled out more than once.
480 joost kremers
(8) Linear Correspondence
If X is structurally external to Y, then F(X) is linearly external to F(Y).
5
Linear Correspondence states that a morphosyntactic representation
such as [[root sf
1
] sf
2
], in which sf
1
is frst attached to root, and sf
2
then to the complex [root sf
1
], is mapped onto a morphophonological
representation in which the order of the two sufxes is retained, i.e., /
rootsf
1
sf
2
/. It can in principle not be mapped onto a structure in which
both sufxes have switched places: */rootsf
2
sf
1
/.
Te second general mapping principle is Input Correspondence :
(9) Input Correspondence
If an affix selects (a category headed by) X, then F(affix) takes F(X) as its
host.
Input Correspondence states that if an affix in morphosyntax attaches
to an element X of a specifc category, or to a projection of X, then the
morphophonological form to which affix is mapped must attach to
the morphophonological element associated with X. In other words, the
phonological form of a nominal affix cannot attach to a phonological
form that is associated with a syntactic V head ; it must attach to some-
thing that is associated with an N head .
Note that both Input and Linear Correspondence only apply when
F(affix ) has an overt phonological form. When F(affix) is empty (i.e.
//), they do not apply (or apply vacuously). Since both principles regu-
late the distribution of phonological material, it stands to reason that
they do not apply when there is no phonological material to distribute.
Apart from these general mapping principles , there are also lexical
mapping principles , as in (7). Note that not only words, but also afxes
can be described with lexical mapping rules. For example, the English
agentive sufx er is normally mapped onto the phonological form /
/:
to write o writer. Tis fact is recorded in the lexical entry for the afx :
5
The notation F(X) refers to the phonological structure onto which the syntactic
structure X is mapped. It is equivalent to the slash notation /x/ that Ackema and Neele-
man use, but in my opinion less confusing. I define it as in (i):
(i) a. D(X): the subtree that has X as root
b. F(X): the phonological material onto which D(X) is mapped.
mas dar formation 481
(10) agentive noun l er[N,sg] l /
/
Tere are, however, specifc cases in which er receives an idiosyncratic
mapping . One example is the case of the verb to type, of which the agen-
tive noun is typist, not *typer. In other words, we can say that there is an
idiosyncratic mapping rule of the form in (11):
(11) [[type] er] l /taip/ /ist/
Tis rule says that a syntactic structure of the form [[type] er] is
mapped onto a phonological structure /taip/ /ist/. Note that mentioning
the phonological form of the stem in the rule is in fact redundant: there
is no reason to assume that the lexical entry for type cannot provide its
phonological form in (11). Terefore, I adopt a slightly diferent nota-
tion for such idiosyncratic mapping rules . Instead of (11), I will write
the following:
(12) er/type l /ist/
(12) expresses that er, when attached to type, is mapped onto /-ist/.
6
Afxes are well-known to have selectional restrictions. For example, the
English agentive sufx -er must attach to a word (i.e., it cannot attach
to a phrase), and moreover, this word must be a verb . Te point that
Ackema and Neeleman make is that these selectional restrictions are in
fact of non-uniform nature, and must therefore lie in diferent modules
of the grammar.
By representing a sufx such as -er as is done in (10), it becomes pos-
sible to express this point: Ackema and Neeleman state that the syntactic
afx er has the requirement that it must attach to something of cat-
egory V, while the phonological afx /
f i
n a
l
Te top row in (23) represents the syllabic tier : the frst two syllables
are given by the stem VII marker . Te frst syllable is extrametrical (it
is later syllabifed by the insertion of an epenthetic /i/), and its coda
position is flled by the /n/. Te third syllable is the non-fniteness suf-
fx , which is heavy (i.e. has two mor ), as indicated. Te fnal syllable
is again extrametrical , and is added by default, since every Arabic stem
ends in an extrametrical syllable .
Te bottom row represents the segmental tier . Crucial is of course
the question how the various slots in both tiers are associated with each
other. Te initial /n/ is straight-forward: the stem VII marker speci-
fes that it is associated with the frst (extrametrical) syllable . Just as
straightforward is the fnal root consonant /l/. Because a stem must end
in an extrametrical syllable , the third consonant must always take this
position.
Te remaining segments, /f/ and // of the root , and /a/ and /i/ of
the nominalizer , are associated through the principle of Lef-to-Right
Association (Leben 1973). Te root consonants fll the coda position s of
the second and third syllable, and the vowels of the nominalizer fll the
peaks. Because the third syllable has an additional mora , and because
there is no segmental material anymore to fll it, the vowel /a/ spreads to
the second mora , which results in a long vowel .
Most of the mas dar forms can, mutatis mutandis, be analyzed in this
manner: for the mas dar forms of stems IV and VIIXV, all that needs to
change in (23) is the stem template .
21
As is clear from Table 1, however,
21
Note that the mas dar of stem IX /iKTaBB/ is formed on the base of the underlying
form /iKTaBaB/. The gemination of the third root consonant is the result of deletion of
the /a/, a common process in this context.
490 joost kremers
things are diferent for the remaining forms: they appear to be quite
idiosyncratic. In spite of the apparent complexity, however, we can show
that they all make use of the same stem template that the corresponding
verb forms use. Where they difer from other mas dar s is the form of the
nominalizer and the non-fnite morpheme , which is sometimes absent.
Let us look at the various forms. First, mas dar s V and VI have an
idiosyncratic nominalizer , that takes the form /au/ rather than /ia/.
Tey also lack the non-fnite morpheme , since the forms are not
*/taKaTTB/ and */taKTB/. Here, we can either say that they have an
idiosyncratic non-fnite marker -, or that the lack of a second syllable
in the template causes not just the stem afx but the entire verbal stem
template to be used as a basis for the mas dar . Either way, we obtain the
same result:
(24) stem V/VI:
nominalizer: /au/
I will assume that the nominalizer does not associate with the prefxed
syllable in the stem template . Terefore, the /a/ of the nominalizer asso-
ciates with the second (heavy) syllable of the stem template , and the /u/
with the third. Presumably, the /a/ of the frst, prefxed, syllable is a copy
of the /a/ of the second syllable.
22
Stem II mas dars can be dealt with in a similar way. On the face of it,
the stem II mas dar does not seem to be formed on the template of the
stem II verb, as it contains a prefx /ta/ which the verbal template lacks,
and its vowel pattern is /ai/, not /ia/, as the default nominalizer speci-
fes. Furthermore, it does not show gemination of the second consonant,
which seems typical for the verbal template .
However, McCarthy and Prince (1990b) observe that Arabic pho-
nology does not distinguish between CVV and CVC syllables : all that
counts is their prosodic status as heavy syllables. Te templates of stems
II and III (and likewise stems V and VI ) are therefore identical: their
frst syllable is , without any specifcation how the heaviness of the
22
One fact that supports this assumption is that Classical Arabic had an alternative
mas dar form for stem V verbs, namely /tiKiTTB/. On the assumption that the vowel
in the prefix is a copy of the vowel in the next syllable, this form is completely regular:
it contains the mas dar morpheme /ia/ and the non-finite suffix V, as indicated by the
fact that the final vowel is long.
t
-
mas dar formation 491
syllable is obtained. An additional gemination rule operates on stem II
and V verbs, but not on stem III and VI .
Suppose now that we say that the gemination rule does not apply in
stem II mas dar s (unlike stem V mas dar s ), and that the nominalizer of
stem II is idiosyncratic. We then obtain the following:
(25) stem II:
nominalizer: /tai/
non-fnite: -
Here, the stem II template is not specifc to the mas dar , it is the same
template that forms fnite verb forms of stem II . In order to see how
these morphemes yield the mas dar form /taktb/, let us see how they are
associated with the template . First, we add the non-fnite sufx to the
stem base , and associate the nominalizer with the resulting template:
(26)
Te frst syllable here has two mor because this is specifed in the stem
II template , and the second syllable has two mor because it is the non-
fnite sufx . Te vowel /i/ cannot be associated with the second mora of
the frst syllable because Arabic does not allow two vowels in a single
syllable.
23
When the root is intercalated into the template , the second
mora of the frst syllable can be associated with the frst root consonant .
Te second root consonant can be associated with the onset of the sec-
ond syllable, and, as usual, the last consonant is associated with a (newly
created) extrametrical syllable . At the same time, the /i/ associated with
the frst mora spreads to the second, creating a long vowel :
(27)
()
23
Long vowels and diphthongs are either the result of lengthening or because the sec-
ond mora is filled by a semi-consonant, not because two vowels end up in one syllable.
t a
k
t i
t
a
i
492 joost kremers
In this way, we can derive the stem II mas dar by just assuming an idio-
syncratic nominalizer , in the same way that we have analyzed stem V
and VI mas dar s .
24
Te stem III mas dar is quite idiosyncratic as well. Its form is
/muKTaBa(t)/, where the fnal /-a(t)/ is the regular feminine ending .
Tis mas dar is a so-called mas dar mmy or m-mas dar . Te m-mas dar is
an alternative mas dar formation found in Classical Arabic , in which the
feminine form of the passive participle is used as mas dar . Tis mas dar
formation has mostly disappeared, but in stem III , it is the dominant
one.
25
I will not go into the details of participle formation in Arabic
(see McCarthy 1981 for some discussion), but like mas dar formation ,
it is templatic, with an /m-/ prefx that marks the participial form and a
vowel pattern that indicates voice.
Summarizing, we can say that there is a regular nominalizer /ia/,
which applies in the majority of cases. Tis afx is combined with a non-
fnite sufx V. Stems II , V , and VI have an idiosyncratic nominal-
izer , and stem III has an idiosyncratic mas dar formation . Stems V and
VI lack the non-fnite sufx , and instead get a default V as second syl-
lable. Lastly, stem I mas dar s are all idiosyncratic. In the next section, I
will discuss the syntactic and phonological processes that underlie the
mas dar formation in more detail, and discuss the mapping rules needed
to account for them.
4.3 Te mas dar afx
As we have seen, the syntactic properties of mas dar s are essentially
identical to those of gerunds . Terefore, the starting point for the analy-
sis should be that the syntactic structure of mas dar s is the same as well:
there is a syntactic mas dar afx that attaches to V or to a projection of
V.
If mas dar attaches to V, we obtain the equivalent of the English Ing-of
construction : mas dar +li . Mapping of such a structure to phonology
24
Note that the /ta/ element in the mas dar morpheme is not a prefx , contrary to the
/t/ element in stem V and VI forms. If it were, a form /taKaTTB/ would result. Instead,
it must be part of the nominalizer .
25
Classical Arabic had three other stem III mas dar s , /KiTB/, /KTB/ and /KiTTB/,
but none of these were as common as the m-mas dar .
mas dar formation 493
is unproblematic in any event, as we have seen above, because V and
mas dar are sisters in the tree. Input and Linear Correspondence are
easily satisfed.
However, Linear Correspondence is only apparently satisfed. As we
will see, there is no way in which mas dar in a mas dar +acc construction
can adhere to Linear Correspondence . We must therefore assume that
mas dar is not subject to it at all, so that even in mas dar +li construc-
tions, adherence to Linear Correspondence is only apparent.
(28) shows the tree of a mas dar + acc construction. mas dar attaches
to VP , which yields a structure in which the object can be assigned accu-
sative case:
(28) DP
D NP
MAS DAR VP
Subj V
V Obj
However, on the assumption that mas dar is subject to Linear Corre-
spondence , the mapping to phonology would violate at least one map-
ping principle . Te reasoning is identical to the one discussed in section
3.2 for the English gerund constructions . mas dar attaches to the VP , so
Input Correspondence requires that /mas dar / attaches to /v/. Doing so
would violate Linear Correspondence , however: mas dar is external to
the VP , but having /mas dar / attach to /v/ leaves it internal to it, as it then
occupies a position between /subject/ and /v/.
Te analysis that made this confguration unproblematic for English ,
saying that the nominalizing /afx / is phonologically null, is not avail-
able for Arabic . As we have seen above, the Arabic mas dar formation
uses an overt nominalizing morpheme . Another possible solution that
easily comes to mind is to adjust the structural relations in the tree in
such a way that mas dar and its intended host V are adjacent, so that /
mas dar / can attach to /v/ without violating Linear Correspondence .
Such structural rearrangement obviously implies movement. As
argued in many works (e.g., Ritter 1991, Kremers 2003), in possessive
constructions in Arabic and Hebrew (of which (28) is one, because the
494 joost kremers
subject is assigned genitive case ), N moves to D. If we apply this move-
ment to (28), the V head
26
and mas dar would end up adjacent.
Such an analysis might seem attractive at frst sight, because the order
in mas dar constructions is always NSO. Terefore, the V head must
move to a higher position: the tree in (28) could not derive the correct
order. Moving V to a position adjacent to mas dar in (28) would enable
Linear Correspondence to be adhered to while at the same time deriv-
ing the correct word order.
Tere are some problems with this analysis, however. First, there is no
real consensus that N-to-D movement really takes place: Borer (1999)
argues against it, for example. Furthermore, even if one assumes that it
can take place, N does not always move to D in the Arabic noun phrase .
If there is no genitive-marked DP in the noun phrase , N-to-D movement
presumably does not take place. Terefore, it cannot be guaranteed that
/mas dar / and /v/ will always, under any conceivable construction, end
up adjacent.
Te root of these problems, however, is the tacit assumption that
mas dar is a prefx .
27
Te phonological shape of the mas dar afx , how-
ever, suggests that this assumption is incorrect: as discussed above, much
of Arabic morphology consists of prosodic templates that map onto a
prosodic tier in phonology (see, e.g. McCarthy 1981). Te mas dar afx
is such a prosodic template , as we have seen in section 4.2. It is neither a
prefx nor a sufx , and hence does not appear before or afer the verbal
root . Rather, one would say it appears simultaneous with it.
At frst sight, this seems a problematic conclusion. mas dar is attached
to a syntactic structure , and syntactic structures are at some point lin-
earized. Te common assumption is that linearization is total: it applies
to every terminal element in the tree. Tat is, the linear structure that
results from a syntactic tree contains all the terminal elements in that
tree, and for any pair of distinct elements x,y, a linear order is defned,
either x>y or x<y (Kayne 1994). Linear Correspondence is basically a
formalization of this assumption.
26
Of course, there is no N head in (28), so we would have to assume that V moves
to D.
27
Note that the entire solution would be impossible if /mas dar / were a sufx : the
object would then always end up between /v/ and /mas dar /, no matter where V moves
to.
mas dar formation 495
However, as mentioned above, Ackema and Neeleman assume that
Linear and Input Correspondence do not apply when a syntactic element
maps onto // in phonology . Tat is, the mapping principles are sensi-
tive to the phonological form of the elements they apply to. It should be
no surprise, then, that given the prosodic nature of the mas dar afx , it
is also not subject to Linear Correspondence .
More precisely, the mas dar morpheme maps onto the prosodic tier ,
and therefore does not (and in fact cannot) require linearization with
respect to the material on the segmental tier . However, it does require
association with material on the segmental tier . Tis is obtained through
Input Correspondence : because mas dar attaches to the VP , F(mas dar )
must attach to F(V), which is the root morpheme . Since the root mor-
pheme consists of segmental material, the required association is
obtained.
Having said that, let us look at how mas dar formation takes place
in detail. In section 4.2, I have argued, following McCarthy and Prince
(1990b), that the morphosyntactic process of mas dar formation involves
two morphemes: a nominalizer , which I will indicate with noml , and
a non-fnite sufx n-fin . Above in this section, on the other hand, I
have assumed a mas dar afx that must be able to attach to a structure
in phrasal syntax , in order to derive the Arabic equivalent of Poss-ing
constructions . Te question we must ask, then, is what is the relation
between mas dar on the one hand, and noml and n-fin on the other.
Obviously, noml and n-fin in some way make up mas dar . I propose
the following structure:
(29)
NOML
NOML N-FIN
Tis is a morphological structure, which is then inserted into the syn-
tactic structure at the position of the mas dar afx :
496 joost kremers
(30)
I assume that the V head is composed of the root and the verb stem
marker , here stem VII . Te mas dar afx , as indicated, is composed of
the nominalizer and the non-fniteness sufx . When the tree in (30) is
mapped onto phonology , Input Correspondence will make sure that
F(mas dar) is properly associated with F(V). Te formation of the
mas dar form /infl/ will then proceed as described above.
Note that whether a syntactic element is subject to Linear Correspon-
dence is not a function of its morphological form alone. F(V) itself con-
sists of autosegmental morphemes , but cannot be exempt from Linear
Correspondence : the entire mas dar form consist of autosegmental mor-
phemes , and at least one of them must be subject to Linear Correspon-
dence , otherwise the form could not be linearized with respect to the
other terminal elements in the structure. Te natural assumption is that
the root , which is not a syntactic affix , unlike all the other morphemes,
is this element. Note that the root is a phonological afx , because it can-
not form a stem of its own. Syntactically, however, it is not, because it
does not require adjunction to a structure of a specifc category. Tis
syntactic diference between the root and the other elements is presum-
ably the result of a semantic diference: the root is a lexical item (in the
traditional sense of the word) and as such member of an open semantic
class. All the other morphemes are functional or derivational, and part
of closed semantic classes. It therefore seems reasonable to assume that
members of open classes must always be mapped onto the segmental tier
in some way, and are thus always subject to Linear Correspondence . Te
V
DP
D
NP
mas dar VP
Subj
V
Obj
noml
noml n-fin
V
vii
root
mas dar formation 497
result of this assumption is that the mas dar form is linearized according
to the position that the head V takes in the syntactic structure.
28
Stem V /VI and stem II mas dars difer very little from regular mas dars.
All that is required is to assume the following idiosyncratic mapping
rules :
(31) stem V: noml/V l /ua/
n-fin/V l V
stem II: noml/II l /tai/
Tat is, noml in the context of a stem V template is mapped onto /ua/,
and n-fin onto -V.
29
Furthermore, when attached to a stem II template ,
noml is mapped onto /tai/ rather than the default /ia/.
Te system can also account for the stem III mas dar . Recall that this
mas dar is a so-called m-mas dar , i.e., it is not formed with some genuine
mas dar afx . Rather, a more or less suppletive strategy is used, which
employs the feminine passive participle . Let us say that there is an idio-
syncratic mapping rule of the following form:
30
28
Note that this means that V must move, because mas dar constructions have the
order NSO. At frst sight, it might be tempting to argue that the NSO order is derived by
spelling out the mas dar in the position of mas dar, but this would not work for mas dars
that license their objects with the preposition li: the analysis states that in such mas dars
the mas dar afx attaches to V rather than VP , which would predict a surface order of
SNO for such constructions. Te actual surface structure is NSO, however, the same as
for mas dar s that assign accusative.
29
Te stem V rules obviously also apply to stem VI , and, although not discussed
here, to the second quadriliteral stem as well. Te rule may in fact refer not to the stem
templates but to some other property: as McCarthy and Prince (1990b) discuss, these
three verb stems share properties with each other that are not found in other verb stems ,
indicating that they form a class of their own. Presumably, the mapping rules refer to
this class.
30
I mentioned in footnote 6 that Ackema and Neeleman s formulation of idiosyn-
cratic mapping rules cannot be read as a one-to-one mapping in all cases. (32) is one
such case. Ackema and Neeleman s rules difer in that they mention the phonological
form of the host as well as of the afx . But doing that in (32) would still not establish
that a form consisting of F(prt.pass.f)+F(III) is equivalent to a stem III mas dar , as it
could (obviously) also be a feminine passive participle of a stem III verb. Note that this
is indicative of a general asymmetry between syntax and phonology : a syntactic struc-
ture is always mapped onto one particular phonological structure, but a phonological
structure may have more than one syntactic equivalent. In other words: phonological
structures can be ambiguous.
498 joost kremers
(32) [
noml
noml n-fin]/III l F(prt.pass.f)
I will not go into the formation of the participle here. What is relevant
is the fact that an idiosyncratic mapping rule of the form in can exists.
Rather than specifying the phonological material that the syntactic
structure under consideration is mapped onto, the rule specifes a dif-
ferent (morpho)syntactic form whose phonological mapping must be
applied.
5. Conclusions
Ackema and Neelemans (2004) theory on word formation can give a
straightforward account of so-called mixed categories in languages
where morphology is concatenative. Non-concatenative morphology ,
such as that of Arabic mas dars , at frst sight does not yield to an analysis
in terms of Ackema and Neeleman . If, however, we adopt the common
analysis of non-concatenative morphology in terms of autosegmental
tiers , we fnd a natural way to exclude the mas dar morpheme from Lin-
ear Correspondence , which is the greatest obstacle to the application of
Ackema and Neeleman s analysis to mas dars : Linear Correspondence
only applies within an autosegmental tier in phonology .
At the same time, Input Correspondence still applies to the mas dar
morpheme , accounting for the fact that it takes the verb as its host, and
not some random root in the rest of the structure.
6. References
Abney , Steven. 1987. Te English Noun Phrase in its Sentential Aspect. PhD thesis, MIT:
MIT Working Papers in Linguistics.
Ackema , Peter and Ad Neeleman . 2004. Beyond Morphology: Interface conditions on
word formation. Oxford: Oxford University Press.
Borer , Hagit. 1999. Deconstructing the construct. In Kyle Johnson and Ian Roberts,
eds. Beyond Principles and Parameters. Dordrecht: Kluwer Academic Publishers,
4390.
Eid , Mushira and John McCarthy, eds. 1990. Perspectives on Arabic Linguistics II.
Amsterdam: John Benjamins.
Fassi Fehri , Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Dor-
drecht: Kluwer Academic Publishers.
Grimshaw , Jane.1990. Argument Structure. Cambridge, Mass: Te MIT Press.
Jackendof , Ray. 1997. Te Architecture of the Language Faculty. Cambridge, Mass: Te
MIT Press.
mas dar formation 499
. 2002. Foundations of Language: Brain, meaning, grammar, evolution. Oxford:
Oxford University Press.
Kayne , Richard. 1994. Te Antisymmetry of Syntax. Cambridge, Mass: Te MIT Press.
Kremers , Joost. 2003. Te Noun Phrase in Arabic: A minimalist approach, PhD thesis,
University of Nijmegen. LOT Dissertation Series 79.
Leben , William. 1973. Suprasegmental Phonology. PhD thesis, MIT.
McCarthy , John. 1981. A prosodic theory of nonconcatenative morphology. Linguistic
Inquiry 12(3). 373418.
McCarthy , John and Alan Prince . 1990a. Foot and word in prosodic morphology: the
Arabic broken plural. Natural Language and Linguistic Teory 8. 209283.
. 1990b. Prosodic morphology and templatic morphology. In Eid and McCarthy,
eds. Perspectives on Arabic Linguistics II Amsterdam: John Benjamins. 154.
. 1996. Prosodic morphology 1986. Ms. University of Massachusetts, Rutgers Uni-
versity.
Reuland , Eric. 1983. Governing ing. Linguistic Inquiry 14(1). 101136.
Ritter , Elizabeth. 1991. Two functional categories in noun phrases: Evidence from
Modern Hebrew. In Susan Rothstein, ed. Perspectives on phrase structure: Heads and
licensing. San Diego: Academic Press. 3762.
Selkirk , Elizabeth. 1980. Te role of prosodic categories in English word stress. Lin-
guistic Inquiry 11(3). 563605.
Truckenbrodt , Hubert. In press. Te syntax-phonology interface. In Paul de Lacy, ed.
Cambridge Handbook of Phonology. Cambridge: Cambridge University Press.
Wright , William. 1981. A Grammar of the Arabic Language. Beirut: Librairie du Liban.
Yoon , James. 1996. Nominal gerund phrases in English as phrasal zero derivations.
Linguistics 34. 329356.
MTHODOLOGIE LINGUISTIQUE :
ORGANISATION DE LA LANGUE ARABE
ORGANISATION GNRALE DES LANGUES
Andr Roman
Lyon
1. Introduction
Archimde (~ 287~ 212), ayant tabli la thorie du levier , aurait
demand pour soulever le monde un point dappui .
Lon peut voir dans le levier le symbole des mthodologies . Dans le
point dappui , le symbole de leur premier point dapplication.
1
Un exemple admirable, clbre, de cette double dmarche est donn
par Ren Descartes dans son Discours de la mthode (1819 ; 29 ; 32) :
Je crus que jaurais assez des quatre [prceptes ] suivants [. . .]
Le premier tait de ne recevoir jamais aucune chose pour vraie, que
je ne la connusse videmment tre telle [. . .]
Le second, de diviser chacune des difcults que jexaminerais, en
autant de parcelles quil se pourrait, et quil serait requis pour les mieux
rsoudre.
Le troisime, de conduire par ordre mes penses, en commenant
par les objets les plus simples et les plus aiss connatre, pour mon-
ter peu peu, comme par degrs, jusques la connaissance des plus
composs ; et supposant mme de lordre entre ceux qui ne se prcdent
point naturellement les uns les autres.
Et le dernier, de faire partout des dnombrements si entiers , et des
revues si gnrales, que je fusse assur de ne rien omettre.
Ces longues chanes de raisons , toutes simples et faciles, dont les gom-
tres ont coutume de se servir, pour parvenir leurs plus difciles dmon-
strations, mavaient donn occasion de mimaginer que toutes les choses,
qui peuvent tomber sous la connaissance des hommes , sentre-suivent en
mme faon, et que, pourvu seulement quon sabstienne den recevoir
aucune pour vraie qui ne le soit, et quon garde toujours lordre quil faut
1
Le symbole est la peinture dune mtaphore.
502 andr roman
pour les dduire les unes des autres, il ny en peut avoir de si loignes
auxquelles enfn on ne parvienne, ni de si caches quon ne dcouvre. Et
je ne fus pas beaucoup en peine de chercher par lesquelles il tait besoin
de commencer : car je savais dj que ctait par les plus simples et les plus
aises connatre.
Au contraire [des sceptiques] tout mon dessein ne tendait qu massurer,
et rejeter la terre mouvante et le sable, pour trouver le roc ou largile.
Et remarquant que cette vrit : je pense, donc je suis, tait si ferme et si
assure, que toutes les plus extravagantes suppositions des sceptiques
ntaient pas capables de lbranler, je jugeai que je pouvais la recevoir, sans
scrupule, pour le premier principe de la philosophie que je cherchais.
Ce problme du premier point de lapplication dune mthodologie ,
Ferdinand de Saussure lvoque dans une question, Unde exoriar ,
Do commencer ? (2002, 281). Sa question nest pas, semble-t-il, une
citation. La langue latine , dans laquelle il la formule, est la langue de la
premire voie qui sest ouverte lEurope, la voie romaine.
2
La langue
latine est la langue de la culture europenne.
1.1 Une langue et lhomme
Voici le linguiste devant une langue et devant lhomme . Il peut com-
mencer son tude par lhomme car lhomme est, dans le monde, le seul
animal qui parle. La capacit propre lhomme, lhomme seul, qui fait
de lui un homo loquens , un h ayawn nt iq,
3
est, selon toute vraisem-
blance, refte par les langues que lhomme parle.
4
Si cela est,
ou bien la capacit de parole de lhomme , cette capacit constitutive
de lhomme, tant reconnue, la constitution gnrale de la langue ,
qui est une actualisation de cette capacit, pourra tre reconnue dans
cette capacit ;
ou bien la constitution gnrale de la langue tant reconnue, la consti-
tution de lhomme , qui en est le crateur, pourra tre saisie dans la
langue .
2
Voir R. Brague , 1992.
3
Al-Maydn cite dans son Majma al-Amt l 2 : 291, n3958, ladage : m l-insnu
law l l-lisnu ill s ratun mumat t alatun aw bahmatun muhmalat .
4
Il ne sagit pas ici des capacits physiologiques de lhomme , ses capacits auditives,
articulatoires, car celles-ci ne portent que sur les sons que lhomme doit entendre, doit
produire comme il parle, et non pas sur la grammaire de la langue .
mthodologie linguistique : organisation de la langue 503
Faut-il commencer par lhomme ? Faut-il commencer par la langue ?
Traditionnellement les savants qui se sont intresss aux langues ont
tudi les langues et non pas lhomme . Ou, exceptionnellement, quand
ils se sont intresss lhomme , cest pour afrmer que la langue a t
donne lhomme par Dieu, ou, comme Noam Chomsky , pour avancer,
axiomatiquement, que la langue est inne dans lhomme .
1.2 Mthodologie
Lexemple de Descartes incite rechercher dabord une mthodologie .
La premire esquisse de la mthodologie requise peut se rsumer dans
les deux oprations suivantes :
classer ;
dfnir .
Classer , cest--dire constituer en un ensemble des entits runies sur
quelque trait commun , reconnu, un trait propre, prsent ici, absent l,
qui les difrencie, les oppose aux autres entits . Les plantes . Parmi
les plantes, les arbres . Parmi les arbres, le palmier . . . Les fruits
des arbres . Les fruits rouges , la fraise , la framboise , la gro-
seille . . .
5
Dfnir , cest--dire tablir, dans un ensemble donn par un classe-
ment , une certaine identit de chacun de ses lments. La plante est
un vgtal . L arbre est une plante dune certaine hauteur . . . Clas-
ser est une opration humaine , immdiate, constante. Dfnir est
tout autant une opration humaine , immdiate, constante.
La casuistique est une manifestation, universelle, clatante, de mise
en uvre, voulue objective, de ces deux oprations.
6
Le jugement de
valeur , si commun, est une autre mise en uvre, subjective, de ces
5
Les langues ngro-africaines ont un systme de classes nominales qui repose
sur une rpartition des tres et des objets et, postrieurement sans doute, des abstrac-
tions en un certain nombre de catgories , in Les langues du monde, 2, 740741.
6
On a soutenu avec vraisemblance que la tradition grammaticale arabe avait
emprunt la casuistique le cadre gnral de sa dmarche (Carter 1972, particulire-
ment). Versteegh (1980, 1114 ; 1993, 3336) a replac cette hypothse dans une per-
spective plus assure.
504 andr roman
deux oprations. Classer et dfnir apparaissent comme des op-
rations binaires, simples.
7
Tous les chercheurs ont class. Exemples de classement des gram-
mairiens, les parties du discours : le verbe ; l adverbe . . . ou, dans
le discours, la phrase . . . Et ils ont dfni.
8
Comme tous les hommes ,
toujours.
Mais, la difrence de lhomme naf, le chercheur , doit dfnir en
attribuant chaque lment dun ensemble donn une identit tablie
de telle sorte que cet lment puisse tre, constamment, reconnu par
une difrence raisonne, irrductible, persistante, qui loppose rguli-
rement ce qui nest pas lui.
Et le chercheur , la difrence de lhomme naf, doit rviser ses clas-
sements au fur et mesure du progrs de ses dfnitions . Lidentit du
dernier classement auquel il aboutira ne pourra tre que la mme iden-
tit des lments quil a identifs, qui le composent.
Ce faisant, le chercheur doit inscrire lobjet de sa recherche dans le
temps , car rien sur terre na dexistence hors du temps. Il lui faut donc ne
pas mconnatre lhistoire de lobjet de sa recherche.
2. Sbawayhi
Les dfnitions donnes par Sbawayhi (al-Kitb , I : 12) des trois parties
du discours sont brves :
9
[La particule ] est [dans la langue] pour un sens qui nexiste ni par le nom ,
ni par le verbe . Le nom , cest homme, cheval.
10
Le verbe , ce sont les
7
Schuler (1990, 252) a pu prsenter la dmarche gnrale de la plaidoirie dune
cause, de sa premire position sa dernire position dprcatoire , sur un algo-
rithme exactement binaire.
8
Voir infra les dfnitions des parties du discours donnes par Sbawayhi dans son
Kitb . De mme al-Frb , lun des plus minents et des plus clbres philosophes
musulmans [. . .] surnomm le second matre, le premier tant Aristote [. . .] , mort
Damas, en 339/950, dans son Ih s al-ulm, p. 5. Le nom h add, singulier de h udd,
signife communment limite et manire . Il a, chez Sbawayhi , o il est un terme
de mthodologie , gard son sens de manire (Troupeau , 1976, s.v. h add). Cest, sem-
ble-t-il, dans le Kitb al-h udd du grammairien et philosophe mutazilite ar-Rummn ,
mort en 394/994, quil a pris, suivant sa pente, le sens de dfnition . Semblablement, le
terme orismos, dfnition , employ par Aristote, est rapprocher de horos, bord .
9
Voir sur la dfnition des parties du discours Versteegh (1995, 2242).
10
Certains manuscrits du Kitb , la somme de la tradition grammaticale arabe
naissante, livre imparfaitement dit, ajoutent mur h it ).
mthodologie linguistique : organisation de la langue 505
schmes drivs de lexpression phonique des vnements lis aux noms et
qui sont construits pour ce qui est pass, ce qui sera, ce qui est encore.
11
fa l-kalimu : ismun wa-flun wa-h arfun ja li-manan laysa bi-smin wa-l
fl fa-l-ismu : rajulun wa-farasun. wa-amm l-flu fa-amt ilatun uxidat
min lafz i ah dt i l-asmi wa-buniyat li-m mad wa-li-m yaknu wa-
lam yaqa wa-m huwa kinun lam yanqat i.
La seule partie du discours recevoir dans ce court paragraphe une df-
nition est le verbe . Le verbe se conjugue. Sa conjugaison ofrait une prise
Sbawayhi . Mais, remarquablement, ce que Sbawayhi dit ici du verbe ,
il ne le dduit pas du verbe. Lexpression du temps quil prte au verbe ,
est lexpression du droulement du temps quil constate dans le monde.
Du nom , il sest born donner deux exemples. De fait, dans la langue
arabe de son temps, le nom nest pas construit. Il nest saisissable que par
sa relation ce quil nomme. Il faut donc, pour le reconnatre, regarder
dans le monde lentit quil nomme.
12
Au demeurant, les grammairiens arabes , avant mme Sbawayhi ,
avaient class, faussement, le relatif h ayt u, l o , parmi les noms .
Linclusion dans lensemble des noms dun lment qui nest pas un nom ,
rendait impossible la dfnition du nom. Il fallait sortir h ayt u de len-
semble des noms , recomposer le premier ensemble, encore mal assur,
des noms .
De la particule , qui ne lui apparat pas construite non plus, Sbawayhi
relve quelle assure, complmentairement, les tches smantiques , rf-
rentielles, que le verbe et le nom nassurent pas.
13
Ces tches sont les
tches de relation et de localisation , que lexprience immdiate montre
comme tant ncessaires.
Ce texte, qui sera exactement repris, perptuellement, apparat
comme une image du monde : ce qui est , le nom ; ce qui est dans
11
La constatation que les modus nexistent que dans ou par les res .
12
Un autre grand grammairien, Ibn Fris , mort Rayy en 395/1004, a, dans son livre,
as -Sh ibi f fqh al-lua wa-sunan al-arab f kalmi-h, reproch Sbawayhi , p. 85,
davoir, pour le nom , donn non pas une dfnition mais des exemples (tamt l).
Lui-mme a retenu, comme tant plausible, la dfnition suivante : Le nom est ce qui est
implant sur le nomm comme on le mentionne et qui lui reste attach. , d ukira l an
badi ahli l-arabiyyati anna l-isma m kna mustaqirran al l-musamm waqta d ikri-ka
iyy-hu wa-lziman la-hu wa-hd qarb . Mais cette dfnition nest rien dautre que la
reconnaissance du fait que le nom est rapport immdiatement au nomm, cest--dire
sans le truchement du systme de la langue .
13
En fait plusieurs particules sont analysables ; exemple la ngation , /law/, du mode
rel , faite de la ngation /l/ et du morphme /w/ (< /u/) du mode rel ; voir A. Roman
1998.
506 andr roman
le temps , le verbe ; avec, entre lun et lautre, le lien , indispensable,
verbalis par la particule qui les situe, qui est difrente, par son rle,
du verbe et du nom .
Ici, la tradition grammaticale arabe a choisi, semble-t-il, dobser-
ver dans la langue le monde, non pas lhomme donc et non plus, tout
dabord, la langue. Or la langue ne peut refter limmense complexit
du monde.
14
Cette porte du monde qui ouvrirait sur la langue est une
fausse porte.
15
Quant la quatrime unit fondamentale de la langue que compose-
raient ensemble le nom , le verbe , la particule , cette unit qui serait leur
fnalit dans la langue, cette unit, la phrase , les grammairiens non
arabes en ont prsum lexistence. Ils ont chou la dfnir.
16
Les gram-
mairiens arabes , eux, ne se sont pas attachs cette unit incertaine.
3. Discussion
Sil faut commencer ltude des langues smitiques par un ensemble
taill dans ces langues, pourquoi ne pas commencer par un ensemble ,
caractristique de ces langues, lensemble des suites ordonnes de trois
consonnes sur lesquelles les units de nomination de ces langues appa-
raissent construites, le plus souvent.
14
Les systmes matriels qui composent le monde, dune part, sont ouverts, la
difrences des systmes linguistiques ; dautre part, la combinatoire qui structure les
systme linguistiques est binaire voir infra , tandis que les combinatoires multiples
qui structurent les systmes matriels sont n-aires, avec n suprieur 2 .
15
La proposition de Tom , rapporte dans Petitot, Entrevue avec Ren Tom, est,
senzaltro, plus pittoresque. J.P. : Votre hypothse est [. . .] que les actions archtypales
comme capturer, prendre, couper, lier etc. sont devenues par ritualisation les matrices
de toutes les structures syntaxiques R.T. : Oui. Elles ont captur les structures plus
complexes. La meilleure preuve cest quil nexiste pas de verbes de valence suprieure
quatre. Cest la rgle des phrases de Gibbs . Ce passage ici repris de P.M. Lavorel 1980 :
I, 475).
16
G. Mounin , 1960, a recens, semble-t-il, toutes les dfnitions de la phrase . Aucune
nest opratoire. G. Kleiber , 2003, qui rfute le nouveau dcoupage du discours en clauses
et en priodes, concde qu on a tout gagner tenir compte de la dimension mmori-
elle introduite par Berrendonner , qui tait totalement absente des analyses tradition-
nelles et dont la nouveaut consiste montrer que les relations de discours narticulent
pas deux segments textuels, mais un segment textuel et une information en mmoire
discursive . Mais comment voir une pice de la langue dans cette dimension mmori-
elle ou dans la priode dfnie hypothtiquement comme une unit ergonomique
(A. Berrendonner , confrence, Lyon, 23.10.2003) ?
mthodologie linguistique : organisation de la langue 507
Ainsi, pour reprendre lexemple de toutes les grammaires , la mme
squence, ktb, se retrouve dans : /katbun/, crire (nominatif) ;
/kitbatun/, temps dcriture, criture (dun manuscrit) (nomina-
tif) ; /aktubu/, Jcris (mode rel ) ; /ktibu-n/, crivant (nomi-
natif) ;
17
/maktabun/, objet dcriture, bureau (nominatif) .
Cette squence de trois consonnes , la tradition grammaticale arabe
la nomme, tardivement, as l , souche .
18
Mais ce nom prometteur res-
tera lettre morte faute dune hypothse sur le rle de la squence quil
nomme, faute de lhypothse dun plan gnral,
19
faute de la prise en
compte du temps .
Aprs la tradition arabe , la tradition orientaliste a vu dans cette
squence la racine de ces formes toutefois sans prendre en considra-
tion ni son rle, ni lorganisation de la langue quelle implique. En cons-
quence elle posera la question du nombre des consonnes composant ces
squences non pas dans le cadre dune organisation globale reconstruite
mais uniquement en diachronie, la recherche donc dun nombre pri-
mitif et non pas dun nombre dtermin systmatiquement ou, sinon,
rfrentiellement : elle cherchera tablir si les racines primitives ont
compt deux consonnes ou trois consonnes .
Ainsi la tradition orientaliste a, dans son tude de la racine ,
pris le temps en compte mais non pas lhypothse dune systmatique
gnrale.
Les deux traditions, arabe et non arabe, accepteront donc des racines
de deux consonnes , de trois consonnes , de quatre consonnes et mme,
parfois, de cinq consonnes . Ce faisant, elles auront empch la recher-
che dune organisation de la nomination . En efet, une telle irrgularit
des signifants de racines autrement comparables exclut la rgularit
essentielle tout plan densemble.
17
Sens secondaires : secrtaire, crivain .
18
Dans le Kitb , as l dsigne le principe (dune loi de la langue), ltat primitif dun
h arf (i.e. dun son de la langue) hors de tout conditionnement ou encore dun schme
qui est donc non anomal. Dans le Sh ib (29), le as l englobe ce qui traite de la constitu-
tion de la langue (mawd
al-lua ), de sa materia prima (awwaliyya ), de sa construc-
tion (mana ) puis des rgles suivies par les Arabes dans leurs communications (rusm
al-Arab f muxt abti-h), des ressources multiformes de lloquence (ifinn) qui sont
leur disposition sur les deux modes rel (tah qq ) et fgur (majz ).
19
Ibn Fris est original par sa proposition, qui semble tre sans autre exemple dans la
tradition grammaticale arabe , dune saisie totalisante de la langue. La langue est, dans le
Sh ib , prsente, dentre de jeu, dune part, dans son plan gnral et, dautre part, dans
sa double relation un Dieu crateur et la crature humaine que ce Dieu a dou dune
parole qui est la matrice de sa pense.
508 andr roman
Dans toutes les langues du monde, les racines apparaissent comme les
squences de phonmes ou, sinon, de syllabes,
20
qui chanent les units
de nomination des entits et des expriences que lhomme a inventes
dans le monde : /katb/, /kitbat/, /aktubu/, /ktib/, /maktab/ . . .
Les racines qui regroupent les entits , les expriences , reconnues par
lhomme comme des units parentes, sont la premire nomination de
ces entits, de ces expriences.
Les racines sont leur premier tablissement dans les langues .
Le premier rle des racines est de leur donner dans les langues une
forme et une dimension telles que les langues puissent les manier.
Le nombre des lments qui composent les racines doit rpondre
cette exigence de maniabilit .
Les racines , condition de leur maniabilit , ne compteront que quel-
ques lments. Cest l une premire rduction, trs forte, impose par
les langues la nomination du monde.
Une deuxime rduction ncessaire est ralise par labstraction qui
aboutit donner, en efaant leurs difrences, des entits diverses un
mme nom commun , des expriences diverses un mme verbe
commun .
Les racines , porteuses de ces sens abstraits , cest l leur deuxime
rle, devront compter, chacune, assez dlments pour satisfaire aux
besoins de nomination des entits et des expriences que lhomme veut
nommer.
Dans les langues smitiques , les racines des noms communs et des
verbes communs comptent, rgulirement, trois consonnes parce
que la combinatoire de trois consonnes est la premire combinatoire
mme de produire en nombre sufsant les arrangements qui seront
leurs signifants.
21
20
Les langues smitiques sont les seules avoir construit leurs systmes de nomina-
tion sur des racines de consonnes . Les langues tons les ont construits sur des racines
de voyelles . Les autres langues, sur des racines de syllabes . Les systmes de nomination
des langues smitiques et des langues tons vont se rorganisant sur des racines de syl-
labes .
21
Manifestement, les racines produites par la combinatoire de deux seules consonnes
sont en nombre insufsant. Les racines produites par la combinatoire de trois consonnes
sont, par contre, en nombre plus que sufsant. Ce sont donc ces racines que la langue
arabe a utilises, rgulirement, dans son systme de nomination. Jamais elle na utilis
dans son systme de nomination de racines de deux consonnes. Les racines de qua-
tre consonnes , secondaires, C
1
C :
2
C
3
C
4,
ont t inventes dans les formes du verbe
racines de trois consonnes et modalit ditration ; le signifant de cette modalit tait
la longueur de la deuxime consonne , C :
2
; cette consonne longue a t rinterprte
mthodologie linguistique : organisation de la langue 509
Ce nombre trois , sufsant, est un nombre trs bas. Il est donc satis-
faisant.
22
La tradition arabe et la tradition non arabe ne poseront pas non plus la
question du statut des autres consonnes qui fanquent ces racines. La
reconnaissance par elles de //, dans /aktub-u/, comme tant la premire
personne du verbe ne touche que le signif de cette unit de nomina-
tion . De mme lidentifcation par elles du tanwn /n/ comme un arti-
cle indfni dans les autres formes cites. Quant lidentifcation de
/t/ dans /kitb-a-t-u-n/ comme un sufxe , de /m/ dans /maktab-u-n/,
comme un prfxe , elle est encore plus imprcise : ces deux conson-
nes , /t/ et /m/, sont, pour elles, des zawid , des lments ajouts , des
formans dans des wazn , des schmes , flat pour /kitbat/, maf al, pour
/maktab/. Mais ces wazn , ces schmes, ne sont que les mimes des formes
quils prtendent analyser.
23
4. laboration en schmes
4.1 Schme 1
Il est incontestable :
que les racines de la langue arabe sont, gnralement, des racines de
trois consonnes ;
que les units de nomination construites sur ces racines reoivent,
gnralement, dans les phrases, une voyelle dsinentielle , brve, qui
est soit le signifant dun mode , dans les seuls verbes , soit le signifant
comme une consonne double ; cette consonne double a, par dissimilation dabord, donn
naissance deux consonnes difrentes ; exemple : /faqqaa/ > /farqaa/, craquer qqc. .
Naissance dun nouveau paradigme. Quant aux racines de cinq consonnes , ce sont des
chimres. Voir Roman , 2005, le chapitre Une brve histoire de la langue arabe .
22
Il est remarquable quune morphologie de larabe toute construite sur des racines
de deux seules consonnes est possible. Mais elle ne sufrait pas aux besoins de la nomi-
nation . Au demeurant la nomination smitique , par racines de consonnes , a succd
une nomination par racines de syllabes , dj nombreuses ; voir A. Roman , op. cit., loc. cit.
23
En ralit, ces wazn , ces schmes sont produits en trompe-lil par le sous-sys-
tme syllabique de la langue. En efet, ses deux seules syllabes , CV et CVC, produ-
isent mcaniquement des squences rgulires ; mais chacune de ces squences nest le
plus souvent que la fgure dun sens global, la fgure dune unit de nomination constru-
ite sur une racine syllabique, cest--dire sur un radical .
510 andr roman
dun cas dans les autres units de nomination (la voyelle /u/ dans
chacun des exemples donns) ;
que jamais une consonne na un signif modal ou casuel.
Ces constatations suggrent un premier schma de la langue arabe :
SCHMA I DE LA LANGUE ARABE
24
CCC V
4.2 Les schmes II et III
Hypothtiquement une langue peut tre dfnie comme lintercon-
nexion de deux systmes interdpendants : un systme de nomination
produisant des units de nomination , et un systme de communication ,
sa syntaxe , dans le cadre duquel ses units de nomination entrent en
relation pour dire une exprience .
Dans le premier schma prsent comme lesquisse dune organisa-
tion gnrale probable, CCC reprsente, videmment, le systme de
nomination ; V , le systme de communication :
SCHMA II DE LA LANGUE ARABE
CCC V
systeme de nomination systeme de communication
La raison de cette rpartition des tches entre les consonnes , seules
pouvoir entrer dans les racines de la langue, et les voyelles , seules
signifer les cas et les modes , se dcouvre dans la structuration du sys-
tme syllabique de la langue : en arabe classique , tout arrangement de
consonnes et de voyelles qui, dans le cadre dune syllabe , ne serait ni
CV , ni CVC , constitue une syllabe a-systmatique produite par
une contrainte phontique ou par une pause syntaxique ; un tel sys-
tme syllabique , , dtermine, dans le fonctionnement de la langue
la disjonction du sous-ensemble des phonmes consonnes , {C}, et du
sous-ensemble des phonmes voyelles , {V} ; en efet deux squences de
syllabes CV et CVC ne peuvent composer les signifants de deux
units de nomination qui, elles-mmes, constituant une paire lgitime,
sopposeraient, en bonne rgle, par lopposition simple dune consonne
24
est pour racine ; C , pour consonne ; V , pour voyelle .
mthodologie linguistique : organisation de la langue 511
et dune voyelle ; ce que montre sommairement le colonage ci-dessous
de deux squences difrentes :
CV C.CV
CV.C VC
dont lopposition {C vs V} est non pas simple mais double.
25
Cette disjonction, ds lors que les consonnes et les voyelles peuvent tre
utilises indpendamment les unes des autres, permet efectivement
lattribution systmatique de tches difrentes aux consonnes et aux
voyelles.
Cest ainsi que sest trouve constitue la structure fondamentale de la
langue arabe , des langues smitiques.
26
Cette raison syllabique trouve, la rgularit de son efet implique
la rgularit du plan gnral qui a t suppos.
SCHMA III DE LA LANGUE ARABE
= {CV, CVC} {C} {V} =
Systme Systme
de nomination de communication
5. Systme de nomination
5.1 Le temps
Dans lensemble des units du systme de nomination , quel lment
choisir qui serait susceptible de lordonner ? Le temps , sans doute.
En efet, le temps est essentiel aux expriences . En consquence il est
essentiel aux langues . Cest grce sa saisie et sa mmoire du temps ,
25
En franais , difremment, les paires { aorte vs porte }, { pote vs porte }
opposent une voyelle une consonne ; en franais, consonnes et voyelles peuvent occu-
per, dans les formes, les mmes positions.
26
Invitablement, les consonnes et les voyelles , disjointes par le sous-systme syl-
labique, sont conjointes dans les syllabes mais les racines slectionnent dans les syllabes
les consonnes qui sont radicales .
512 andr roman
que lhomme a pu crer ses langues . Heureusement, le temps peut tre
saisi, navement, comme une donne foue sans doute mais linaire et
simple.
Lespace , lautre coordonne du monde, est toujours ralis dans des
entits multidimensionnelles , souvent irrgulires, que leur complexit
a mises, demble, hors de porte des ressources dont les langues dispo-
sent mme si telle ou telle forme, telle ou telle silhouette, peut tre ab-
straite en un type : les dimensions trop nombreuses de lespace dbordent
les possibilits des langues.
27
Les langues se sont donc constitues, avec la materia prima de la voix,
dans une premire classifcation, labore sur lopposition rfrentielle :
[+ temps ] vs [ temps]
Elles ont, en consquence, produit, ncessairement, leurs units de
nomination ,
soit comme des units inscrites dans un droulement apparent du
temps , dont le temps est lune des composantes, leurs modus exem-
ple : /ktib/, crivant ;
soit, contre la ralit du monde, comme des units trangres au
temps , leurs res exemple : /ayx/, cheikh .
5.2 Le temps et la racine
Selon toute vraisemblance, le systme de nomination se sera construit,
sur les deux fondements reconnus : le temps et la racine .
En arabe , aucune unit de nomination , res ou modus , ne peut com-
mencer par deux consonnes implosives . Le patron syllabique impose
leur sparation par une voyelle :
27
Le nom commun na pas t dtach de lespace . Il tait concret. Ou, plutt, son
caractre concret, ainsi dsign, par hypallage , est apparu quand lhomme a invent des
entits abstraites , des entits dtaches de lespace et par l-mme du temps , partir
donc, chaque fois, dun modus . Mais ce modus rif, ce nouveau nom, reconnu comme
le nom dune entit noccupant aucun espace, ce nouveau venu, dsign comme un nom
abstrait , par la mme hypallage, na pas t signif comme tel par un nouveau mor-
phme : la langue, pour le discriminer na invent aucun morphme dnotant la prsence
ou labsence de lespace dans les entits nommes par ces noms. Aussi la premire langue
aura-t-elle t vocale, le geste tant une criture sur un espace suggr, sur une donne,
donc, non prise en compte par la systmatique de la langue.
mthodologie linguistique : organisation de la langue 513
*/CC . . . / > /CVC . . . /
28
La premire voyelle des res communes construites sur trois consonnes
radicales , cette premire voyelle que le systme syllabique impose, sem-
ble avoir t le signifant de lanimit .
Exemples :
/kalb/, chien vs /milh /, sel
Nulle voyelle entre les deuximes et troisimes consonnes radicales . Le
genre et le nombre , qui sont les autres dterminants propres aux res ,
sont, systmatiquement sufxs.
Exemples :
/kalb-a-t/, chienne vs /kalb--t/, chiennes
La premire voyelle des modus communs galement construits sur des
racines de trois consonnes , cette mme voyelle impose, tait, est res-
te irrgulirement, le signifant de la diathse, subjective ou objective .
Paradigmatiquement, les modus ne pouvaient se distinguer des res que
par la prsence dune voyelle entre les deuxime et troisime consonnes
radicales . Cest efectivement cette voyelle qui les a distingues des res.
Exemple :
/kalab/, rage vs /kalb/, chien
Cette voyelle, la h arakat al-ayn de la tradition grammaticale arabe , est,
dans /kalab/, le signifant de la prsence du temps . Elle est le premier
signifant du temps . Un temps non prcis.
/kalab/ est, dans la tradition grammaticale arabe , un mas dar .
Le terme mas dar , source, origine , a t choisi parce que certains
grammairiens, les grammairiens dal-Bas ra , les premiers semble-t-il, ont
fait de cette forme, hors plan, la forme mre du verbe.
29
28
Ibn Fris , as -Sh ib , p. 3839 : inna l arab-a [. . .] l t-ajmau bayna skin-ay-ni
wa l t-abtadiu bi skin-i-n .
29
Voir Ibn al-Anbr , Al-Ins f f masili l-xilf bayna n-nah wiyyna l-bas riyyna wa
l-kfyyn vol. I, chapitre XXVIII ; al-Frb , op. cit., p. 6. ; Fleisch , 1979 : 149 ; 1995.
514 andr roman
La tradition na donc pas dfni le mas dar en lui-mme.
Il apparat ici comme un modus infnitif .
Lopposition du modus infnitif , /kalab/, la res anime , /kalb/, est un
exemple de la relation biunivoque , gnrale, l , qui sefectue, dans le
systme de nomination de chaque langue, comme une opposition, vs ,
entre deux paradigmes ; exemple :
2 1 1 2
/katab + t a/ vs /t a + ktub(u)/
Tu as crit vs Tu cris
ou entre deux schmes ; exemple :
/katab + t a/ vs /katab + t i/
Tu as crit (homme) vs Tu as crit (femme)
Dans le systme de nomination , cette relation biunivoque reste une,
car, tant elle-mme opposition , elle ne peut gnrer, par opposition,
aucune autre relation, difrente, qui serait univoque.
5.3 Des racines dune seule consonne
Les racines de trois consonnes ne sont pas les seules racines des langues
smitiques . En efet, ct des units de nomination commune , qui
sont des images difrencies du monde, et sont donc trs nombreuses,
il existe dautres units de nomination non plus commune mais gnrale
ou banale qui sont des images indifrencies du monde, peu nombreu-
ses donc.
30
Le grand nombre des units de nomination commune a impos leur
construction sur des racines de trois consonnes .
Le petit nombre des units de nomination gnrale ou banale a per-
mis leur construction sur des racines dune seule consonne .
Les res banales sont :
la premire et la deuxime personne , remployes, tour tour, dans
chacune de leurs occurrences ; exemples : //, je , de racine , la
30
Banal , qui se met la disposition de tout le monde , Littr.
mthodologie linguistique : organisation de la langue 515
racine de la premire personne ; /ka/, te, de toi , de racine k, la
racine de la deuxime personne ;
les res montres , remployes semblablement, tour tour ; exemple : /
t/, celle-ci , de racine t, la racine des res dostension ;
les res reprsentes , remployes semblablement, tour tour ; exem-
ple : /hu/, lui, le, de lui , de proto-racine *c, la racine des res de
reprsentation.
31
Les res gnrales sont :
les res vagues , exemple : /m/, quoi ? , de racine m, la racine de la
res gnrale ; t, racine du temps gnral , homophone de la racine t
des res dostension ;
32
n, racine du lieu gnral.
33
Les modus de racines monoconsonantiques sont :
le modus /f/, avatar de lancienne racine , p, du modus gnral ;
34
le modus /s/, galement ralis //, faire , avatars de lancienne
racine *c ;
35
le modus /yy/, tre , de proto-racine *c ;
36
le modus d assertion , de proto-racine *m ;
37
31
La proto-consonne */c/, occlusive medio-palatale sourde, se transformera en per-
dant, par lnition, son occlusion : */c/ > *// > // > /s-/ > /h/ > /-/ ; ou, sa sourdit : */c/ >
*/g/ > /y-/ /-yy/. Seules les consonnes qui sont afectes dun tiret sont encore employes
dans la langue arabe historique en tant que pices de son systme. La branche sourde est
termine par locclusive glottale // qui a remplac, contre-courant, la constrictive /h/
dont elle est issue. En efet // est moins difrente que /h/ des voyelles devant lesquelles,
gnralement, elle se trouve. Quant /yy/, cest la mme consonne vocalique /y/, mais
elle a t allonge pour assurer sa dfense contre les conditionnements des voyelles car
elle est toujours, en fn de forme, en position intervocalique ; /y/, difremment, est tou-
jours en dbut de forme.
32
Lhomophonie de certaines racines monoconsonantiques est annule par leurs dis-
tributions difrentes.
33
la difrence de /m/, dans /m-aktab/, de /t/, dans /kitba-t/ (chacun de ces deux
morphmes appartient la forme dans laquelle il se trouve), /n/, est une autre unit
de nomination en fonction dexpansion annective de /maktab/. Unit de nomination
non spcife, /n/ est la tte du paradigme des expansions annectives, spcifes, de la
langue arabe ; /maktab-u-n/ signife, littralement, bureau dun lieu indtermin >
un bureau ; do lidentifcation traditionnelle de /n/ comme un article indfni ;
do son nom traditionnel, tanwn , cest--dire nounation , prsence de /n/ .
34
Exemple dans /kayfa/, Comment ? . Ce modus gnral , concurrenc par les deux
modus tre et faire tait sans emploi ; il a t cr par le jeu des oppositions binaires
constitutif de la langue. Il nexiste que dans /kayfa/.
35
Exemples dans les verbes de paradigmes / + af ala/ et /(i)s + taf ala/ : / + ahadta/,
Tu as fait que qqn tmoigne ; /(i)s + tahadta/, Tu as fait que toi-mme sois tmoin .
36
Exemple : /m + aktab i + yy/, bureaucratique ; littralement : tre bureau .
37
Exemple : */allh - u + m/ > /allh + u/, Allh.
516 andr roman
le modus d invocation , de proto-racine ;
38
le modus d exclamation , de proto-racine .
39
La combinatoire des consonnes dans le cadre des racines de trois conson-
nes est double par une combinatoire de racines .
Exemples :
/ - a - ktub u/, Jcris , construit sur les deux racines et k-t-b ;
/kitbat/, temps dcriture , construit sur les deux racines k-t-b
et t ;
40
/m a ktab/, objet dcriture, bureau , construit sur les deux raci-
nes m et k-t-b ;
/m a ktab i yy/, bureaucratique , construit sur les trois racines
m, k-t-b et *c > /yy/.
Ainsi le systme de nomination exploitait deux combinatoires :
une combinatoire de consonnes radicales ;
une combinatoire de racines associant :
soit une racine monoconsonantique une autre racine monoconso-
nantique ;
soit une ou deux ou trois racines monoconsonantiques une racine
triconsonantique .
5.4 Units non construites sur des racines
Face aux units de nomination construites sur des racines , la langue
a produit des units non construites sur des racines . Ces pices de la
machinerie de la langue, ses modalits , sont dans le systme de
nomination les partenaires des racines. Elles jouent par rapport elles
un rle de dterminants .
la difrence des racines , les modalits sont les images linguistiques
dentits du monde qui peuvent tre apprhendes, navement, non pas
comme des entits communes mais comme des entits universelles , le
temps, la vie . . .
41
38
Exemple : /a + ayx-u/, Cheikh !
39
Exemple : */allh - a + / > /allh + a/, Allh !
40
Exemples dans le verset II :280 : /fa naz irat-u-n il maysarat-i-n/, Alors attendre
le temps [quil faudra] jusqu ce que [le dbiteur] soit dans une aisance [sufsante] .
41
Navement, une entit peut tre perue comme universelle si elle est apparat tou-
jours prsente. Sa prsence constante implique sa prsence partout. Luniversalit ainsi
mthodologie linguistique : organisation de la langue 517
Les modalits de temps sont employes particulirement dans les
conjugaisons . La modalit de vie a t employe comme une modalit
d animit , mle, femelle . . .
42
6. Les res , les modus et le temps
Les langues dtachent les res du temps .
Elles rattachent les modus au temps , or les res nexistent pas hors du
temps,
et les modus nexistent que dans ou par les res.
Les langues ne peuvent donc garder les res hors du temps et elles ne
peuvent davantage garder les modus hors des res . Il faut donc que, dans
le discours, elles accouplent les res et les modus .
Le premier couple {res modus } construit le noyau de chaque phrase
structure.
43
Ses deux composantes sont ainsi relies, lune lautre, ipso
facto, par une relation intrinsque, biunivoque , galitaire : l .
cette relation biunivoque , galitaire, de dclaration du lien , indis-
sociable, entre deux units de nomination,
44
sopposent, ventuellement,
deux relations univoques , les deux seules relations mme dentrer en
opposition avec elle :
lune galitaire, la coordination , + ;
lautre non galitaire, la subordination , n .
entendue lie le temps et lespace, indissociablement. Et elle nimplique avec lespace et
le temps que la vie, qui nat du temps, qui est avec le mouvement dans lespace lautre
mesure du temps. Le temps et la vie . . . sont signifs par des modalits ; lespace nest
jamais dnot que par des noms .
42
Les modalits aspectuelles spcifent difremment non pas la relation des modus
au temps mais, par hypallage , le droulement intrinsque des modus.
43
Les modalits aspectuelles spcifent difremment non pas la relation des modus
au temps mais, par hypallage, le droulement intrinsque des modus.
44
Les phrases non structures sont les onomatopes, units unaires trangres
lorganisation de la langue. Les langues animales sont unaires.
518 andr roman
SCHMA IV DES RELATIONS BINAIRES
[BIUNIVOQUE ] [UNIVOQUE ]
[HIRARCHISE] [NON HIRARCHISE]
DCLARATION SUBORDINATION COORDINATION
l n +
Les trois relations ainsi dfnies, ds lors quelles sont les seules relations
possibles, sont communes toutes les langues .
Ds lors quelles sont communes toutes les langues , le systme quel-
les constituent est universel.
Ainsi le systme de communication sest tabli pour toutes les phrases
de toutes les langues du monde sur le plan suivant :
PLAN UNIVERSEL DE LA PHRASE
{x (+ e . . . ) } {y (+ e . . . ) }
{e (+... )} {e (+... )}
{x (+ e . . . ) } {y (+ e . . . ) }
. . . .
. . . .
. . . .
Dans ce plan, les units de nomination x et y sont, in prsentia
ou in absentia, les deux lments structurellement ncessaires du noyau
de la phrase , ses lments fondamentaux. La relation biunivoque , l ,
qui les solidarise, les constitue en une structure unique : un duo de
deux voix .
Dans ce plan, x , y ..., e , e . . . sont des units de nomina-
tion ventuellement appeles par le locuteur, selon son besoin ; ce sont
des extensions , des complments sans aucune ncessit structurelle.
Un tel systme est simple.
noyau
+
extensions
extensions
+
extensions
mthodologie linguistique : organisation de la langue 519
Sa simplicit tient au trs petit nombre des oppositions en jeu, deux
seulement.
Sa simplicit tient aussi au caractre abstrait des trois relations nes
de ces oppositions.
Commun toutes les langues , ce systme est difrent dans chaque
langue, superfciellement, par les morphmes spcifcateurs propres
son systme de communication .
Le recours ces morphmes est rendu ncessaire par le caractre ab-
strait des trois relations en jeu. Et cest ce mme caractre abstrait qui le
rend possible.
Les morphmes spcifcateurs propres la relation galitaire, biuni-
voque , de dclaration , sont principalement :
La diathse subjective ou objective ;
Le mode rel ou potentiel ou irrel ;
45
Les modaux afrmatif , ngatif , interrogatif , impratif . . .
Les morphmes propres aux relations univoques sont :
les morphmes dfnisseurs de la relation univoque , galitaire, + ,
qui sont les coordonnants ;
les morphmes dfnisseurs de la relation univoque , non galitaire,
n , qui sont les subordonnants , la modalit exceptive.
Ainsi les morphmes qui spcifent la relation biunivoque et ceux qui
spcifent les relation univoques sont complmentaires.
7. Lorganisation de la langue arabe
Ont t constates :
dans le systme de nomination ,
la prsence dune structure radicale dans les units fchies ; lab-
sence de structure radicale dans les units amorphes ;
la prsence du temps dans les modus ; labsence du temps dans les
res ;
46
45
La tradition grammaticale arabe na pas reconnu les modalits de mode.
46
Cette opposition, fondamentale, {res vs modus } nest plus vivante dans les langues
historiques .
520 andr roman
la relation, biunivoque , dopposition, entre les paradigmes eux-
mmes et les schmes qui les composent ;
dans le systme de communication ,
les relations biunivoque , de dclaration ; univoques , de coordination ,
de subordination .
Ces quatre oppositions, binaires , retraces dabord, dans la langue arabe
primitive rvlent un plan de la langue arabe exactement binaire.
47
ARBRE DE LORGANISATION GNRALE PRIMITIVE DE LA LANGUE
ARABE
[SIGNIFIANTS ] [SIGNIFIS ]
[NOMINATION ] [COMMUNICATION ]
[TEMPS ]
[RACINES ] n +
phonmes modus aspects res animit
syllabes
8. La structuration binaire de la langue
La reconstruction prsente de la systmatique de la langue arabe ne
saurait surprendre. Cest cette mme reconstruction quaboutirait une
tude qui commencerait, non plus par lobservation de la langue , mais
par lobservation de lhomme .
En efet, la parole de lhomme est convenue, ncessairement : il ne
saurait parler au hasard.
Si la convention qui rgle sa parole tait brute, cest--dire amorphe,
chaque entit , chaque exprience du monde, serait dite, immdiatement,
absolument, par une onomatope.
48
47
La relation biunivoque , l, est d opposition dans le systme de nomination ; elle
est de dclaration dans le systme de communication .
48
Ouille ! , sil faut un exemple, ne peut que crier une douleur de linstant.
Ouille ! ne peut crier le souvenir dune douleur passe, la crainte dune douleur future.
l
mthodologie linguistique : organisation de la langue 521
Les limites, limprcision, dune telle langue sont manifestes.
Une telle langue ne peut tre la langue de lhomo sapiens , faber.
La convention qui rgle sa parole nest donc pas amorphe mais
structure .
La premire structuration qui existe est la structuration binaire ; au
de, cest--dire hors structuration, le son profr est unaire ; il est
ralis comme une onomatope ; au-del de la structuration binaire la
structuration n-aire (avec n suprieur 2) tait et reste inaccessible
lhomme sans autre recours que son corps. Lhomme a forcment struc-
tur ses langues binairement, ds lors que la structuration binaire est la
premire de toutes les structurations possibles et quelle est la seule quil
puisse matriser.
9. Conclusion
La systmatique de ses langues a t impose lhomme par sa capacit
de combinatoire binaire , son langage naturel , sa premire langue com-
mune , la premire langue de ses conventions.
49
Le plan primitif de lorganisation gnrale des langues humaines
naturelles , ne de la capacit binaire de lhomme , ne peut tre difrent
du plan gnral de la langue arabe .
49
Les langues se sont constitues, par pas de deux , chaque signifant propos
par le systme des sons trouvant, ventuellement, son signif . Exemples de pas
de deux : C
1
V
1
C
2
C
3
est le schma des res ; C
1
V
1
C
2
V
2
C
3
est le schma des modus ;
C
1u
C
2
C
2
R
3
est objectif ; C
1a
C
2
V
2
C
3
est subjectif ; V
2
= /u/, si le sujet nest que le lieu
du modus ; V
2
= /a/, si le sujet produit le modus par son action ; V
2
= /i/, si le sujet
produit le modus par sa raction ; si le sujet produit le modus par son action, la transitiv-
it ainsi ralise est soit afciente, alors V
1
= /a/ ; soit dfciente, alors V
1
= // . . . Jeux
dopposition qui produisent les difrences ncessaires mais aussi jeux analogiques qui
produisent les paradigmes ncessaires et, parmi eux particulirement, les jeux iconiques
qui mlent les analogies abstraites et les analogies concrtes ; exemple : laspect achev est
signif par lordre {C - CCC} des racines ; laspect inachev par lordre inverse, {CCC
- C}.
522 andr roman
Lorganisation primordiale retrouve par la dmarche qui commence dans
ltude de la langue et par la dmarche qui commence dans lobservation
de lhomo loquens est lorganisation binaire , simple et puissante.
La capacit de combinatoire binaire est le seuil de lhominisation .
10. Rfrences
Brague , Rmi. 1992. Europe, la voie romaine. Paris : Criterion.
Carter , Mike G. 1972. Les origines de la grammaire arabe, Revue des tudes islamiques,
40, 6997.
Descartes , Ren. (15961650). 1987. Discours de la mthode, texte et commentaire par
tienne Gilson. Paris : Vrin.
al-Frb (m. 339/950). 1350/1931. Ih s al-ulm, dition tablie et prface par Ut mn
Muh ammad Amn. Le Caire.
Fleisch , Henri. 1961. Trait de philologie arabe, Vol. 1 : Prliminaires, phontique,
morphologie nominale; 1979, vol. 2 : Pronoms, morphologie verbale, particules.
Beyrouth : Dar El-Machreq.
Ibn al-Anbr (513/1119577/1181). s.d. Al-Ins f f masil al-xilf bayna n-nah wiyyna
l-bas riyyna wa-l-kfyyn (publi avec le Kitb al-Intis f min al-Ins f de son diteur
Muh ammad Muhy d-Dn Abd al-H amd), 2 vols.
Ibn Fris (m. 395/1004). 1382/1963. as -Sh ibi f fqh al-lua wa-sunan al-arab f kalmi-
h, d. Mus t af uwaym. Beyrouth: Badrn.
Kleiber , Georges. 2003. Faut-il dire adieu la phrase ?. LInformation grammaticale,
98. 1722.
Lavorel , Pierre Marie. 1980. Aspects de la performance linguistique Contribution
neurolinguistique et psycholinguistique lanalyse des systmes langagiers, Tse dtat.
Lyon : Universit lyon II.
l
CAPACIT BINAIRE
[SIGNIFIANTS ] [SIGNIFIS ]
[NOMINATION ] [COMMUNICATION ]
[TIME ]
[RACINES ] n +
phonmes modus tenses res animit
syllabes aspects
ARBRE DE LORGANISATION GNRALE DE LANGUES
mthodologie linguistique : organisation de la langue 523
al-Maydn Ah mad (m. 518/1124). s.d. Majma al-amt l, d. Muh ammad Muhy
d-Dn Abd al-H amd (2 vols.). Damas-Beyrouth : Manrt Dr an-Nas r.
Meillet , Antoine et Cohen , Marcel. d. 1952. Les langues du monde, nouvelle dition,
CNRS, 2 vols. Paris : Atlas.
Mounin , Georges. 1960. Dfnitions rcentes du langage. Diogne 31, 99112.
Roman , Andr. 1998. Les particules dcomposes ou la reconnaissance des com-
posantes des morphmes de ngation de la langue arabe . In D. Leeman et A. Boone,
eds. Du percevoir au dire. Hommage Andr Joly, Paris : LHarmattan, 8796.
. 2005. La cration lexicale. 2
e
d. Lyon-Kaslik : Presses Universitaires de Lyon-
Presses Universitaires de Kaslik.
Saussure , Ferdinand de (18571913). 2002. crits de linguistique gnrale, texte tabli
par S. Bouquet et R. Engler. Paris : Gallimard.
Schuler , Bernard. 1990. Rhetorica. Rhetorica, 8, 229254, University of California
Press.
Sbawayhi (m. circa 180/786). 13851397/19661977. Kitb, 5 vols. Abd as-Salm
Muh ammad Hrn, ed. Cairo.
Troupeau , Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris : Klincksieck.
Versteegh , Kees. 1977. Greek Elements in Arabic Linguistic Tinking. Leiden : E.J. Brill.
. 1980. Te Origin of the term qiys in the Arabic grammar. Zeitschrif fr arabische
Linguistik, 480, 730.
. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden : E. J. Brill.
. 1995. Te Explanation of Linguistic Causes : Az-Zajjjs theory of grammar Amster-
dam : John Benjamins.
DIALECTS
HOW TO BE KOOL IN ARABIC WRITING:
LINGUISTIC OBSERVATIONS FROM THE SIDE LINE
Gert Borg
Nijmegen University
1. Introduction
Students of Arabiccertainly those of a former generationhave ofen
been led astray by a fatal fallacy: the Arabic language they study, in its
grammar and syntax , shines out as a neat structure of blissful regular-
ity. Tey learn by heart a set of rules with a high rate of predictability,
that govern the language that was used to convey Arab thought and cre-
ativity for 15 or 16 centuries. One might say that just like the Orient
is the dream of the Orientalists , Arabic seems to be the dream of the
Arabists .
But most Arabists are confronted with the harsh reality of the lin-
guistic situation in the Arab world during their frst visit: a reality called
diglossia . It is bitter in a way to have to learn a second Arabic language
in order to be able to survive in everyday situations.
Te actual situation of the foreign learner in the Arab world is not very
helpful either: if he or she gives the impression of being able to speak
some Arabic , the addressee may well decide to speak a higher variety
of the language out of politeness and respect, or start rattling in ver-
nacular assuming that for a foreigner to speak MSA would imply even
more strongly that he or she has an easy command of lower language
levels. As a result the linguistic confrontation between native speaker
and foreign language learner in the Arab world is hardly ever a natural
process.
Nevertheless the gifed and dedicated learner of Arabic can achieve
astonishing results in mastering this complicated linguistic situation .
Diglossia of course has some consequences for the teaching process
of the Arabic language; fundamental choices have to be made. Do we
teach Modern Standard Arabic , one of the modern dialects or some
kind of mix between the two? Must this mix be presented synchronic-
ally or should it be taught diachronically, in this case leaving open the
question with which variety of Arabic to begin.
528 gert borg
Te choices and options become aggravated by growing pressures to
make university programs more time efcient and productive. Andas
if this were not enoughtraditional academic ambitions aim at a wider
study of history of the Arab world , of culture and of religion as well,
apart from the obvious purpose, that students of Arabic Studies are
expected to know their way around in Arabic and in the present day
Arab world.
Although we usually consume Arabic in written form, we soon come
to realize, that the linguistic reality of the Arab world is by no means
as static as the written material suggests: among Arabs new needs and
concepts emerge, new answers and solutions to the diglossic situation
are implicitly or explicitly formulated and this momentumshaped by
the mobility and fexibility of actual peopleis unfolding in a typically
linguistic way: anarchy.
It is obvious, that something static as a language course, a grammar or
even recently collected audiovisual material cannot possibly follow the
volatile dynamics of this process.
2. Ih n, s awt jl bi-h luh
In this contribution I will discuss a single phenomenon that represents
the present stage (August 2005) of this fexibility and anarchy in written
Arabic in Egypt : the monthly Ih n, s awt jl bi-h luh is a glossy maga-
zine in its frst year. It is printed and distributed in Cairo where it sells
for LE 5 (= around 0.75). Te names of the main editors are d arf
al-Jaz r, Jhn Mah m d and arf al-Alf . Te number of contributing
editors (muh arrirn) is 13.
It is not easy to identify the intended readership, but on the basis of
the various contributions and the goods advertised in it (mobile phones,
cars, beer, traveling and leisure) one might assume roughly that the
readers would be between 18 to 30 year olds, well to do and with a wide
range of interests from pop music, sex and vintage cars to (anti-Amer-
ican) politics, recent Egyptian history and backgrounds of hot news
items in Egypt and abroad.
Te general character of this magazine can be qualifed as:
critical open minded
easy to read fairly independent
cultural humorous and satirical
how to be kool in arabic writing 529
Te political sympathy of the editors lies apparently with the Kif ya
popular movement in Egypt .
Te most signifcant feature of this magazine is its use of two lan-
guage levels in various ways: contributions in MSA , others in Egyptian
colloquial (mm iyya) and some with a mixed use of language.
I conducted a basic round of questioning about the opinion of the
reading public and the following image emerged: it is immensely popu-
lar, also in circles that are not intended, like 14 year old school girls and
their mothers, who appreciate the easy reading of this magazine. Some
readers however regret the use of colloquial whichin their opinion
hampers children in learning real Arabic , but they enjoy the reading
nonetheless.
In a lecture at the Netherlands-Flemish Institute the Egyptian lin-
guist Madh a Dos s emphasized, that the editors of this magazine use
Egyptian colloquial , not because they do not know how to write MSA /
fus h they actually seem to publish in MSA elsewherebut that they
choose to use colloquial according to the character of this particular
magazine.
To illustrate the use of modern colloquial in Egypt and to give an
impression of the purposes for which it is used I selected one issue of
this magazine to be discussed here: nr. 8, published in August 2005.
2.1 Fragment 1
1
Let us frst have a look at a contribution that is purely in MSA . It is the
beginning of a contribution entitled: ahr al-irhb al-aswad
2
. . . black
month of terror, about the terrorist attacks in London and the murder
of the Egyptian Ambassador in Iraq :
(yawm al-irhb al-aswad) kna hd huwa l-unwn ar-ras li-jardat
al-ahrm as -s dira s abh yawm al-juma [blank] yliy wa-knat
al-jarda bi-hd al-unwn turu il l-ah dt al-irhbiyya llat waqaat f
l-yawm as-sbiq, h ayt u knat iddat infjrt qad waqaat f qalb al-s ima
al-bart niyya london mdiya bi-h ayt al-aart wa-kna tanz m
al-qida f bild al-rfdn qad alana an qatl as-safr al-mis r f l-irq
hb arf . . . ,
1
By the nature of this contribution many quotes are from texts in which it is not
always clear whether the Arabic should be understood as MSA or mmiyy a. Terefore
the transliteration may sometimes be inconsistent.
2
Cf. pp. 1011. All loose page numbers refer to this specifc issue of Ih n.
530 gert borg
(a summary of al-Ahrm s report of the terrorist attacks in London and
the killing of the Egyptian Ambassador in Iraq) .
Up to this point the choice of language and register may have been dic-
tated by the source that is quoted, al-Ahr m, but afer this the article
continues:
ill anna al-irhbiyyn lam yaktaf bi-hd al-yawm li-l-qiym
bi-hajamtihimi llat stamarrat t iwl ahr yliy mustahdifah (sic: h)
amkin mutafarriqa min al-lam. Fa-qad istayqaz a al-mis riyyn s abh
yawm as-sabt 23 yliy amalan f l-istimt bi-ajzt at -t awra qabla an
yufja bi-anna bos la at-tafrt al-irhbya qadi ttajahat il madna
arm a-ayx al-mis riyya . . . (the author turns to the recent terrorist attacks
in Egypt ).
In his own words the editor continues the language and register as set
by the caption, referring to the terrorist attack in Sharm el-Sheikh . Te
remaining part of this article is more of the same, a language register
that I would qualify as just above the level of average Media Arabic . Te
level even rises a little towards classical in a context in which the editor
compares these erring groups of Muslims ( fraq isl miyya d lla) with
the Xawri j, quoting extensively from H ad t and Qur n .
Te choice of language in this contribution is quite predictable:
commentary on news item o Media Arabic or higher
religious/historic context o Classical Arabic
2.2 Fragment 2
Te Ih n-editor al-Alf conducts an interview with an elderly man who
is fshing from the kbr ag-gm ia. Te photographer he has with him
is named Huseyn .
ajba h usayn f taraddud la, mi hanistann li-h add
e
m samaka tit la.
ibtasam ar-rajul wa-ml li-yafah ant atah wa-axraj minh samaka
ka-annah muidda xis s s an li-t-tas wr wa-abakah f xut t f as-sinnra
wa-rh yahuzzuh h att taz har ka-annah s h iya wa-btilab amm
al-kmr
(Huseyn refuses to wait to shoot his photographs because of the failing
light:) No, were not going to wait till a fsh pops up. Te man smiles
and turns to his bag, producing a fsh as if prepared for the photo, hooks
it to the rod again and starts shaking it till it looks like its alive, playing
for the camera
how to be kool in arabic writing 531
saaltuh d nh h y ustd h asan?. anm. anm?. anm . . bi-
l-mm. amm?. anm. anm . . . ah . . .
(the interviewer asking about the kind of fsh: a very realistic conversation
indeed).
2.3 Fragment 3
A truly remarkable article is wijha naz ar,
3
point of view, a contribu-
tion physically surrounded on its page by a quotation from the Qurn :
rabban l tuxid n in nasn aw axt an . . . (Q. 2/286). Te title of
this contribution is y rabb . . . anta rif (O Lord, You are aware . . . )
qla llh: wa-id saalaka ibd ann fa-inn qarbun ujbu dawata
d-di id dan (= Q. 2/186) y rabb anta qarb minn, lh ana bad
annak? y rabb anta rif kull
e
h ga, wa-rif (q)add
e
h ana muh tglak
wa-rif (q)add
e
h gahl wa-duf byixalln abad annak wa-amil h gt
tzaallak aw mamal al-h gt ell turdk aw amilh bass mi bi-t -t ar(q)
a ell turdk. Ana rif ana (q)add
e
h wih i mak, wa-rif enna l-h gt
wi-niam ell anta addthl akbar min ayy
e
h ga a(q)dir amilh
alanak, lh ba(q) l-whid byistashal ennuh yamil (q)all
e
h ga wa-
huwa fkir ennuh kedah xals amal kull al-mat lb? . . . lh al-duny dalma
kedah quddm al-wh id wa-lh al-h ayt s aba? Ana mi rif wa-mah ad
rif lh kull
e
h ga btih s al f l-duny btih s al kedah lh . . . I feel guilty about
all my wrongdoings in the light of your benevolence, o Lord, and I dont
understand the world anymore . . .
Tis text is very personal in tone, it is a monologue with God, almost
like a prayer and it is in mmiyya throughout, except verbal forms like
turdk.
What we have here is a religiously inspired text, in the context of
Qurnic Arabic , but written for a purpose that isapparentlybest
served in colloquial .
2.4 Fragment 4
H ukkm l yahbna amrk, an article about world leaders who defy
the US and its politics,
4
is also mainly in MSA and Media Arabic . Typical
markers for this second type of Arabic are for instance passages like:
3
Cf. p. 18.
4
Cf. pp. 2526.
532 gert borg
t umma d ahab
5
Castro il qat ar li-tazz al-alqt bayna l-baladayn
xs s atan wa-anna l-alqti llat tarbut qat ar bi-l-wilyt al-muttah ida
wat qa. t umma d ahab il sriy wa-was afa ziyratahu li-dimaq bi-annah
muhimma wa-annahu yaqif il jnib sriy . . . . . . .etc.
. . . then Castro travelled to Qatar to strengthen the relations between the
two nations, especially because Qatars relations with the US are solid.
Ten he went to Syria and described his visit to Damascus as important
and (stated) that he stood next to Syria . . .
Te type of language corresponds neatly to the subject of this contribution.
2.5 Fragment 5
Announced as a grain of seriousness (h abba jadd) we fnd another
contribution in MSA on the daring subject of the role of sex in matrimo-
nial relations: m hiya h udd al-alqa l-h amma bayna l-azwj?
6
(what
are the boundaries of the intimate relationship between married part-
ners?). With captions like an-naz ar il awrat az-zawja (looking at your
naked wife), h urriyyat al-wad al-jasad at n al-jim (freedom of body
positions during intercourse), it does not shy away from daring subjects.
Te whole article is written in MSA except for some anonymous reac-
tions by boys (ray a-abb) and girls (amm l-fatayt fa-tah addat na
ka-t-tl). Tese reactions follow separate patterns:
in the boys section the introductory remarks are in MSA , which
gradually turns into mmiyya. Finally the boys are cited in full
mmiyya:
wa-al l-rami mimm qad yaz unn al-bad min annahum l yatah adda t n
f tilka l-mawd . . . . ad-d ukr arab an rayihim bi-bagh a
7
wa-ka-
annan gn lahum a-ag-garh . qla ah aduhum: ih n makbtn
8
wa-madt n min (!) al-mugtama . . . . although some expected them not
to talk about such subjects, the men expressed their view with impudence
as if we were touching a sore point. One of them said: we are oppressed
by society . . .
5
Te use of the verb d ahaba is extremely rare in Media Arabic : compare the Nijme-
gen Arabic corpus by Jan Hoogland .
6
Cf. pp. 2829.
7
A typical mas dar for mmiyya , not found in classical and MSA dictionaries , but in
Badawi-Hinds .
8
A typical mmiyya word.
how to be kool in arabic writing 533
the girls section is less predictable:
wa-ih d l-fatayt qlat f tamallul: wa-h-yifd bi-h marifat ray f
al-awd ag-ginsya . . . whats the use of knowing my opinion about sex-
ual positions!
wa-law yiz tifam nafsiyyat al-bant mila izzy f l-mawd dah f flm
woman on top fa-huwa yah k qis s at fatt rmnsiyya tazawwajat bi-bb
wasm (if you want to know how womans psychology works in this matter,
than look at the flm woman on top; it tells the romantic story of a girl
that married a handsome boy . . . )
Ten the story of the flm is being told in MSA .
2.6 Fragment 6
A constant mix between MSA and colloquial is a contribution by Yumn
Bassioun i, called al-istina bi-mumat t iln kmbrs min ajl az-ziyrt
ar-rasmiyya,
9
calling in extra representatives at ofcial visits for help,
a hilarious story about the hiring of extra patients to accommodate
the ofcial opening of a hospital wing for children on the 4th foor of
the Ab R Academic Hospit al. Te frst line of this story is indica-
tive for its mixed character: wa-alan uakkid annahu mi mugarrad
kalm, hunka mit ln h adat muak k aran yuakkidn al mawd
al-istina . . . to make sure that its not only a rumour, here you have
two examples that occurred recently, confrming the topic of calling
in . . . .
From a linguistic point of view the use of mmiyya in this article is
interesting: a remarkable colloquial factor is the use of mmiyya con-
junctions instead of MSA ones: apparently this contributes to the fuent
style of writing:
. . . h att az-zar (!) byitl alan yith at t f makn ziyra rasmiyya tnya
(for li . . . )
. . . wa-lk in inn al-mawd yaws al (for lkinna)
. . . wa-alan uakkid . . . (for li . . .)
. . . - zayy
e
m binf f t-tilifzyn(for kam)
. . . kullu deh wa-l-umr m zlat tabd t abiyya li-l-ya . . . (for hd
wa-)
9
Cf. p. 31.
534 gert borg
Te mmiyya can also be used to indicate direct speech without using
quotation marks. Doing this makes the scene lively and realistic, because
we can hardly do anything else than picturing the usual hanger around
in such buildings, all the more so because typically the foor that is said
to be still closed is the foor that was meant to be inaugurated from the
beginning:
. . . f d-dr ar-rbi ell lissa mufatah nuh . . . on the fourth foor that
they didnt open yet.
2.7 Fragment 7
Ih n apparently did some investigative research into the behavior of
female singers on the various MTV-like channels and the impact it has on
the audience in an article with the title man tufaddil min al-muanniyt?,
Which of these singers do you like most?.
10
Some youths give their
commentary in which the choice of language seems signifcant for
emphasizing the point of view. A conservative 23 year old boy says:
masxara al-an ntija an s ah wl al-mujtama wa-laysat al-sabab fh
wa-l-ns mud nibn h ayt u annahum hum yuajjin tilka al-an . . . the
cause of those ridiculous songs is the deterioration of society, not the other
way around; people have to blame themselves because they are encourag-
ing these songs . . .
A 25 year old girl sounds a bit more committed:
al-fdiy klb ba isff wa-axjal min muhadat qanawt al-an amm
wld video-clips are just pulp; I feel embarrassed to watch them when
my father is around
Tis 24 year old girl is very dismissive:
hirt hd az-zamn as bah na yulinna an anfusihinna f t-tilifzyn f
akl fdiy klb the whores of this age have come to express themselves
through the video-clip
And this 27 year old girl sounds critical:
law nrakkaz al urbitn mumkin namil h gt original bass al-Arab
(ih n yan) dyn if we concentrate on our Arab identity we could come
up with something original, but the Arabs that means weare lost
10
Cf. pp. 3233.
how to be kool in arabic writing 535
A 36 year old male:
bs samr law m amalit kedah mh ad hayiuf if Busi Samir wouldnt
act that way, nobody would notice her
Te background of the speakers shifs from:
conservative and intellectual to
sympathetic but dismissive
conservative and denouncing
conservative and nationalistic
realistic
and with these attitudes the language variety shifs correspondingly from:
MSA to
popular (mmiyya)
MSA
street wise (mmiyya)
popular (mmiyya)
2.8 Fragment 8
Walid Irfas contribution is a review of Paulo Coelhos book Te Zahir,
which was popular at the time. Te language level chosen for this article
is close to Classical Arabic .
2.9 Fragment 9
A special page is reserved for reviews of DVDs. Tese are all in MSA . Te
writer, Marwn Qadr , also comments on the Broadcast and Television
Festival in July 2005.
11
On this subject his criticism is harsh: wa-ka-da
kull al-qimn alayhi (sc. al-mahrajn) faal al-mahrajn faalan d aran
min h ayt u t-tanz m wa-l-ifith wa-l-xitm . . . as usual with all organiz-
ers the festival failed completely in organization and during opening and
closing sessions. Only once Qadr deviates slightly from MSA : muz am
ad-duyf wa-l-mutarikn lays . . . , most guests and participants were
(plural) not . . . , but this is far from abnormal in Media Arabic . When
11
Cf. pp. 3637.
536 gert borg
the author voices his indignation about the unavoidable balad/fellh
dancer performing at these occasions he shifs to mmiyya:
l in anh f ayyi h tifl h att wa-law kna t ahr ibn al-jrn wa-llad
yablu min al-umr 7 sanawt. Lh .. lh al-raqs a l-fallh ? H arm alk
zihiqn gaddad, huwa int mi btitfarrag al ayy
e
stirdt aw h aft
xitm aw ifith f ayy
e
h itta barra mas r. Kulluh km wa-l-wufd aw
bi-man as ah h ruas al-wufd km tn . . . there is no avoiding her
at any festive occasion, even if it be the neighbors son circumcision whos
only seven years. Why . . . why this peasant dancer? Shame on you, were
fed up with it, fnd something new(?), its because you havent seen any
festivities of openings or closures anywhere outside Egypt . All that is one
thing, but these delegations or more specifcally the delegation leaders are
another!.
2.10 Fragment 10
In an article with the title bsbr axdar m yisw
12
(a green passport
isnt worth anything) Yumn Bassioun i complains about the frst class
treatment visitors to Egypt receive from the authorities if they carry for-
eign passports. As witnesses to this practice she cites a few personal sto-
ries, parts of which are in mmiyya. One account is by Mohamed Sami
(22) who traveled in Egypt in the winter season with a relative of his
who carries an American passport:
13
ya ibn xlat f amrk wa-ndiran m yat il mis r . . . . . wa-kunn
rjina min xarga
14
bi-l-layl mutaaxxir(!) jiddan . . . wa-awqafan ad-
d bit wa-kna kik(!) fn li-anna ibn xlat kn lbis rt f izz al-bard!
wa-awd ah a muh ammad anna ad-d bit akka f annahum mutat n
h ga muxaddara muxalliyhum mi h ssn bi-nafsihim wa-lbsn kedah,
xs s atan wa-anna lukna qarbih f l-arab mi salma wa-s ammam ad-
d bit annahum yat la al al-qism. al-h ga l-wah da ell anqadatn
annan gibn li-d-d bit gawz safar ibn xlat al-amrk wa-hadadnhu
annan hanatakh f s-sifra l-amrkya an al-bahdala ell h as alit/h as alat
la-n wa-bi-t-tl mumalat ad-d bit itayyarat tamman wa-tad ar kitr
alan m namil makil! . . . yan ibn xlat lamm kn byistamal
al-bsbr al-mas r maa dalika d-d bit , m la r s mumla wa-awwal
m az har bsbruh al-agnab kn lahu al-ihtirm a-add wa-l-adab f
l-mumala
my cousin lives in the US and rarely comes to Egypt . . . we were return-
ing from Kharga very late at night . . . and a policeman stopped us and
12
Cf. p. 41.
13
Passages in colloquial are in italics.
14
Written is xurga which seems improbable.
how to be kool in arabic writing 537
suspected us because my cousin was wearing shorts in this extreme cold!
Mohamed made it clear that the policeman was wondering whether they
had taken some drug that lef them not feeling themselves while they were
dressed that way, especially because the accent of his relative in Arabic was
not ok. Te policeman decided that they had to go to the police post. Te
only thing that saved us was that we handed the policeman my cousins
foreign passport and we threatened him that we would complain about
him at the American Embassy about the illtreatment that we encountered.
All of a sudden the policemans behavior changed completely and he apol-
ogized deeply so that we wouldnt cause him any problems! So when my
cousin used his Egyptian passport (in dealing) with that policeman, he
encountered maltreatment and as soon as he waved his foreign passport
he was treated with deep respect and decency.
In this passage the story changes rapidly from account to direct speech
and the language changes accordingly.
2.11 Fragment 11
In the frst lines of the contribution tanfsah
15
(reassurance?) we fnd a
remarkable blend of MSA and mmiyya:
Kun Anta. Ql iraf nafsak dah ahamm wa-as ab wa-akt ar al-ns
l yarifn!! Amm d h ga arba kull
e
m inn abus s f bit qat arif ana
mn wa-ahdat at-taxarrug wa-ahdat ag-g wa-al-bsbr . . . Be your-
self. Tey say: Know yourself. Tat is the most important and the most
difcult thing, while most people dont!! Tat is something strange: every
time I look at my ID I know who I am . . . , my degree, my army card and
my passport . . .
But then the mm ya takes over almost completely:
Huwa deh ana wa-l deh elli itfarad alayy enn aknuh . . .
Amil ell inta yizuh . . . yiz arabiyya bi-rmt bitru mayyah (unz ur
namd aj 2)
16
yiz tilab mazk! Yll nirh ri muh ammad al bukra
mumkin yigibak gtr wall darabkah . . . yiz tiqr!?wa-llhi fkra mi
bat t laindin maktabt fh kull al-mawdt ell titxayyilh, mamak
fuls?! inzil sq
17
al-ezbekya fa (!) al-ataba . . .
Tat is me and not the one who imposed on me that it is me . . . Do what-
ever you want . . . You want a car with remote control that sprays water (see
15
Cf. p. 42.
16
Referring to a photograph.
17
Te text reads sr, but that seems improbable.
538 gert borg
example
18
2). You want to play music!
19
Lets go to Muhammed Ali Street
tomorrow. Maybe youll like a guitar or a darabukka . . . you want to read
sure, not a bad ideawe have bookshops with all subjects that you can
imagine, you dont have money?! Go to al-Ezbekiyah and Ataba .
In this fragment MSA is used for the general introductory remarks; the
more subjective fctitious monologue is in colloquial .
2.12 Fragments 1216
A few remaining articles about (muscle) cars, sporting exercises, qual-
ity food, gadgets and the unavoidable horoscope are all completely in
MSA.
20
2.13 Fragments 17 and 18
In a magazine that is so obviously playing with the possibilities that lan-
guage levels ofer it is only natural that the reader would fnd contribu-
tions about language itself. In this issue of Ih n we fnd two, both under
the common and signifcant title Arbk slng.
21
We fnd two lists of
say and dont say items (qul and l taqul), here represented as
opposites:
Say Dont say
22
Say Dont say
mat ar nat ar dafar dfa/ir (?)
silsila sinsila muhandisn muhandizn
gurnl gurnn bsbr bzbr
bart amn bat ramn
23
isbniy izbniy
fngn fngl dalwaq/t dawaq/t
And under the heading qms Ih n (Ih ns dictionary) we fnd a qua-
si-classical explanation of the word sakalns
un
(sic):
kalima taxdum jam al-ard al h asb al-man l-murd: yumkin
istixdmuh bi-man kullih f t-tamm kulluh f s-sakalns, wa-yum-
kin al-ira li-l-by frnd aw al-grl frnd al annahum as-sakalns.
18
Example refers to a photograph accompanying the text.
19
Mark the calque for azaf.
20
Cf. pp. 4450.
21
Cf. p. 6.
22
Tese lists seem to be provocative to me; please go ahead and say gurnn!
23
Badawi-Hinds mentions both but apparently prefers bat ramn, a glass or plastic
jar with a lid, a jam jar.
how to be kool in arabic writing 539
kam yumkin al-qawl bi-anna axs (m f t-tarwa?) bi-annahu f
s-sakalns. s-sakalns tustaxdam f l-qah w li-tasmiyat al-kktlt:
qahwa al yzabd bi-l-fawkih . . . lx. f l-mat im, as-sakalns tur
il sndwitt al-fl bi-t -t amya bi-l-bayd aw sandwitt as-samak bi-g-
gambar bi-s-sabt (?) bi-t -t ah na wa-llat turaf ayd an bi-sandwitt
al-fygr. sakalns tams r li-kalima aksalns . . .
an all purpose word depending on the intended meaning: it can be
used as a generic term: its all sakalns. It can refer to your boy friend
or girl-friend: they are sakalns. One can also say that someone . . . (?) is
in sakalns. sakalns is being used in cofee shops to indicate cocktails:
cofee on tea, cream on fruit etc. In restaurants sakalns refers to bean
sandwiches with t amya and egg or fsh sandwiches with shrimps with ?
with t ah na that are also known as viagra sandwiches. sakalns is an egyp-
tianisation of the word excellence.
A wonderful play of words and styles.
3. When is mmiyya being used in this magazine?
Te use of mmiyya seems neither inconsistent nor indiscriminate in
this magazine; I will try to clarify the preference of the contributors for
using mmiyya in its various contexts:
1. mmiyya is used for representing direct speech and to add a realistic
favor
la, mi hanistann li-h add
e
m samaka tit la (fragment 2)
also for fctitious direct speech : yiz tiqr!?wa-llhi fkra mi
bat t la(fragment 11)
2. for adressing the reader
zayy m binf f t-tilifzyn (fragment 6)
3. for contexts with humorous, hilarious or ironical purposes
f d-dr ar-rbi ell lissa mufatah
nuh (fragment 6)
h att z-zar byitl alan yith at t f makn ziyra rasmya tnya
(fragment 6)
4. for intimate and personal expression
fragment 3: the prayer
5. for clarifying the social status or political point of view of the speak-
ing individual
the interviews in fragment 7
540 gert borg
6. for emotive contexts, like anger or indignation
the way indignation that is expressed about the baladi dancer in
fragment 9
fragment 5: ih n makbtn wa-madt n min al-mugtama
4. Orthography
It is obvious that for an Arabic periodical in most Arabic speaking
countries to use written mmiyya is like walking a minefeld: it will be
frowned upon by the conservative cultural and intellectual elite and the
periodical might very well be unable to penetrate a wider Arab market
although in the case of Egyptian colloquial this risk is limited. It opens
however a thrilling space of new and unexplored opportunities. It has to
be said though that this experiment faces some practical problems, one
of which is orthography .
For Classical and Modern Standard Arabic the orthography is fairly
straightforward. Even in written Media Arabic the orthographical prob-
lems are mainly confned to transliteration of loan words and foreign
names. But for written colloquial the situation is completely diferent:
the Arabic graphemes do obviously not cover the mmiyya stock of pho-
nemes and no consistent convention has been developed yet. If this trend
of writing colloquial will persist, it would mean, that new orthographical
standards will emerge. It will be hard to continue the standard of one-
to-one equivalents as in written CA and MSA . For written colloquial
the orthography of the future might well be less stable and predictable.
In a few instances we can already get an impression of some of the
difculties: do we spell as we speak or spell as the conventional spelling
of MSA tells us to spell: for example the (ayn ) is clearly disappearing
from the urban Cairene dialect . As a consequence Ih n writes byarif
for the MSA yarifna. To do this on the cover even seems to be meant
as an elegant provocation, because the same phenomenon ( for ) can-
not be witnessed in the remaining articles except for the article that the
cover announces: istif ih n: 95% min a-abb byarif (with fnal
alif ) man bs samr . . . inquiry by Ih n: 95% of the kids know who
Busi Samir is (fragment 7).
In other issues of Ih n, not quoted here, I noticed a tendency to
represent stressed vowels by lengthening them, using matres lectionis ,
which may be confusing for the unaccustomed eye.
how to be kool in arabic writing 541
Furthermore we fnd inconsistencies for instance in prepositions with
the frst person sufx : liy, alayy, but biyyah.
Te conjunctions wa and fa are increasingly represented and
therefore considered as separate words, not only in advertisement, but
also within running contexts like these: kull an-ns knit rfa mas lah t
aktar minn: mm wa bb, al-mudarrisn, . . . everyone knew better
than me what was in my interest: ma and pa, the teachers, . . . (from
fragment 11); inzil sq al-ezbekya fa al-ataba . . . go to the Ezbekiya
market or Ataba (fragment 11).
5. Conclusions for teaching
For the present and coming generations of students of Arabic it might
be important to become acquainted with the notions and fashions of
their own generation in the Arab world . For them to learn about dif-
ferences and similarities will probably contribute to motivation and
mutual understanding.
Tese notions and fashions are for a large part expressed in language.
Te present state of the Arabic language and its day-to day develop-
ments cannot however be covered by course books and vocabularies
because these cannot possibly keep up with the tempo of modern urban
life, the media or the internet.
Te pace of changes in the Arab world as refected in the quick devel-
opment of the written Arabic language confronts us, teachers of the
Arabic language and culture, with the question whether we can still
convey the concept of the written Arabic as a continuum; or should we
actually split the current language acquisition curricula into the catego-
ries classical and modern? Te implications of options like these are
far reaching.
Tere is at least one practical consideration: teachers of modern Ara-
bic can hardly do without regularly being in touch with the Arab world
in order to adequately follow the linguistic developments in the Arab
world. Tis can be achieved either by following the (satellite) media
orbetter stillby regular visits to Arab countries to witness the fast
developments frsthand.
542 gert borg
6. References
Hinds-Badawi: Hinds, Martin and El-Said Badawi . 1986. A Dictionary of Egyptian Ara-
bic: Arabic English. Beirut: Librairie du Liban.
Ih n, s awt jl bi-h luh. 2005. August issue.
Nijmegen Arabic Corpus: a collection of MSA texts from the written media compiled by
Everhard Ditters and Jan Hoogland .
HELLO, I SAY, AND WELCOME! WHERE FROM,
THESE RIDING MEN?
ARABIC POPULAR POETRY AND POLITICAL SATIRE :
A STUDY IN INTERTEXTUALITY FROM JORDAN
1
Clive Holes
University of Oxford
1. Introduction
In Arab literary studies, popular poetry , that is, poetry composed in a
non-standard form of the language, remains a relatively unexplored res-
ervoir of creative activity. Tere has been a tendency for native and west-
ern critics alike to ignore it, or at best pigeonhole it as folklore, devoid
of literary value, and written in a debased form of the language. One
of the few great Arab writers to stand out against the prevailing opinion
was the 14th century historian, sociologist and polymath Ibn Xaldn ,
a writer who in this, as in so much else, was sui generis. His words are
worth quoting, since he puts his fnger squarely on the main reasons for
the prejudice against popular poetry (ignorance) which applies with as
much force today as it did six centuries ago:
Most contemporary scholars, philologists in particular, disapprove of
these types (of poems ) when they hear them, and refuse to consider them
poetry when they are recited. Tey believe that their (literary) taste recoils
from them, because they are (linguistically ) incorrect and lack vowel end-
ings. Tis, however, is merely the result of the loss of the habit (of using
vowel endings ) in the dialect of the (Arabs ). If these (philologists) pos-
sessed the same (speech) habit, taste and natural (feeling) would prove
to them that these poems are eloquent, provided that their (own) natu-
ral dispositions and point of view are not distorted. Vowel endings have
1
Te present paper is based on data gathered in the course of a feld-based investiga-
tion into the practice of Bedouin poetry in Jordan and Sinai , involving the collection of
poetry composed over the last 50 years, and particularly over the last 20, direct from the
poets themselves. Te feldwork was carried out by Dr Said Salman Abu Athera , a Bed-
ouin originally from Beersheva but now resident in Jordan , and the translation, glossing
and annotating of the poems was done by me. Our book Y Kundlzza Rys ! Politics
and Popular Poetry In the Contemporary Arab World is to be published by Brill in 2007.
544 clive holes
nothing to do with eloquence. Eloquence is the conformity of speech
to what one wants to express and the requirements of a given situation,
regardless of whether the u -ending indicates the subject and the a -ending
the object , or vice versa. (Ibn Xaldn , tr. Rosenthal 1958, 3:4145).
With the beginnings of western interest in Arabic literature, from the
mid-19th century until the early decades of the 20th, there was a good
deal of academic efort devoted to collecting examples of Arabic popular
poetry , but with a mainly philological rather than ethnographic or liter-
ary inspiration (e.g. Wallin 1851; 1852; Socin 19001; Landberg 1901;
Meissner 1903, though Musil 1928 is an exception). In recent years there
has been a fresh burst of interest, particularly in the contemporary Bed-
ouin poetic tradition (Sowayan 1985; Abu-Lughod 1986; Ingham 1986;
Bailey 1991; and Kurpershoek s monumental fve-volume study of the
Bedouin poetry of Najd , 19932005). Tese later studies have focused
on the practice and social functions of poetry , and on its ethnographic
value as both the product and record of Bedouin societies and modes
of thought that are now rapidly passing into history. Tere is evidence
to suggest, however, that the tradition of popular poetry in todays Bed-
ouin society is far from dead; on the contrary, it seems to be transform-
ing itself into a voice of socio-political commentary and criticism which
transcends purely local tribal concerns and addresses regional and even
international issues. During the 20th century, Arabic popular poetry
was frst deployed by urban poets , in the absence of any other public
forum for protest, as a potent political weapon, the two prime examples
being Mah md Bayram at-Tnis s barbs against Egypt s British colo-
nists during the early part of the century (Booth 1990), which led to his
expulsion from Egypt , and his heir, Ah mad Fud Nigm s coruscating
attacks on Anwar Sadat and the policy of economic and cultural infth
pursued by the Egyptian government during the 1970s (Abdul-Malek
1990), a campaign which landed him in prison on several occasions.
Popular Bedouin poetry , on the other hand, has until relatively recently
been little concerned with politics and society outside those of the tribal
dra ; only with post-Second World War political independence, and
a perceived encroachment of the state on the concerns of their com-
munities have Bedouin poets gone down the same path as their urban
counterparts and turned their poetry into an instrument for expressing
communal popular protest.
But what is meant, in the early 21st century, by the term Bedouin in
the phrase Bedouin popular poetry , since virtually no one now leads the
arabic popular poetry and political satire 545
Bedouin life of nomadism ? As a statement that today refers to member-
ship in a genealogy-based social structure, rather than to a way of life,
ih na badu , or more frequently ih na arab, is still a proud boast. Individu-
als whose families may have been settled for many generations continue
to use this phrase to identify themselves, however far their life-style has
deviated from the Bedouin stereotype of yesteryear. In countries like
Jordan , Iraq and Saudi Arabia , the concentric circles of family ties, clan
membership and tribal afliation into which self-proclaimed Bedouin
are born continue to exercise a strong pull on their imagination and pro-
vide the backdrop against which their social relationships are conducted
and political attitudes forged, even in large cities.
2
Contemporary Bedouin poetry is composed in a stylised form of
the poet s spoken dialect , but this does not mean that it is a purely oral
form, or that the poet has had any formal training. Poets come in all
shapes and sizes, and not from any particular class or background: he
(or she, as there are some skilled female poets ) can be a tribal elder, a
supermarket owner, a smuggler, or a doorman;
3
what counts is a gif for
composing, honed through years of listening to other poets . Te oldest
generation of poets now in their 70s and 80s are almost without excep-
tion illiterate, and use a vocabulary and forms of poetic diction now
difcult for younger generations to understand.
4
Nowadays, however,
many of those who would describe themselves as Bedouin poets , like
the one discussed here, are literate in Modern Standard Arabic , but, for
a variety of audience- and topic-related reasons, some of which I will
discuss below, choose to compose in dialect , and use a vocabulary closer
to that of everyday speech. Written versions of their poems not infre-
quently appear in newspapers and are published in locally and cheaply
produced dwns .
2
In Egypt , in the mouths of the settled farmers and town-dwellers of the Nile valley,
arab is a term of opprobrium, signifying backwardness and stupidity; but for the Sinai
Bedouin , resentful of the power over their lives which the settled population of Egypt
now exercises, these farmers and town-dwellers are contemptuously referred to as ban
frn the sons of the Pharaoh, an insult that alludes to their supposed slave origin and
habit of subservience.
3
Poetry was collected from poets ftting all these descriptions in the course of the
present project.
4
A good example is the poetry of the illiterate Unz Ab Slim at-Turbn of south-
ern Sinai , considered by many the premier Bedouin poet of the region, who spent many
years in Egyptian prisons for smuggling ofences, and died in his 80s in 2000.
546 clive holes
2. Ghassn Surr a-baylt ( Ab Surr)
One of the Jordanian poets from whom we have collected a large number
of unpublished poems , though many of them have circulated by word
of mouth and on cassette, is Ghassn Surr a-baylt (Ab Surr) .
Ghassn was born in az-Zarq , north Jordan , in 1954, the son of the
paramount sheikh of the al-Uwayst section (ara) of the Ban H asan ,
a north Jordanian tribe . He now lives in a large house in al-Mafraq . He
describes his social background with the tongue-in-cheek epithet ladu:
an abbreviation of l flh wa l badu neither cultivator nor nomad, a
term used to describe people historically of Bedouin origin who are now
fully settled. Ghassn was the eldest son in his family, with seven broth-
ers. When his father died, he was around 14 years old. His father lef him
virtually nothing (300 JD), having sold of or squandered almost all the
family wealth. As a result, Ghassn was forced to leave school immedi-
ately and fnd work to support his siblings and mother. At that time, in
the late 1960s, the Iraqi army still had units in Jordan following the 1967
war with Israel . Tis presented opportunities for smuggling. At that
time the Iraqi army was using Russian weapons and ammunition, but
the Bedouin in Iraq still carried British weapons from the days of Brit-
ish infuence before the revolution of 1958, just as the Bedouin of Jordan
still did. When Iraqi units rotated home, the young Ghassn would sell
them British weapons and ammunition, which they smuggled across the
border and sold on at a proft to Iraqi Bedouinthere were no border
checks on troop movements. Tis ammunition and weapons smuggling
subsequently broadened into a more extensive illicit tradecigarettes
and other goodsand extended also to Syria , which was a much nearer
border. By the early 80s, as a young man, Ghassn had begun to travel
regularly to West Germany and the countries of the eastern bloc, from
where he brought back cars and other goods to sell in Jordan . Eventu-
ally, the profts from this enabled him to buy land and start businesses,
including supermarkets. He is now, at around 50 years of age, a rela-
tively wealthy man. Tat is the bare bones of Ghassn s journey from
childhood rags to adult riches. He is at pains to point out that none of
what he owns and has achieved was through family infuence, but rather
through thirty years of his own eforts and honesty with business part-
ners: what he has achieved has been, as he put it, bi s -s adga wa l bi
l-garba (through friendship and not through family ties).
arabic popular poetry and political satire 547
Ghassn is a complex character. Despite his lack of formal educa-
tion he has read quite widely, and is in the habit of quoting Plato in
his conversation. It became obvious from talking to him that he has a
thorough knowledge of the history of the Levant . He also, despite (or
perhaps because of the means by which he acquired) his wealth, has a
well-developed social conscience. Tis is strongly refected in his poetry ,
most of which was composed during the 1980s when he was struggling
to make a living and continually coming up against economic and social
barriers. His constant poetic targets are: the economic inequalities of the
Jordan of this period; the incompetence, hypocrisy, cupidity, corruption
and arbitrary powers of government ministers and ofcials; the nepo-
tism and snobbery endemic in Jordan at all levels; and perhaps most
emphatically, the loss of the sense of honor and dignity which, in his
view, no longer underpins the fabric of Jordanian society in the way it
once did. All of these he had to battle against to achieve what he has.
Te satirical poem presented here, as well as many others, could not
be published in Jordan , and the poet ran into serious problems with
the intelligence services at the time when his poetry frst began to cir-
culate informally some twenty years ago. How did, and does, Ghassn s
poetry circulate? Most ofen a fnished poem or poems would be recited
in a gathering in the poet s majlis, and recorded there on a cassette. Tis
would be copied and distributed to local shops and sold for a very low
price. Bus drivers would ofen play such cassettes to their passengers. Te
fax machine, when it became common in the 1980s, was another means
of circulation of the poem in written form. Te latest method is via text
messaging on mobile phones. Although Ghassn still composes poetry ,
he has turned away from the provocative and ofen personal attacks
of the 1980s, and stays behind what he describes as al-xut t al-h amr
(the red lines) that must not be crossedin particular anything that
could be construed as critical of the Hashemite royal family . His mission
remains, however, to speak up for the little people of Jordanian society,
in particular the Bedouin , whose voice and concerns can rarely be heard
above what he describes in another poem (al-Jarid the Newspapers)
as the sycophantic blather of the government-controlled media.
Ghassn commented thus on the poem presented below:
A famous Jordanian song, popular in the 1960s, was y marh ab, y hal,
minn ar-rakb, minn? Te song praised the heroes of our nation and
encouraged manliness (rugla), gallantry (ahma), self-respect (karma)
and freedom (h urriyya). But peoples ideas have changed, as have their
dealings with each other. People have become so preoccupied by earning
548 clive holes
a crust and making ends meet that they have no time to think about any-
thing else, and if they do they get distracted and lost.
3. Rashd al-Kln in Abdh Mss musical setting
Te words of the original song (Ghawnmah 2002, 1501) are by Rashd
al-Kln . A famous Jordanian gypsy singer and rebec player of the time,
Abdh Ms (19271977), made them famous by putting them to
music. Ms was well known for his popular nationalistic songs prais-
ing the army, national heroes, and the Hashemite dynasty , as well as
sentimental ditties for religious occasions. Ghassn takes many of the
original lines of Abdh Ms s song and, while preserving the original
meter and rhyme scheme , alters the words to provide a series of poetic
snapshots of the state of Jordan in the early 1980s. Te efect of insert-
ing such material into the structure of the song, whilst at the same time
preserving much of the original bombastic phraseology , is pungently
satirical and bathetic. Te impact of Ghassn s poem on a local audience
depends precisely on the fact that Ms s song from an earlier, more
confdent period of Jordanian history was so well known to ordinary
Jordanians from its frequent airing in the local media.
Rashd al-Kln s original words are as follows:
5
1 y marh ab, y hal, mnn
ar-rakb, min wn?
6
Hello, I say, and welcome! Now
where from, these riding men?
agbal aln d-dih y zna
gblah
The mornings come upon us,
see how beautifully, again!
2 h inn dr
7
al-id, t allba li
d-dn
Against the foe
8
were ruthless;
our debts we neer disclaim,
5
Each verse of the poem is divided into two hemistiches , all the opening hemis-
tiches as a group, and all the closing hemistiches as a group, being metrically identical,
and each group having a diferent rhyme , -n (or -n) for the opening hemistiches and
-lah for the closing ones. Tis is a common arrangement. Unfortunately it is almost
impossible to imitate in an English translation, so I have rendered the poem into English
rhyming couplets , with fourteen syllables to each hemistich , and tried to be as faithful as
possible to the meaning and tone of the original.
6
A standard Bedouin greeting.
7
dir pl dr ruthless, remorseless.
8
Lit aggression < CLA ad.
arabic popular poetry and political satire 549
wa l-gr m yigbalah illa
r-rad xlah
For only he accepts abuse whose
menfolk have no shame.
3 xd il-marik lin min ymn
s alfn
9
The fray we enter swaggering,
our heads we hold up high,
wa l-hm d
illan wa r-rh
fadw lah
The Hashemites provide our
shade
10
for them we live or
die!
4 wa t -t r krin
11
lina, y t rn
bi t rn
Revenge is second nature; for
each wrong we take back two,
y s bin h aggan l budd m
nnlah
From any who usurps our
rights, we wrest them back
anew!
5 nizh af ala ll ba wa xn al-
ahd wa d-dn
We march against oppressors
who break their word and creed,
wa nds ala ll t a wi nigzh
b af lah
We stamp upon the tyrants
head, and punish his misdeeds!
6 h ayyhum nim l-wat an,
h ayyhum gund ih sn
So salute our homeland brave
hearts! So salute them, Hussein s
pride!
rab il-kaff l-h umur wa
l-ugul mayylah
Theyre the ones who wear
red head-cloths, head-ropes
slanting to the side!
7 y mgana
12
bi n-nay
13
kaf
dum il-n
O wife who grieves in widows
weeds, weep not, nor solace
seek,
kaff dum il-as a l-xadd
sayylah
Weep no more tears of sorrow,
no tears trickling down your
cheek.
8 g k n-nim, laf,
14
wi
tbair y zn
Our gallant lads have come to
you, theyre here, be of good
cheer!
9
salf pl salfn brave, swaggering. Cf. Musil 1928, 561 salf a strong gust of wind, Clas-
sical Arabic salif vainglorious, boastful.
10
I.e. protect us.
11
kr pl. kiyr custom, habit. Cf. Bailey 1991, 451.
12
mgana woman wearing a guna = womans black head covering reaching down to
her behind.
13
nay sadness, dejection. < Classical Arabic n-w-. Cf. na bi l-h iml to groan under
a burden.
14
lafa (yilf) come, arrive.
550 clive holes
igbn fg il-higin, w usd
xayylah
Like eagles riding camels, like
fierce lions, those cavaliers!
9 add l-bandig zurug
rah saffn
15
They wield machine guns,
double-clipped, the metal
glinting blue,
u mudarratin uhub nrn
gattlah
And armoured cars, their colour
grey, death-dealing bullets
spew!
10 madf mwallafa, tirm ala
l-gln
16
Their guns co-ordinated, and
they let fly to each side,
s aly
17
l-qanbil raad u burg
alah
A thunderous volley of shellfire,
it lights up the countryside.
11 wa nsr gu ww s-sam
mitlaf a bi l-n
18
And vultures
19
hover in the
sky, wrapped round by cotton
clouds,
tixw
20
ala l-mitad lil l-mt
ayylah
They plunge down on the
enemy, and bring him his death
shroud!
12 tixw ala l-mitad mixlbh
bi h addn
They plunge down on the
enemy, their claws as sharp as
nails,
m d g t am as-salma min
nis at glah
And none who dares to
challenge them has lived to tell
the tale.
21
13 aswad s abh al-id ymin
alhum n
Black morn for the aggressor,
the day him ill befell!
u ubr yim d
af il-galb an
h lah
By dust his weak heart blinded,
his fate he could not tell!
14 min kull dabbbtin ummt
ganzrn
There comes from every battle
tank, twin caterpillar-tracked,
tuhdur hadr il-bah ar, a
l-gm
22
s ayylah
A roar deep as the oceans,
gainst the foe they launch
attacks!
15
saf cartridge magazine or clip of a frearm.
16
g l pl. gln side, direction.
17
saly blazing volley (of shots, cannon fre).
18
Sc. m clouds. Te change to n is a poetic licence to maintain the rhyme .
19
I.e. fghter pilots.
20
xawa to plunge, dive.
21
Lit: None tasted safety that went in their direction.
22
Normal Bedouin term for enemy.
arabic popular poetry and political satire 551
15 u fhd bn il-h ufar tuhgum
ala rigln
And darting tween the battle
lines
23
two-legged cheetahs
growl,
ganns a li l-id w inmr
gawwlah
And hunt down the aggressor,
just like panthers on the prowl.
16 h ayyhum nim l-wat an,
gu nd ih sn
So salute our homeland brave
hearts! So salute them, Hussein s
pride!
rabi il-kaff l-h umur wi
l-ugul mayylah
Theyre the ones who wear
red head-cloths, head-ropes
slanting to the side!
4. Text comparison
In his version, Ghassn takes the sentimental nationalism of Ms s
refrain:
h ayyhum nim l-wat an, So salute our homeland brave
h ayyhum gu nd ih sn hearts! So salute them, Hussein s
rab il-kaff l-h umur wa l-ugul pride!
mayylah Teyre the ones who wear red
head-cloths, head-ropes slanting
to the side!
(line 6 of the original, repeated in line 16)
and transforms it into a variable refrain of his own in which he in turn
ridicules Jordan s political leaders:
h ayyhum iyx il-arab il So salute our Arab leaders, on their
takstuh
24
bi antnn cars two aerials fxed!
rab it-taks l-h umur wa Teyre the ones who drive red
d
-d
-d
bi l-bn
26
with our cruel fate content,
wa l-kull rd
l-gr h att
Each one of us accepts abuse, even
-aham xlah
27
those of high descent.
wa l-amr tanfd ah lin u min We do as we are ordered to; its
drn xann not our place to ask,
wa l-gir sr ar-rabb wa r-rh Money rules our lives; our souls are
fadw lah mortgaged to that task.
wa s-samt krin lin wa Keeping quiets second nature; we
a l-h agg m nin dont stick up for peoples rights,
wa l-h agg min il-mursil
28
Our rights we beg from minions,
wallah m nnlah then give back without a fght.
(lines 14)
Te symbolic widow of lines 78 of the original, weeping for a husband
killed in battle, and comforted by the arrival of the army of the nim
(gallant lads) of the army:
25
< y m ab igblah lit: how awful its coming!
26
= w illi nartadi bi l-bn those who are content with adversity.
27
Ironically echoing the sentiment of the popular Bedouin saying dawwir li wil-
dik xl fnd a maternal uncle for your son, i.e. fnd a brave man and marry his sister
because the bravery will run in the family. Nowadays, the poet is saying, the once noble,
independent Bedouin are ordered about by moral pygmies, and money has become the
yardstick of a persons worth.
28
mursil servant, messenger-boy.
554 clive holes
g k n-nim, laf, Our gallant lads have come to you,
wi tbair y zn theyre here, be of good cheer!
igbn fg il-higin, w usd Like eagles riding camels, like ferce
xayylah lions, those cavaliers!
is told in Ghassn s version (line 7) to cry even more, as there is no lon-
ger any comforting presence to reassure her:
rh n-nim u mad
in yxk
30
u s in
31
Were lost, my brother, wander-
wn il-harba wn? ing . . . oh where is our escape?
u h litn b hal-balad By God, the nations state is dire:
wallah mi h lah no order here, no shape.
wis iln dr il-h ad
bi l-bn
We act like world-class donkeys,
with our cruel fate content,
wa l-kull rd
l-gr h att
-aham xlah
Each one of us accepts abuse,
even those of high descent.
3 wa l-amr tanfd ah lin u min
drn xann
We do as we are ordered to; its
not our place to ask,
wa l-gir sr ar-rabb wa r-rh
fadw lah
Money rules our lives; our souls
are mortgaged to that task.
4 wa s-samt krin lin wa a
l-h agg m nin
Keeping quiets second nature; we
dont stick up for peoples rights,
wa l-h agg min il-mursil wallah
m nnlah
Our rights we beg from minions,
then give back without a fight.
5 h ayyhum iyx il-arab il
takstuh bi antnn
So salute our Arab leaders, on
their cars two aerials fixed!
rab it-taks l-h umur wa
d
-d
aym bi iglah
Theyre the ones who drive red
limos, head-ropes full of despots
tricks!
6 y mgana bi n-nay zd dum
il-n
O wife who grieves in widows
weeds, weep on, the futures
bleak,
zd dum il-as a l-xadd
sayylah
Weep yet more tears of sorrow,
let them trickle down your cheek.
7 rh n-nim u mad
aw, l
tafrah y zn
Our gallant lads have gone for
good, theres nothing left to
cheer,
b syf il-h arb u m d
all
xayylah
They sold their swords off long
ago; their horsemen disappeared.
8 kull nad il h kim bihum h ukm
il-and i-n
Now any numb-skull scoundrel
treats them like his abject tools,
u xall wgh il-balad bi s-sg
attlah
Makes porters out of highborn
men who once this country
ruled.
35
35
Te reference is to Mud ar Badrn , the Jordanian Prime-Minister of the time. Tere
is also a whif here of a sense of lse-majest: Jordanians of pure Bedouin descent regard
the Badrn family, which has its roots in urban Syria , with disdain.
558 clive holes
9 awmrah mh ad
arah yirm
ala l-gln
Hes got his orders ready, and he
lets fly to each side,
saly l-awmir raad u alt
gattlah
A thunderous volley of memos,
death-dealing stuff inside!
10 saww tnakit il-kz
36
bi h dd
dnrn
A jerry-can of kerosene put up to
two JD,
37
wa a h sb g in-ns titkt ar
amwlah
To make his stacks of cash pile
up, the people go hungry.
11 il-wisk lah wi l-z lah wa li
h sbah hal-girn
Hes got his whisky, young roast
lamb, hes got our money too,
wa -aab b h amdullah m
yiba nxlah
The people feed on wheat-chaff
and just thank the Lord they do.
12 fawwad
ard
in yxk u s in wn il-
harba wn?
Were lost, my brother,
wandering . . . oh where is our
escape?
u h litn b hal-balad wallah
mi h lah
By God, the nations state is dire:
no order here, no shape.
15 wis iln dr il-h ad
min xn
wa l-n
To blame? Our sheikhs and
senators; the mess were in is foul:
ill ala abwb id-dant inmr
gawwlah
They spend their time in sleazy
dives, like panthers on the prowl.
16 l yiurrak kubr i-anab wa
l-fard maa mit n
Dont be fooled by big
moustaches, guns with double
cartridge clips,
bi d-d abd aba yibd u bi l-madh
yinh maglah
What they write begins with
weasel words and ends with
flattering quips.
36
A twenty-litre jerry can of kerosene, used for cooking on primus stoves.
37
I.e. (in 1982) very expensive.
arabic popular poetry and political satire 559
17 iyh t t lah gu ww z-zarf
h ukkmn girn
Our rulers slip them money, in
brown envelopes its brought,
u min agal hal-girn yanh ar
lak iylah
Theyd kill their kids to get
one, slit their throats without a
thought.
18 y marh ab y hal! imnn
il-il min wn?
SoHello, I say, and welcome! But
now wheres this price rise from?
agbal aln l-bal, y maba
igblah!
A scourge is come upon us, and
its hit us like a bomb!
19 u ym id
-d
gt wn rh
in-nim wn?
Where did they go, now times
are hard, those brave heart lads
with pride?
rabi il-kaff l-h umur wi l-ugul
mayylah,
The ones who wear red head-
cloths, head-ropes slanting to the
side . . . ?
20 win xilsat il-arkt, yallah is-
salmah
But when the battles over, and
the situations calm,
itxalln il-wh id min id-d ill
h ltah h lah
You lot will mess things up again;
youll do us yet more harm!
Tus where the original song was a jingoistic, bombastic romp, in
Ghassn s hands it is transformed into a mordant diatribe against the
Badrn s governments economic mismanagement, its cavalier treatment
of ordinary Jordanians , and the corruption and cupidity of its ofcials.
It is precisely the parodying of the diction of the original that gives the
poem its punch as a piece of political satire.
Te relationship of language level, poetic form and subject matter is
worth pondering. Historically, Bedouin popular poetry was concerned
solely with the local tribal milieu: with vaunting the merits of ones tribe
over those of a neighbour, with praising heroic warriors and generous
shaikhs, with ridiculing enemies, against a backdrop of descriptions
of desert fora and fauna and the movement of the tribes in constant
search for water and good pasture. In this, as has ofen been observed,
Bedouin popular poetry is the direct descendant of the old pre-Islamic
poetry , and is similarly restrictive in matters of meter and rhyme . For
all these topoi, the only possible linguistic vehicle was the vernacular ,
both because it was the only form of Arabic the poets and their audi-
ence knew, but also because of the organic nexus between locale, sub-
ject matter and language. But now that the traditional subject matter for
the exercise of poetic talent has all but disappeared, this ancient poetic
560 clive holes
tradition has shown its protean nature: it continues to be used, even by
Bedouin poets who are literate, to serve many of the same functions,
though now on a much broader social and political canvas. If one wants
to emoteto praise, to damn, to ridicule, to gloat, to cajole, to bewail, to
lovehow much more efective is the vernacular because of its infnitely
greater allusiveness, colour, concreteness and idiomaticity for the ordi-
nary person, compared with the sanitised and institutionalised bland-
ness of the modern standard language .
As I write this article, I have to hand the latest in a long-running
series of poems written since 2003 by a south Jordanian Bedouin poet ,
Muh ammad al-Fant il al-H ajy on the American-British invasion of
Iraq , some of which have been published in local newspapers and maga-
zines. Te poems recycle standard Bedouin poetic tropes, such as the
journey (similar to the Classical rah l), whereby a rider, afer frst having
had the noble qualities of his camel
38
described by the poet , is sent with
a message (the poem ) to kinsmen encamped far away; or the ascent
whereby the poet climbs to the top of a nearby hill (rigm), there to wres-
tle with the cauldron of his emotions (higs, hgs, hawgs) and fnally
be seized by poetic inspiration to write his poem
39
. . . except that in these
poems , it is leading international politicians who are thus depicted! For
example, the poem Fadas wa Fads
40
opens with George Bush making
an ascent in his home state of Texas to ponder on his woes during the
2004 Iraq insurgency:
41
araf rigmin f marb takss Ah climbed atop a Texas peak, out
west in our proud nation,
rigmin t awl min al-baar A peak so hah no folks live there, to
r mns fahnd some isolation.
38
Nowadays, it is quite common for the poet , without a trace of irony, to describe
the qualities of a modern steeda Toyota Landcruiser, for instancein terms similar
to the ways a fne riding camel would once have been described: its strong chassis, its
smooth gear-change, its fne performance, etc. See, e.g., Kurpershoek 1999:110114 for
examples from Najd . We recorded similar descriptions of cars in Bedouin poetry from
Sinai .
39
Kurpershoek 1999:3134.
40
Tese pet names are ofen given by the Jordanian Bedouin to dogs (rather like
Rex and Rover in English). In the poem they are insultingly applied to the pet rulers
of unnamed Gulf States who, in the poet s view, treacherously provided bases for the
Americans at the time of their invasion of Iraq .
41
And speaking in a Texas accent, which I have tried to imitate in my translation!
arabic popular poetry and political satire 561
u darabt min hamm xms Ah brooded deep upon mah woes,
bi l-asds Ah felt at mah wits end, when
u h asst f rs kam t arg ngs Ah heard a bell toll in mah head
what did that bell portend, then?
u gadt wagt bn higsin u hgs My worries crowded in on me, my
state it was concernin,
wi l-galb tgad bih kam nr Mah heart afame as if it were a
fns lanterns oil-wick burnin!
Another character who has had poems put in his mouth by this poet
is Saddam Hussein , in prison in Iraq , but stillpoetically shaking a
defant fst a the Americans . And to George Galloway , founder and sole
representative of the UK s Respect political party, he has dedicated a
typical piece of Bedouin madh , praising Galloway s opposition to Tony
Blair and what the poet sees as his dogged, unpopular but heroic sup-
port for the Iraqi people.
Popular poetic commentary on contemporary events such as has
been exemplifed in this paper is nothing new: its just that the focus
has broadened. Before the birth of the nation state in the Arab World ,
and even in the period since, Bedouin tribal poetry was one of the main
means by which partisan communal sentiments could be articulated. A
good example of this was a dispute between the H uwt t , a Jordanian
tribe , and the Ban At iyya of Saudi Arabia , which rumbled on into the
late 1980s in the form of an extended poetic debate of claim and counter-
claim. Te dispute centred on T ubayq , an area in the far southeast of
Jordan which the Jordanian government ceded to the Saudis in 1964
in return for a stretch of Red Sea coastline which would enable Jordan
to improve its naval facilities away from the prying eyes of the Israelis .
Initially, this exchange of land and redrawing of the border caused no
problems, and the H uwt t on the Jordanian side continued to drive
their animals to seasonal pastures on the Saudi side, exactly as they had
always done. Tey incidentally benefted from the fact that many goods
were cheaper in Saudi Arabia , and border controls were lax or non-exis-
tent. A number of events, however, changed all this. During the Jordan-
ian civil war of 1970, Russian machine guns became easily available, and
enterprising H uwt t began a lucrative smuggling operation into Saudi
Arabia . Drugs were also smuggled in. Tis led to a clampdown by the
Saudis , and the blacklisting of many H uwt t . Ten, afer the shock-
ing siege of the grand mosque in Mecca by Islamic fundamentalists in
1979, the Saudis put even stricter border controls in place. One measure
562 clive holes
was the digging of a ditch, 3m wide by 3m deep along the border with
Jordan in T ubayq . No camel or car could cross it except at designated
control points situated far apart. H uwt t who had been used to driving
their animals 10 miles to pasture were now faced with driving them 100
miles to fnd the nearest border post and 100 miles back again. Customs
controls became much stricter. A system of registration documents
was introduced for the family members who moved with the migrating
focks. Tere were cases of H uwt t mothers who had given birth while
in Saudi Arabia being detained there because, when they tried to return
to Jordan , the number of family members did not correspond with
the number on the registration document. Tere were intimate body
searches of all females at the border checkpoints. Tis was bad enough
for the socially conservative Bedouin , but it also closed of the last
avenue for smuggling. On the other hand, the Jordanian borders
remained open to Saudis without let or hindrance, a fact that caused
huge resentment. Tis unequal treatment, coupled with what was per-
ceived by the Jordanian Bedouin as the central governments lack of
economic help to the people of the south, was one of the factors that
precipitated rioting in Maan and Al-Jafr in 1989.
A well-known H uwt poet , Barrk Dish Ab Tyih , wrote several
emotional poems about the dispute that the Ban At iyya , on the Saudi
side of the border, did not like. Ban At iyya poets replied with poems
which were recorded and passed back to the H uwt t . Te initial skirm-
ish might have ended there. However, it was followed by a long and
insulting poem from a young H uwt poet , Nad T mn Ab Tyih . Te
poetic tit-for-tat then escalated, with ten poems by Ban At iyya poets in
reply, releasing much pent-up personal vituperation. Tis caused an out-
raged reaction among the H uwt t , who had always regarded the Ban
At iyya as inferior to themselves. Seventy years ago, the result would
undoubtedly have been a tribal war; on this occasion, the dispute was
fnally defused in 1990 by mediation, and the signing of an agreement
that neither side would write any more poems on the subject of T ubayq .
Te key point here is that poetry was, and is, regarded by the Bedouin
as a suitable vehicle for airing important issues of the moment, rather
than a letter to the provincial governor, still less lobbying a remote and
seemingly uncaring central government.
arabic popular poetry and political satire 563
5. References
Abdel-Malek, Kamal . 1990. A Study of the Vernacular Poetry of Ah mad Fud Nigm .
Leiden: E.J. Brill.
Abu Athera , Said Salman and Holes , Clive. Forthcoming 2007. Y Kundlzza Rys !
Politics and Popular Poetry in the Contemporary Arab World . Leiden: E.J. Brill.
Abu-Lughod, Lila . 1986. Veiled Sentiments: Honor and Poetry in a Bedouin Society.
Berkeley: University of California Press.
Bailey , Clinton . 1991. Bedouin Poetry from Sinai and the Negev. Oxford : OUP.
Booth , Marilyn . 1990. Bayram al-Tunisi s Egypt : Social Criticism and Narrative Strate-
gies. Exeter: Ithaca.
Ghawnmah , M . Abdh Ms : Ridan wa-Mubdian, Amman, Dr al-Kind, 2002.
Ibn Khaldn , tr. Rosenthal , Franz . 1958. Te Muqaddimah. An Introduction to History.
London: Routledge and Kegan Paul.
Ingham , Bruce . 1986. Bedouin of Northern Arabia : Traditions of the l-D hafr . London:
Kegan Paul International.
Kurpershoek , P. Marcel . 1994. Oral Poetry and Narratives from Central Arabia I: Te
Poetry of al-Dindn . Leiden: E.J. Brill.
. 1995. Oral Poetry and Narratives from Central Arabia II: Te Story of a Desert
Knight. Te Legend of lwh and other Utayba heroes. Leiden: E.J. Brill.
. 1999. Oral Poetry and Narratives from Central Arabia III: Bedouin Poets of the
Dawsir Tribe . Between Nomadism and Settlement in Southern Najd . Leiden: E.J.
Brill.
. 2002. Oral Poetry and Narratives from Central Arabia IV : A Saudi Tribal History.
Leiden: E.J. Brill.
. 2005. Oral Poetry and Narratives from Central Arabia V : Voices from the Desert.
Glossary, Indices, and List of Recordings. Leiden: E.J. Brill.
Landberg , Carlo le Comte de. 1901. tudes sur les dialectes de larabie mridionale .
Leiden: E.J. Brill.
Meissner , Bruno . 1903. Neuarabische Gedichte aus dem Iraq. Mitteilungen des Semi-
nars fr orientalische Sprachen zu Berlin VI. 57125.
Musil , Alois . 1928. Te Manners and Customs of the Rwala Bedouin . New York: Ame-
rican Geographical Society.
Socin , Albert . 19001. Diwan aus Centralarabien . Leipzig: Teubner.
Sowayan , Saad Abdallah . 1985. Nabati Poetry : Te Oral Poetry of Arabia . Berkeley: Uni-
versity of California.
Wallin , Georg August . 1851. Probe aus einer Anthologie neuarabischer Gesnge in der
Wste gesammelt. Zeitschrif der Deutschen Morgenlndischen Gesellschaf 5. 123.
. 1852. Probe aus einer Anthologie neuarabischer Gesnge in der Wste gesam-
melt. Zeitschrif der Deutschen Morgenlndischen Gesellschaf 6. 190218, 369378.
Yassin, M.A . 1977. Bi-polar terms in Kuwaiti Arabic . Bulletin of the School of Oriental
and African Studies 40. 297330.
NOTES ON THE DIALECTS OF THE LGT AND H AMD
AH
OF SOUTHERN SINAI
Rudolf de Jong
ACLC, University of Amsterdam
1. Introduction
Te subject of this contribution
1
is the dialect of the Lgt (or Ulayqt) ,
a bedouin tribe who live in the western central part of southern Sinai
(see map below).
2
In addition, notes on the dialect of the H amd
ah have
been included. Some texts recorded among the Lgtwith additional
information in footnotesmay serve for further illustration.
Te H amd
ah are only few, and are ofen regarded as a clan (or fam-
ily) of the Lgt , although various sources claim that they were present
in Sinai before the Lgt .
3
We shall see that there are some notable dif-
ferences between these two varieties of speech.
1
With great pleasure I dedicate this contribution to Kees Versteegh . My dedication
is with deep respect for his stature in our feld of Arabic Studies , with gratitude for the
inspiring thoughts he has shared with our community and with fond memories of the
(too few) occasions I had the honor to work with him.
2
Te material used for this article was collected in the framework of my own research
into the bedouin dialects of southern Sinai . Tis project is funded and supported by Te
Netherlands Organization for Scientifc Research (abbreviated in Dutch as NWO ) and
the Amsterdam Center for Linguistics and Communication (ACLC) at the University
of Amsterdam. I am sincerely grateful to both organizations, as I am thankful also to
Manfred Woidich for his advice and support during my researches. I am no less grateful
to my desert guideand now dear friendId Ab uw Silm (al-At ra at-Turbniy ), who
has been my ally in this project and without whose contributions this research would
not have been possible.
3
Murray (1935, 291), for instance, writes that the H amd ah [. . .] are now only about
twelve tents strong, and live mostly near Serabit el-Khadim and Bir Nasb , where they are
reckoned as part of the Aleiqat. For the location of Sarbt al-Xdim , see Google Earth
at (appr.) 29 00 05 N33
28 01E.
At -T ayyib (1997, 480481) mentions that the H amd ah joined the Lgt under the
Lgiy x at that time (A.D. 1914) Mdaxxal Slmn of the Lgiy clan named Awld
Silmiy . Von Oppenheim (1943, 162, 164) writes that the H amd
ah were angegliedert to
the Olk t (in his transcription).
On the webpage http://members.nova.org/~lroeder/alegat.htm (authored by Larry
Roeder ), under History of Hamada, it is reported that the H amd
ah [. . .] probably
[joined the Alegat ] shortly afer the capture of Nuweiba . We do know that from then
until the 1880s this was a sub-tribe (or section) of the Alegat . Ten in 1880 internal
566 rudolf de jong
Reported to have arrived in Sinai at some time in the fourteenth
century,
4
the Lgt today number around 1,000,
5
and are found predom-
inantly in an area named ar-Raml ah , near where Jabal Sarbt al-Xdim
6
stands, situated just south-west of the Th escarpment, and north of
Wdiy Fr n and in and to the east of the town of Ab uw Znmah .
7
Also
farther north, to the south of Rs Sadr and in Wdiy arandal (not far
north of H ammm Farawn ) families of the Lgt are said to reside, as
well as farther south, near the town of at -T r .
8
Approximate distribution of bedouin tribes in southern Sinai
9
politics forced a split when the Hamada requested permission to live under the protec-
tion of the Muzeina . By 1935, they had become integrated enough in Muzeina afairs to
be considered an integral part of that tribe.
4
See Bailey 1985, 48. For more information on subdivisions of this tribe , their xs,
history, territories, etc., see at -T ayyib 1997, Part 2, 475489.
5
Such numbers are of course approximations.
6
Sarbt al-Xdim , some 40 kilometres east of Ab uw Znmah (on maps usually
spelled as Abu Zinima ), is famous as the site of turquoise mines operated since early
pharaonic times, and the temple of Hathor , which is the only pharaonic temple built
outside Egypt proper. See also the webpage about the Sinai at http://www.arcl.ed.ac.uk/
arcx/remot esense/sinai/.
7
Also in Wdiy Isla . Te Lgt are reported to form an alliance in Sinai with the
H amd
ah and Mznah , see At -T ayyib (1993, 705) and Bailey (1991, 5). At -T ayyib (1993,
706) reports that their territory stretches from ar-Raml ah to Wdiy arandal . One of my
own Lgiy informants mentioned these areas too, but said their drah stretches up until
the area named ar-Rynih , north of Rs Sadr . For a map locating their (there transcribed
as Algt ) drah, see Bailey (1991, 4).
8
Wdiy s Sahaw , mnt igit Mbajjmah , Br anNas b and Wdiy Lih yn were mentioned
to me by a Lgiy informant as parts of the H amd
iy drah .
9
Abbreviations: Ah = Ah aywt , Tr = Tar bn , H w = H wt t , Db = Dbr (see remark
below), Ty = Tayhah , Lg = Lgt , Bd = Badr ah , Jr = Jarjrah (see remark below),
dialects of the lgt and h amd
ah 567
1.1 General
In the course of the ongoing research into the dialects of southern Sinai,
a picture has been emerging in which dialects of southern Sinai appear
to constituteto an as yet undefned degreea homogeneous group.
Tis group is typologically related to the dialects of group II in the north
of Sinai: that of the Agylah, andeven more clearly sothat of the
Samnah.
10
As a group, the dialects in the south stand separate from the dia-
lects of group I, or the Negev-type of dialect (to which the dialects of
the Ah aywt and Tar bn belong, and also those of the Tayhah and
Jarjrah) .
11
Te southern group also shows important diferences with
the dialect of the Mznah, who live in the eastern part of southern
Sinai and its southern tip.
12
We shall see that of the two dialects treated
here, H amd
/ (the
latter as a refex for both *d and *d
r
male given namexud
iy , however, the sufx -kiy is also used instead of -k for the 2nd p. f. sg.,
while V(C)
u
k (CC-uk) is regular for the 2nd. p. m. sg. in both H amd
iy and Lgiy .
15
Te spelling here with triple t is for the sake of morphological transparency. Te
pronunciation is, however, not noticeably diferent from doubled t (IPA [(t)]). See also
fn 90.
dialects of the lgt and h amd
ah 569
ing monophthongs and tend to be realized lower when they are
preceded by some of the back spirants (notably x, and h and ), or
by certain (primary or secondary) emphatics: d
my
year.
2.2.1 Raising the short vowel a
Short a is raised in a variety of positions preceding primary stress. In
the cases cited below, such raising is concluded to be optional, since it
does not occur always. Raising of a tends to be towards [u] when it pre-
cedes C, or when followed by w, and towards [i] in otherwise neutral
environments.
Lgiy ilh (~ alh) on him (notice the short base,
17
but in H amd iy
more regularly ilh), d ibh t I slaughtered and in sandhi also i- mb
on the side.
Short a preceding stressed : gibyil tribes, iml north and ibn
hungry, kisln lazy. Short a preceding stressed : kit r much, bid far,
digg four, ijn dough. Although such raising is predominant, forms
with unraised a also appear, e.g. t argah method, kabr large, and even
more regularly so when h , precedes a, as in h add iron, ars groom,
xarf autumn.
16
Ab uw lHl , lit. Te Sphinx, but here as a folk etymology for Te Blue Hole, which
is a popular dive site about 12 km north of D ahab .
17
Tis must have developed in analogy to such variation as luh (~ much less frequent)
lh to him. Comparable forms (alh and alh) were also recorded in the dialect of the
Tar bn of the north and that of the Rmlt (respectively), see De Jong (2000, 181).
570 rudolf de jong
Short a preceding stressed : unb (~ anb) south, isSudiyyih
Saudi Arabia, gud camel that has not yet cut its canine teeth,
18
lugh
impregnated (of a she-camel) and ub y (~ ab y) my father. Absence
of such raising was also recorded, as in ar emperor fsh.
19
Such raising
appears to be less regular when h precedes, as in h amlih animal led to a
party to be slaughtered for those present,
20
ars bride, azz pole. Te
gahawah-vowel (see below) a preceding in an open syllable remains
unafected, e.g. maxarm pierced, maarf known, mah at t placed.
Preceding stressed : il
u
k on you, fdt I sacrifced and middt I
stretched, xit I entered, istinnt I waited. Raising towards [u] pre-
ceding w: in muwwl poems, ruwynah well-watered, suwwt I did.
In some cases comparable raising has led to morphological restruc-
turing.
21
Examples are irbt I drank, nist I forgot, ligt I found. Notice
that in such cases underlying |a| no longer has a surface form a.
22
Not only when preceding primary stress, but also when preceding
secondary stress (and following primary stress), a may be raised in
certain positions. Examples are (Lgiy) : nwikal he was eaten, ttifag
he agreed, jjimal the camel and lmut ar the rain. Raising of a in
these previous four examples is optional, but raising of a has become
steady (i.e. it has become morphologically restructured, although it has
remained underlying |a|) in open syllables of imperfect forms such as
(underlying |a| underlined) ynbis it he rejoices, yttifg he agrees.
23
2.2.2 Final ah and *- ( )
Te fem. morpheme tends to be raised, and not only in pause. Te height
of raising is from [e] to [i] (e.g. s ayyrih # small, madrasih # school,
mirfh # acquaintance, h jih # thing, zibdih # butter) provided such
18
See gad, Bailey (2004, 457) (glossary).
19
See (spelled shaour) webpage http://www.saudi-fsheries.comArabic/fsh_
product_2.htm
20
In other dialects h amlah is used in the meaning of clan , see Blanc (1970, 114).
21
a no longer appears at the surface as a, but as i. Tis i, however, is not dropped
in open unstressed syllables (e.g. iribt, not ribt for I drank, the latter of which can
be heard e.g. in the dialect of alAr , see De Jong (2000, 515)), and is therefore to be
regarded as still underlying |a|, see also following fn.
22
In Negev dialect , on the other hand, the |a| reappears as a in closed syllables , e.g.
arbit she drank, cf. Blanc (1970, 134).
23
In closed syllables however, underlying |a| does indeed reappear as a, e.g. yinbs t uw
they rejoice, yittfgin they (f.) agree. Notice that in the dialect of alAr , this vowel i is
also underlyingly |i|, which may be concluded from its elision in a form like yniwkil it
is eaten and (not ynwikil), see De Jong (2000, 521).
dialects of the lgt and h amd
ah 571
raising is not inhibited by phonetic factors, such as preceding emphat-
ics (e.g. h t ah #, ajar ah # tree, H amd
iy as well.
Some additional examples are ghawah cofee, h awal cross-eyed,
th alib she milks, yazlin they (f.) spin. Notice that when sufxed, the
fem. sufx -ah in gahawah-forms becomes -at, irrespective of whether
the sufx is vowel- initial or consonant-initial, e.g. gahawat my cofee,
gahawathin their (f.) cofee.
26
3.3 Stress in gahawah-forms
In verb forms, the gahawah-vowel a has a tendency to behave like
an ordinary anaptyctic, in the sense that it remains unstressed, where
it would otherwise receive stress, (gahawah-vowels underlined) e.g.
th a rt uw you (m. pl.) plough and yazlin they (f. pl.) spin;
27
compare
these to forms like (anaptyctic high vowel underlined) tguduw you
(m. pl.) sit and contrast with stress in e.g. saltuw you (m. pl.) asked.
3.4 Elision and anaptyxis
Generally, elision of high short vowels i and u (except those that are
underlying |a|) takes place like elsewhere in Sinai.
28
25
Notice that group II (i.e. the dialects of the Samnah and Agylah ) dialects in the
north have stress patterns inCCaC, yinCCiC. On the intermediate position between
groups I and II in the north of the stress-type inCCaC, ynCiCiC, see De Jong (2004,
159).
26
See remarks below.
27
Comparable forms with vowel -initial sufxes were heard in group I, see De Jong
(2000, 105).
28
See De Jong (2000, chapters IIII, 2.4).
dialects of the lgt and h amd
ah 573
4. Morphology
As bedouin dialects do, Lgiy and H amd
iy tend to
be used for the 3rd person masc. sg.
29
Te sufxed form of the 3rd p.
masc. sg. is -u(h) in both dialects. Te 3rd p. m. pl. pron. sufx is usually
-huw in Lgiy, but -hum in H amd
iy was
-k um, while I only heard -k uw in Lgiy, the latter of which is also heard
in dialects farther to the east. Both dialects have intin and -kin for the
f. pl. Te 2nd p. sg. sufxes -k and -k, which are typical for many south-
ern Sinai dialects,
30
are also present in H amd
iy and Lgiy.
Like almost all dialects of Sinai, Lgiy and H amd
r a
vegetable dish
Instances with dropped i (ofen anaptyctics here underlinedare
inserted to eliminate a resulting cluster) ilibt my pack/can, ar abyt
u
k
33
your car, and in sandhi u l t alm ayyah of (gen. exponent) the water. rh t
ilbirdign the smell of the orange, giblt iliLgt the tribe of the Lgt .
4.2 Verbal morphology
A notable feature of verbal morphology is the vowel harmony in pre-
formatives of the imperfect forms, which is common in many Sinai dia-
32
In this respect, Lgiy difers from dialects of the Negev , where one will hear (also)
gahawat(+), but gahawithin, see Blanc (1970, 142). Other dialects in Sinai show yet
other ways of treatment of T in gahawah-forms, see De Jong (2000, various chapters,
3.1.10.3). Treatment of T in Lgiy is much like what was described for Smniy (i.e.
the dialect of the Samnah ) of group II in the north, see De Jong (2000, 279280).
33
Notice that the form is not ar abiyytk! Contrast with the form ar abiyythum listed
above.
dialects of the lgt and h amd
ah 575
lects , but in our dialects discussed here the frst p. c. singular is included
in this rule (so that in measure 1 of regular verbs the 1st p. c. sg becomes
homophonic with the m. sg. imperative form), e.g. iktib I write, ud
rub
I hit and ar ab I drink, but also ugm I rise, il I carry, anm I sleep
and jiy I come.
34
Another characteristic of bedouin dialects in the south of Sinai
(including Mzniy ), though not of all, is the apocopation of 2nd p. masc.
sg. forms of tertiae infrmae verbs . Tis occurs in all measures (indi-
cated in Roman numbers here), e.g. in Lgiy : (I) tim you go, (I) talg
you fnd, (II) tsaww you make/do, (III) tlg you fnd, (IV) tit you give,
(V) taa you have dinner, (VI) talg you meet (with),
35
(VIII) titir
you buy and also (with shortened base vowel ) tij (~ tjiy) you come.
In a similar fashion, masc. imperatives of tertiae y verbs are usually
apocopated , as in the examples ijr run, ans forget, saww do/make!, it
give! etc. Also when sufxes are appended, the apocopated forms are
used, e.g. ansuh forget him!, sawwha make it (f. sg.)!, it uh-yyh give it
to him! and itwarrh-iyyh you show her to him.
36
5. Typological positioning of Lgiy and H amd
iy inside Sinai
Lgiy may not very convincingly be part of the group I or Negev dialect -
type (as illustrated below in a comparison with the group I dialect of
the Tar bn around Rs Sadr , who are the northern neighbours of the
Lgt ), but it is still nearer to group I than H amd
iy .
37
In this respect the
34
Forms like at lub and aktib (like in dialects of northern Sinai , see De Jong (2000,
3.2.1.2 of chapters 1V)) were also recorded, but the forms with vowel harmony in
the 1st p. c. sg. turn out to be much more regular than I had previously noticed (contrast
De Jong (2004, 166), where only at lub and aktib are listed).
35
Another feature typical of many southern Sinai dialects is the reduction of initial
tt- (> t-, sometimes erroneously referred to as a haplological drop of ta-) in ta- initial
measures (V and VI).
36
Te origin of such apocopation may lie in a reasoning by analogy (extrapolation
leading to paradigmatic levelling within the verb system): if forms like e.g. imperfect
tgudiy and tkitbiy (or imperative gudiy and kitbiy) are fem. forms for the 2nd per-
son singular, and the forms tugud and tiktib are used for the 2nd p. sg. masculine, then
dropping the -iy ending from a 2nd p. f. sg. forms like e.g. tit iy will yield the form for
the 2nd p. masc. sg.: tit .
37
Largely illustrating Murray s remark (1935, 263 f, see fn. 13 above). Te dialect of
the Tar bn of Rs Sadr (not very diferent from other varieties of Turbniy ) is taken
here as representative for group I (or Negev -type) dialects . For the latter type, see Blanc
(1970) and De Jong (2000, chapter 1).
576 rudolf de jong
following dialectal featuresfor the greater part in recapitulationmay
be considered (this list is not exhaustive):
3839
Lgiy and H amd
iy * Tur bniy
Phonology
phoneme */k/ /k/ and /k/
38
only /k/
allophones of /j/ [d] ~ often [] [d], rarely []
diphthongs *ay, *aw // and // phon. conditioned ay
and aw
raising of a preceding middt maddt
(extreme) raising *- () ti, lai tiy, laiy
velariz. in pl. of kat r kt a r (no velarization ) kt r (velarized)
Stress CaCC (Lg) CaCC
CCaC (~ few CaCC)
(H m)
lCaCaC (Lg) lCaCaC
alCCaC ~ lCaCaC
(H m)
Parts of Speech
Article and relative
pronoun
al- /-il & alliy / illiy only al- & alliy
Demonstratives: singular m. (h)d a, f. (h)d iy m. hd
a, f. hd
iy
coll., plural d ill (ih) (~ hd il)(Lg) hd
al/hd
l
d ill (ih) (~ hd
l)
(H m)
Negated pronoun 3. sg. m h, m h (~ mha) m h, m h (~ mha)
Pronoun suffix 3. m. sg -uh -ah /-ih
3. m. pl. -huw (Lg), -hum
(H m)
-hum
2 m. sg. C-
u
k C-ak
2 f. sg. V(C)-kiy, CC-ik
39
invariable -kiy
Suffixed prepositions muh mah
ilh (Lg), ilh (H m) alh
fh, f
u
k fah, fak
Verb impf . 1st p. c. sg. iktib, ud
rub, ar ab,
ugm, uxu, etc.
aktib, ad
rub, ar ab,
agm, axu, etc.
Apocopated impf . 2 m. sg. tim, talg, tlg
tit , taadd, etc.
timiy, talga, tlgiy,
tit iy, taadda, etc.
38
But this phonemic opposition is not as widely used in Lg.
39
Tis is not entirely certain; my H amd
ah 577
Verb to come
perf . 3.m. (sg.)
i (pl.) um ~uw (sg.) ja (pl.) jaw
impf . 3.m. yiy, yuw yjiy, yjuw
Suffixed prepositions : muh mah
ilh (Lg) alh
Vowel impf . to eat, to take ykul, yxud (H m) ykil, yxid
ykil, yxid (Lg)
Vowel perf . 3rd p. f. pl. impf . katabin, gl in,
yar ab in
kataban, gl an
yar ab an
-m in 2nd m. pl. perfect katabtuw (Lg), katabtuw
katabtum
40
~-tuw
(H m)
Interrogatives
when? imtn mat (~ wagt)
who? mn min
Adverb
here nih(-niy) hniy
Selected lexicon
mother um m am m
baking sheet z s j
* Where not specifed, the listed form is heard in both dialects .
40
6. Conclusion
Te comparison drawn above shows a number of diferences between
Lgiy /H amd
iy.
7. References
at -T ayyib, Muh ammad Sulaymn . 1997. Mawsat al-qabil al-arabiyya, Part 2, Cairo:
Dr al-Fikr al-Arab.
Bailey, Clinton . 1974. Bedouin Weddings in Sinai and the Negev. Folklore Research
Center Studies, vol. 4, Jerusalem Magness Press, 105132.
. 1985. Dating the Arrival of the Bedouin Tribes in Sinai and the Negev. Journal of
the Economic and Social History of the Orient 28, 2049.
. 2004. A Culture of Desert Survival. Bedouin Proverbs from Sinai and the Negev . New
Haven: Yale University Press.
40
Verb forms with fnal -m were only recorded during direct elicitation. Such forms
also occur in group II of the north, see De Jong (2000, 298299).
578 rudolf de jong
Behnstedt, Peter . 1979. Die nordmittelgyptischen bukar a -Dialekte. Zeitschrif fr ara-
bische Linguistik 3, 6295.
Blanc , Haim. 1970. Te Arabic Dialect of the Negev Bedouins. Israel Academy of Sci-
ences and Humanities, Proceedings, vol. 4, 112150.
de Jong, Rudolf . 2000. A Grammar of the Bedouin Dialects of the Northern Sinai Litto-
ral. Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden:
E. J. Brill.
. 2004, Characteristics of Bedouin Dialects in Southern Sinai. Approaches to Arabic
Dialects. A Collection of articles presented to Manfred Woidich on the occasion of his
sixtieth birthday, ed. by Martine Haak , Rudolf de Jong , Kees Versteegh , Leiden: E. J.
Brill, 151175.
Hava , J.G . 1982. Al-Farid ad-durriyyah, Arabic -English Dictionary (ffh edition). Bei-
rut: Dar el-Mashreq.
Lavie, Smadar . 1990. Te Poetics of Military Occupation. Berkeley: University of Cali-
fornia Press.
Marx, Emanuel . 1999. Oases in South Sinai. Human Ecology 27, 341357. New York:
Kluwer Academic.
Murray , George W . 1935. Sons of Ishmael a Study of the Egyptian Bedouin . London:
Routledge.
Oppenheim, Max Freiherr von . 1943. Die Beduinen , Band II: Die Beduinenstmme in
Palstina, Transjordanien , Sinai , H edjz . Leipzig: O. Harrassowitz.
Stewart , Frank Henderson . 1990. Texts in Sinai Bedouin Law , part 2 (Texts are in Ara-
bic ). Wiesbaden: O. Harrassowitz.
Tckholm, Vivi . 1974. Students Flora of Egypt . 2nd ed., Beirut: Cairo University (Coop-
erative Printing Company).
8. Texts
Te speaker is a member of the Lgt , who was 38 years old at the time
of recording. He was born in Sarbt alXdim (about 40 km east of
Ab uw Znmah ) in the area called ar-Raml ah , where he had lived until
his 29th. He has had no school training. (S) = Speaker of Lgiy , (R) =
interviewer.
8.1 Hunting rabbits
8.1.1 Transcript
1. (R) hatkallimni f s ayd ilar nib? 2. (S) a:ywah. alar nib d illeh
41
. . . f
nnahr bitnm. f nnahr . . . bitnm m btarta.
42
3. ibtarta a r h
billl . . . ugb almaarib . . . ibtusrub. iza nymah f h ajr , f ajar ah . . .
41
d ill (-ih) is the c. pl. demonstrative pronoun for near deixis. Notice the absence of
velarization .
42
Te single negation m + verb form is regular.
dialects of the lgt and h amd
ah 579
ibtusrub h f lll bitdawwr alma. 4. id d
all
43
h srbeh, iw srh ah . . .
l as s ubih . . . . la:
44
lfajir. alfajir xals giddm aam la d
urr
45
b iwayyeh . . .
5. bitdawwir lhe . . . ajar ah . . . bitdawwir lhe jibl, bitdawwir lh-ayyi
h jeh, iw bitgt ir fha bitnm. 6. iygl luk
46
alarnab nymeh. alarnab
d
in td
all.
44
Such prosodic lengthening of the vowel is ofen used to indicate the long duration
of time.
45
d
wiy,
53
w
alh all d
awa, yid
ah 581
iw t niy h jeh byibnh f lgd
. 22. fh
bar m biyt h f ssiyyl, talgha
56
tamalliy f ssiyyl alar nib bitr awwih . . .
8.3.2 Translation
17. (R) And are there . . . (people) who make a trap? (S) Yes, they do.
Tis trap is (a) recent (thing), this trap is something recent, which
people have imitated, not people of the old days. People have started
copying the (method of hunting with a) trap. Tey set it, and they place
in it . . . they place the peel of an orange in it, they put anything in the
trap. 19. And it (the rabbit) is attracted to its smell. Tis (person) sets it
only afer sunset. When the small cattle come back home. And when it
has come back home . . . thats it . . . there is no small cattle (lef roaming
around), (then) they set it. 20. Tey dont set it during the daytime. And
besides this they set it in summer . . . the time when the acacias bloom,
when the acacias have fruit
57
and stuf. 21. Tey also set it next to the
acacia tree. Tere is fruit that falls down. It comes normally in summer,
the month of July, the months of June and those of summer. 22. Tere
is fruit that falls from the acacias, (so) youll fnd rabbits going to the
acacias all the time.
8.3.3 Transcript
23. ibyibnuw lha fa . . . alfaxx f-assiyylah d iy, iw kamn f ssiyyl d
k,
f ttalah
58
d iy byibnuw lhin. 24. iw yulugt ha f- alfaxx . . . yulgut ha
min r agabatta min . . . min rjilha min h jeh. 25. as s ubih ym yjiy l
alfaxx iw lannha malgt ah. 26. iw fh ns ibyugud . . . igr ayyib lha . . . f
lll . . . yasma xabit alfaxx. 27. ym yt ulguh byasmauh (R) aywah . . . (S)
54
f irg, litt. in the root (of) has been grammaticalized as a preposition meaning
next to (presumably next to some standing object), also e.g. f irg alh t ah next to the
wall.
55
Assimilation of bitr awwih + diy.
56
talgha: apocopated 2
nd
p. m. sg. imperfect of the verb ligiy, yalga fnd, followed by
sufx -ha.
57
Lane (1863, part 1, 195): [. . .] the fruit of the t alh [or acacia gummifera, which is
of the trees called idh] [. . .]
58
talah (with t!) is a water course between two mountain peaks, and can be used as
a pass between mountains.
582 rudolf de jong
ibyarf alfaxx s dha . . . ibylgut uh . . . f lll 28. iw fh ns ibyibnh . . . iw
biygt ir ibld bideh . . . zayy xamsah kluh aar ah kluh s s ubh ibyjiy luh.
29. ym yjiy luh linnuh, linnuh
59
lgit ha. hd a larnab. 30. amma lazl
hd a, m kull anns ibtugnus luh, alizln d illah m biyt h in f l . . . f
lwt iy, a r f jjbl, f jjbl albideh . . . 31. (R) a r tat la fg . . . (S) a r tugnus
lhin f lijbl albideh . . . alazl. 32. w alazl law ar wh
u
k
60
. . . ibyun-
guz min induh. 33 law ar wah rh t
u
k . . . yaniy jt a rrh w ar wah , xals
biygt ir min induk. hd a lazl. 34. alliy h lbadan w alh jt d illah
kulluh. hd a . . . hd a wn ilinsn ibylugt uh.
8.3.4 Translation
23. Tey set a trap for it in (under) this acacia, and also in those (other)
acacias, in this watercourse they set (traps) for them. 24. and they catch
it in the trap. . . . they catch it by its neck . . . by its leg . . . by something. 25.
In the morning, when he comes to the trap, there it is, caught. 26. And
there are people who sit (and wait) . . . near it . . . at night . . . hell hear the
snapping of the trap. 27. When it releases hell hear it . . . (R) Yes . . . (S)
and he knows that the trap caught it . . . (and) he seizes it . . . at night 28.
and there are people who set it . . . and they go away to a far place
61
. . . like
fve kilometres, ten kilometres (and) in the morning he comes (back) to
it. 29. When he comes to it, lo, lo, it has caught it. Tat is the rabbits.
30. As far as these gazelles are concerned, not all people hunt them.
Tese gazelles dont come down to the lower areas, (they stay) only in
the mountains . . . 31. (R) You have to go up . . . (S) You have to hunt them
in the far mountains . . . Gazelles. 32. And gazelles, when they smell
you, they jump away from where they are. 33. When they smell your
scent. . . . that is, if you come down with the wind and he smells (you),
thats it, he gets away from you. Tese are the gazelles, 34. which are the
ibex and all these things. Tis . . . this is where a person catches them.
59
See fn. 49.
60
Te (superscript) anaptyctic vowel is voiceless.
61
bld lit. land, here place.
dialects of the lgt and h amd
ah 583
8.4 Snakes
8.4.1 Transcript
35. (R) fh bard
ah 585
8.5.2 Translation
43. (S) Te boy . . . the boy then goes . . . when he is out herding . . . he sees
the girl . . . 44. (R) Where does he see her? (S) He sees her in her herd-
ing place where she is herding her goats. 45. If he likes her. . . . and in
case. . . . he then goes . . . and speaks to his mother. He doesnt speak to his
father, he speaks to his mother. 46. His mother speaks to his father. She
says Te boy wants (to marry) so-and-sos daughter . . . from so-and-sos
family . . . 47. His father then goes to her father . . . and . . . he goes to her
father and speaks to him. 48. And his mother . . . goes to the girls mother,
and speaks to her. If the father of the girl says she is not engaged . . . and
we will marry her into your family . . . (then) theres more talk. 49. If the
father of the girl says Shes engaged . . . (then) there is a talk. If the girls
father says Her cousin also wants (to marry) her, (then) theres more
talk. 50. And there are people who see her, of their neighbours . . . one
of his neighbours that he has sees a girl that he likes . . . 51. hell (then)
speak to his mother. Hell say I want (to marry) so-and-sos daughter.
52. She goes to her mother and speaks to her and asks her and she says
the girl is engaged or she is still an unbetrothed virgin or thus or thus...
53. Afer that his father speaks to her . . . he goes to the man . . . the girls
father, and he speaks to him. 54. He speaks to him. If he says that hell
marry the girl you, she is not engaged, hell say to you okay. 55. If the
girl is young, he says to you wait (until) afer one year, wait (until)
afer fve months, wait (until) afer two years. 56. He then slaughters
for her. He gets his sheep, if he says to him wait (until) afer one year,
two years . . . the girl is (still too) young. 57. He then brings his sheep
and slaughters it. When he has slaughtered it, then they have betrothed
her.
68
58. From the moment they have given him his twig, thats it, he
knows that this (girl) is (going to be) his wife, in the tradition of God
and His Prophet
69
they have given him . . . his twig.
68
at h gas alah, lit. they have given him a twig. Te twig is given to the groom in
betrothal ceremonies, who is then mxid gas alatha holding her twig, i.e. she has been
betrothed to him, see Bailey (1974).
69
Te phrase b sinnt Al l h w rasluh in the tradition of God and His Prophet may
ofen be heard in additionused almost as an excuse in pre-emptionto the descrip-
tion of traditions, of which the islamic origin is doubtful (or non-existent).
586 rudolf de jong
8.6.1 Transcript
59. iw h l . . . gm xal l ha w ugub sanatn . . . z abbat
70
lh at ab, iw jb
id d abyih , iw jb ibyt aar. 60. iw azam anns, iw gl lhuw lfar ah
ind
ah 587
he has prepared the food (to be served before the wedding ceremony),
and the people have come, everyone with his slaughter animal, everyone
with his sheep. Everyone has his slaughter animal and his wife with him.
62. On the day of the wedding ceremony with so-and-so we go over
(to him) and bring him over. Everyone (goes) with his slaughter ani-
mal and goes, and also afer afernoon prayer . . . (there is) slaughtering.
63. He slaughters and prepares dinner. And there are people who make
fat (unleavened) bread . . . farh . . . the men and women make them.
64. And he distributes . . . (to) every man a fat bread and his portion in
which there is his meat. . . . he distributes it to them . . . all of them. 65.
And afer the sun has set, afer that . . . afer the evening prayer . . . round
about it (i.e. that time) . . . the dancing starts. 66. Tere are people who
do a dah iyyah, and there are people who do marbah. 67. Te marbah
is the same as the rafh iy. And the dah iyyya is . . . the clapping of hands.
(during) Te dah iyya everyone says his own lines (of poetry) . . .
8.7.1 Transcript
68. hd a xams t-infr aar ah . . . xamistar nafr biysawwuw, w
arraggs ah h urmah wih dih . . . ibturgus alhuw. hd a b innisbah l
addah h iyyeh. 69. w almar bah hd iy . . . m . . . biyr awbuw fh a
r anns .
. . ijjihhl
79
. . . alliy hinna lbant . . . ibyrugs in . . . w ilwild. 70. ijjihhl hd il
biysawwin . . . almar bah. 71. assmir d a biysawwh a
r aiyyb. anns
alkibir hd il bysumruw. ibysumruw a jl . . . ismuh smir. 72. amma
ddah h iyyah hd iy . . . biysawwha . . . kit r . . . f Sna nih biysawwha
wjdah ddah iyyah. 73. ym alfar[h awt], iw ym azzawyir, iw ym
ald . . . biysawwuw ddah h iyyah. hd a h . . .
8.7.2 Translation
68. Tis, fve, ten . . . ffeen persons do this, and the dancer is one
woman . . . who dances before them. Tis as far as the dah iyyah is con-
cerned. 69. And this marbah . . . only young people take part in the
marbah . . . which is the girls . . . who dance . . . and the boys. 70. Tese
youngsters take part . . . in the marbah. 71. In the smir only older peo-
ple take part. Tese old people take part in the smir. Tey do a smir
on the fringe (of the main festivities) . . . it is called smir. 72. As far as
this dah iyyah is concerned . . . they do it . . . a lot . . . Here in Sinai they do
it a lot . . . the dah iyyah. 73. When there are weddings, and when there
79
jhil lit. ignorant is commonly used in Sinai for young or child.
588 rudolf de jong
are visits (to sheikhs tombs), and when there is the feast . . . they do a
dah iyyah. Tats it . . .
8.8.1 Transcript
74. (R) w alars . . . w al . . . (S) ilurs f lbt (R) f lbt . . . 75. (S)
ywah . . . alurs . . . itsawwiy zzafh hum m a, iw jbha b ilar abiyyt iw
waddha . . . albt. 76. alurs had iy btugud saba t-iyym, iw h f lbt.
77. m btat la minnuh wala biyhiddh bt alfr ah . . . mabniy . . . 78. sabi
t-iyym mn ajjimeh . . . l ajjimeh. 79. iw h gdih
80
lars . . . bat nuh . . . iw
irssa indaha. l(a:) jjimah jjyih. 80. ajjimah jjyih ssawwiy sib,
d abah luh h tn . . . 81. kamn h id
ar,
81
w alliy
mh yiz mh yiz . . . 82. iw ugub kid iy diy[t] waddha druh.
aza druh . . . hlih . . . waddha-yyha. mh hleh druh, iw h l . . . 83.
bitr awwih ind um m ha. ibtasrah ib anamha. 84. lam biysawwiy druh,
iw h l . . . iw hd iy h rumtuh induh. 85. hd a b innisbah l ijjawzeh zayy
kid iy.
8.8.2 Translation
74. (R) And the bride and the groom. (S) And the bride is at home (R)
At home . . . 75. (S) Yes . . . the bride, they organise a wedding procession.
And they have brought her with cars and brought her over to the house
(tent). 76. Te bride spends seven days (while she is) in the tent. 77.
She does not go out of it (i.e. the tent), nor do they take it down the
wedding tent . . . stands (lit. is built) 78. For seven days from Friday till
Friday. 79. And she sits inside it, the bride . . . while her groom is with
her. (all the time) until the next Friday. 80. Te next Friday you celebrate
the weeks feast, he slaughters one or two sheep (for himself) . . . 81. And
also whoever wants to attend is present, and whoever does not want,
does not want . . . 82. And afer that he then takes her to his house. If
his house is . . . ready, he takes her to it. If it is not ready his house, 83.
then the situation is. . . . she goes to her mother. She takes her cattle out
to graze. . . . 84. When he prepares his house, then the situation is . . . and
this is then his wife with him. 85. Tis is with reference to the wedding,
like this.
80
in gdah is quite high; when in neutral environmentsin the CCiC pattern
of the active participle tends to be realized near IPA [:].
81
yah d
ah 589
8.9.1 Transcript
86. (R) bard
ah 591
mellons, when it comes up normally, it grows on rain (water). When the
dam
86
is well-watered it comes up and spreads out 102. It becomes inter-
twined, and the watermellons become plentiful. But the women dont
come there . . . not one. 103. From the moment he sows it, they go away
from it (and stay) on the side. (if not) It is jinxed. 104. If it is jinxed the
watermellons will go bad. Tis . . . is with reference to the watermellons.
105. And this wheat . . . comes up when our Lord wills it. 106. Tis wheat
is what we make four from and this situation, thats how it is. And the
straw . . . which they take (as fodder) to the camels . . .
8.11.1 Transcript
107. (R) bard
yih...
int ibtat h an . . . biddi
90
. . . biddi. biddi . . . yaniy ald r ah . . . itxallha
mi dagg . . . itxallha majrih jari. 113. hd a l albur n . . . l azzamil . . .
itain.
91
hd a kull ym . . . lzim tuj
u
ru lhin . . . imn ald r ah d iy. 114.
algamih hd a . . . hd a byat h aninnuh duriy a rh h . . . dagg . . . dagg hd a
zuwwdeh. 115. itdibb almazwad . . . alh l hd a . . . addagg imn algamih .
ald r ah hd iy . . . l albur n biyjirha. 116. [kull ym] hd a lh l ulitta
btat h an algamih , w ibtat h an ald r ah, iw kulluh. wi . . . w attibin hd a . . . l
ijjml a t l. m biyat h anh walla h jih bass ibyidirsh daris . . . ibydir-
suw lgamih minnuh . . . iw biywaddh a lj[l] . . . 117. (R) iw byidirsh b
h? fh lh ah kid iy walla . . . (S) aywah . . . (R) ismuh h? (S) fh ns ibydir-
suh bi l . . . b jjimal . . . biysawwiy luh madrs, ibydirsuh. 118. iw fh ns
halh n ibyidirsh b lmakan. (R) rif kilmit ilhjal?
92
(S) alhjal . . . (R)
bitgl yaniy alhjal? (S) aywah . . . [. . .] alhjal hd a biysawwuw minnuh,
ibydirsuw minnuh b jjimal. iw a
ah 593
is how it goes (lit. its situation), you grind the wheat, and you grind the
corn, and everything. And this straw . . . goes (lit. is) straight to the camels.
Tey dont grind it or anything, they only thresh it properly. Tey thresh
the wheat from it . . . and they put it aside . . . 117. (R) And with what do
they thresh? Is there like a (wooden) board, or . . . (S) Yes . . . (R) What is
it callled? (S) Tere are people who thresh it with . . . with camels. . . . hell
make (for himself) a threshing foor. 118. And there are people now-
adays who thresh it with machines. (R) Do you know the word hjal? (S)
the hjal. . . . (R) Do you call it a hjal? (S) Yes . . . [. . .] Tis hjal (people)
do with it, they thresh with it with camels. And one has to know (how)
to do it . . .
CLASSICAL AND COLLOQUIAL ARABIC ARCHAISMS
Alan S. Kaye
California State University, Fullerton*
1. Introduction
Tis essay
1
argues that modern spoken Arabic dialects sometimes
retain very archaic Semitic features. In fact, they may even preserve
Proto-Semitic forms that have been lost in Classical Arabicanother
indication that Classical Arabic is not to be regarded as their ancestral
proto-language (see Kaye 1976 for fuller explication of this idea). Rather,
there were other Arabic dialects spoken alongside Classical Arabic all
throughout history that served as the ancestral inputting ones to the
contemporary picture of Arabic dialects .
Te orthographic representation of Hebrew l no as <l>
2
may best be
interpreted as evidence for the reconstruction of Proto-Semitic *la no.
No matter what vowel preceded the glottal stop , I do not think any Semi-
tist can deny that the Proto-Semitic word ended in a glottal stop.
3
Final
* Shortly afer Alans untimely death we received the proofs of his article. In deep
respect we insert his paper in this volume as it was at his passing on in appreciation for
what he contributed to our common feld of interest (the editors).
1
Tis was originally presented to the joint session of the 216
th
meeting of the Ameri-
can Oriental Society and the 34
th
annual North American Conference on Afroasiatic
Linguistics on March 17
th
, 2006, in Seattle . I am thankful for the stimulating discussion
by the audience participants. I also wish to express my gratitude to Vit Bubenik , Gideon
Goldenberg , Wolfart Heinrichs , Bob Hoberman , Jonathan Owens , Adrian Mcelaru ,
Gary A. Rendsburg , Judith Rosenhouse , Aaron D. Rubin , Avi Shivtiel , Laurence J. (Taw-
fq) Surfas , Rainer Voigt , and Bill Young for useful comments on a preliminary version.
2
According to Koehler and Baumgartner (1958, 466) and (1998, 466), this spelling
occurs 466 times in the Old Testament , while the plene spelling <lw> occurs only 35
times. Gary A. Rendsburg [p. c.] notes that the plene spelling occurs especially in Jer-
emiah , and that there are more instances of the interrogative and presentative /hall/
spelled plene than there are defectiva.
3
Let me comment on the notion of a pho(a)ethematic glottal stop , which occurs
in some Arabic dialects in fnal position; e.g., /mia/ he went. Tese are clearly inno-
vations within Arabic (noted by the medieval Arab grammarians as well) and are pa-
ralleled by similar occurrences in Neo-Aramaic dialects (thanks to Bob Hoberman ,
p. c.). Neo-Aramaic /la/h/ is paralleled by Israeli Hebrew /lo/ and, in many ways, by the
English voiceless unreleased bilabial stop in yep and nope.
596 alan s. kaye
graphemic aleph in Biblical Hebrew
4
/qr/ he read presents evidence
for an earlier */qr/ or perhaps an even earlier */qra/ < */qara/
(cf. /qr they read), which is equivalent to Classical Arabic /qaraa/.
5
Similarly, the Hebrew form <l> points to an earlier *l < earlier */la/
6
= colloquial Arabic la, la, or laa (Hinds and Badawi 1986, 775). Te
Biblical Hebrew variant spellings <lh> (Koehler and Baumgartner 1998,
474) and <lw> (Koehler and Baumgartner 1998, 475) prove that the
glottal stop was no longer pronounced. Still, the ubiquitous spelling with
the glottal stop must be accounted for. Moroccan Arabic (so too Alge-
rian Arabic ) even has lla no and lla
7
no way!; absolutely not! (Sobel-
man and Harrell 1963, 130 for the former alternant) demonstrating that
4
Hebrew refers to Biblical Hebrew unless otherwise noted.
5
Te problem of stress is not a relevant consideration here and its discussion is thus
lef unmarked. Let me take up the matter of the Hebrew fnal graphemic glottal stop .
Could this merely be an orthographic convention to distinguish two common words:
no, not and to him, originally spelled <lh> (= Classical Arabic <lh>), making it homo-
graphic to to her, and later spelled <lw>? Te answer is clearly in the negative, since the
Proto-Semitic form contained a glottal stop.
6
Cf. Proto-Semitic ra head > Classical Arabic ras colloquial Arabic rs =
Hebrew r and Proto-Semitic *kas = colloquial Arabic ks = Hebrew ks (spelled with
aleph ), although Gideon Goldenberg remarks that Anton Spitaler thought the glottal
stop in the last example was a hypercorrection a theory that I fnd farfetched. Bob
Hoberman [p. c.] also notes the Hebrew verbs ymar he will say and ykal he will eat,
etc., that are spelled with aleph but where its phonetic realization as a glottal stop disap-
peared. Let me hasten to add that when it comes to a glottal stop, it is important to point
out that Hebrew /yir/ he will see retains it, whereas Classical Arabic /yar/ he will
see does not. Cf. Israeli Hebrew /lo/ absolutely not!. Although it is conceivable that the
glottal stop might go back to Proto-Hebrew times, it is much more likely that the latter
form is evidence of history repeating itself. Although the fnal glottal stop in this word
may originally be due to sound symbolism (cf. English yep and nope that are marked for
informal register, corresponding to more formal yes and no), the fact remains that it is
Proto-Semitic . Proto-Hebrew glottal stop in fnal position disappeared only to surface
again in Israeli Hebrew (see below) as a result of a linguistic cycle to give the root struc-
ture more than a mono-consonantal baseas a kind of Systemzwang . Tis is parallel to
Modern English ask < Old English aks, which is also once again the form (aks) in several
modern English dialects .
Concerning the Israeli Hebrew development, Judith Rosenhouse [p. c.] notes that it
seems to be a recent phenomenon and appears to be restricted to young adults, espe-
cially females, and seems to indicate abruptness or vehemence, particularly as an answer
to a yes-no question in order to stop someones nagging. When I suggested that this
might be a result of colloquial Arabic infuence on Israeli Hebrew , she replied that this
was not possible, since it was used by native speakers of Hebrew who do not have direct
contact with Arabic speakers. Still, I pose the following question in rebuttal: does one
have to have direct contact to have infuence? Bob Hoberman [p. c.] notes that the fnal
glottal stop in Israeli Hebrew is older than Rosenhouse thinks and it is not restricted to
just young adults and females, and is not mainly confned to nagging. Arabic full
investigation of this is, in my view, a desideratum.
7
Te form with the geminated l actually gives a tri-consonantal body to this root : ll.
classical and colloquial arabic archaisms 597
Arabic dialects do not like to tolerate a mono-consonantal root . Te frst
cited Moroccan form at least points to a bi-consonantal or geminated
root with double /l/, although such roots are also not characteristic of
Semitic languages in that C1C1C2 roots are unknown.
Before continuing the discussion of the importance of the aleph in
the Hebrew word for no, not, it is necessary to point out that Classical
Arabic , although usually very conservative in matters of phonology , is
sometimes innovative. I will now discuss two signifcant innovations.
First, many contemporary spoken Arabic dialects preserve the Proto-
Semitic imperfect vowel preformative (the so-called taltalah ) */i/ rather
than */a/ = Classical Arabic /a/. Colloquial Arabic /yifqid/ corresponds
do Akkadian /ipqid/, Aramaic /nepqud/, Hebrew /yipqod/, Geez /
yfqad/, and Classical Arabic /yafqid/ (Proto-Semitic fqd = to seek).
Most Semitic languages and most modern Arabic dialects have /i/ or
a vowel clearly originating from it, as opposed to Classical Arabic /a/.
Although Moscati et al. (1964) postulate the vowel */a/ as Proto-Semitic,
they also remark (1964, 143): Some scholars, however, regard the vowel
i of the prefx as primary, alongside a. I believe there is much more evi-
dence in favor of Proto-Semitic */i/ than */a/.
Second, the Classical Arabic voiced alveo-palatal africate ( jm) is
clearly secondary deriving from Proto-Semitic */g/. Although I recon-
struct a voiced lamino-palatal fricative *// for Proto-Colloquial Arabic ,
the entire story of the development in Arabic dialects is not germane to
this investigation (see Kaye 1972 for the details).
Looking at the cognate Semitic languages, Ugaritic l (variant spell-
ing l,
8
syllabically spelled la-a indicating l) does not support the col-
loquial Arabic archaic forms with glottal stop , but matches up exactly
with Akkadian l and Classical Arabic l. Aramaic , Old South Arabian
and Geez lend further support to the earlier glottal stop with their gra-
phemic alephs , although the Geez all is problematic from a compara-
tive Semitic point of view. (Hebrew does have a negative al for negative
imperatives with cognates in Ugaritic , Phoenician , Biblical Aramaic ,
Geez , and possibly Akkadian ul ).
9
Tus, any notion that colloquial Ara-
bic la() might be secondary is, I believe, an erroneous perspective due
8
Te Ugaritic spelling with a glottal stop is erroneous, although Koehler and Baum-
gartner (1998, 466) cite it. Gideon Goldenberg [p. c.] informs me that the spelling with
aleph was originally postulated by G.R. Driver ; however, others thought that this was a
verb meaning to be weakened or gleaming.
9
See Koehler and Baumgartner (1998, 45).
598 alan s. kaye
to the graphemic evidence from the ancient Semitic languages. Rather,
Classical Arabic l (which might represent /la/ since the grapheme for
the hamza is latesee further below) should be viewed as the later devel-
opment < Proto-Semitic */la/. When the fnal glottal stop disappeared in
this Classical Arabic word, the previous vowel was lengthened as a result
of compensatory lengthening. It should be noted that this form without
the fnal glottal stop breaks with root bi-consonantalism or tri-conso-
nantalism in favor of mono-consonantalisma very rare, if not impos-
sible, situation in the classical and modern Semitic languages, especially
for words (prepositions and other function words, such as b- in or l-
to; for, are mono-consonantal , although the latter mono-consonantal
form is certainly related to the bi-consonantal root l in Hebrew and the
tri-consonantal ly in Classical Arabic (Koehler and Baumgartner 1998,
48).
10
Moreover, it is not impossible that <l> in the very early stages of
Classical Arabic could have represented /la/, but that the hamza was not
written on top of the alif .
11
Tis spelling represents the scripta defectiva,
much in the way the hamza of Modern Standard Arabic is ofen omit-
ted in /uktub/ write! (masc. sg.) as it should be in /waktub/ and write!
(masc. sg.). Tis hamzatu lwas l should, strictly speaking, be written in
word-initial position, at least in some varieties of Modern Standard
Arabic.
12
I can only speculate that the Akkadian graphemes might also
represent an earlier pronunciation in a pre-Akkadian dialect or in early
Akkadian with a fnal glottal stop before the general loss (in East Semitic
as a whole) of the pharyngeals and laryngeals .
Since one is confronted with the fact that Proto-Hebrew *-a > (as
in /qr/ he read discussed above), one may be forced to hypothesize
that the Proto-Semitic word for no has a long , thus yielding a long
via the Canaanite vowel shif. It is important to point out that *qara
he read (previously discussed) does not yield qar or the like; thus the
inputting original forms had to be diferent because of the very diferent
10
Tere are other prepositions in the Semitic languages which clearly have bi-con-
sonantal roots , such as Hebrew m- from < *min (Koehler and Baumgartner 1998, 535)
usually assimilated to the former form, and Hebrew al on < ly.
11
Jonathan Owens notes [p. c.] that the hamza was a post-hoc addition to Arabic
orthography , which is why the orthographic rules relating to it are so complicated.
12
Gideon Goldenberg [p. c.] comments that in early Modern Standard Arabic (e.g.,
Blq Press, Cairo ), fully vocalized Modern Standard Arabic was printed with the hamza
sign on the alif only when the glottal stop is phonemic , such as in urdu I want. He
maintains: Te rule to write the hamza where the glottal stop is a positional alternant
was adopted by the Egyptian Ministry of Education, then became rather popular.
classical and colloquial arabic archaisms 599
outputs of the development. Bob Hoberman [p. c.] suggests that /a/ >
// was only in closed syllables where the // was the syllable coda. Te
relative chronologies of the various sound changes are also important
factors to consider afecting vowel quantity.
2 Biblical Hebrew
13
and Comparative Semitic Evidence for Proto-Heb rew
14
*// in Final Position Proving Proto-Heb rew */l/ < Proto-Semitic */la/
no
1. /b/
15
enter; come
Classical Arabic /ba/, Akkadian /bu/, Ugaritic */bw/,
Phoenician /b/, Old South Arabian /bh/ (Koehler and
Baumgartner 111)
2. /br/ create
Old South Arabian /br/ build (Koehler and Baumgartner
14647)
Classical Arabic /baraa/ is not in Koehler and Baumgart-
ner (1998).
Wehr (1974, 49) lists it (create).
3. /gb/ ditch
= Akkadian /gubbu/ < */gubu/ (Koehler and Baumgart-
ner 163)
4. dr /dern/ abhorrence
= Classical Arabic /daraa/ repel (evil) (Koehler and
Baumgartner 217)
5. /d/ grass
Akkadian /diu/ spring, Biblical Aramaic /dit/, Syriac
/t/ (according to Carl Brockelmann via Rainer Voigt
[p. c.], /te/ ~ /ta/), Classical Arabic /adiya/ be moist,
Old South Arabian /d/ season of grass; spring (Koehler
and Baumgartner 220)
13
Te Biblical Hebrew data in Section 2.0 of this essay have been taken from Koehler
and Baumgartner (1998), an updated and revised version of Koehler and Baumgartner
(1958). Tese 40 roots ofer the evidence from other Semitic languages of roots ending
in aleph , corresponding to Proto-Semitic */la/ no, not.
14
It makes little diference whether this happened in Proto-Hebrew or a pre-Hebrew
dialect .
15
It is important to point out that the aleph is pronounced in various conjugated
forms of tertiae-aleph roots ; e.g., /b/ they entered, /yb/ they will enter.
600 alan s. kaye
6. /h/ he
Classical Arabic /huwa/ but Bedouin Arabic /h/. Cf.
the // in Geez /wt/ (Koehler and Baumgartner
226). /h/
she works the same as /h/. Cf. Geez /yt/. Koeh-
ler and Baumgartner (1998) do not cite Bedouin Arabic
*/h/, but it probably exists (Koehler and Baumgart-
ner 226). Koehler and Baumgartner (226) reconstruct
Proto-Ethiopian Semitic */hatu/ he and */hati/
she. Gideon Goldenberg [p. c.] notes that Qumranic
Hebrew /ha/ he and /ha/ she parallel the Classi-
cal Arabic forms with /-a/.
7. hb /nhb/ hide oneself
Classical Arabic /xabaa/ hide (Koehler and Baum-
gartner 270)
8. /h g/ shame; confusion
Classical Arabic /xajaa/ look ashamed (Koehler and
Baumgartner 275)
9. /h t / miss the mark
Classical Arabic /xat ia/ make a mistake; sin Wehr
(1974, 245)
Koehler and Baumgartner (1998) cite Classical Arabic
/xat iya/. Ugaritic /xt /, Biblical Aramaic /xt /, Old
South Arabian /xt /, Geez xt (Koehler and Baum-
gartner 288)
10. /h l/ fall ill
Classical Arabic /xalaa/, Old South Arabian /h l/ fall
ill (Koehler and Baumgartner 298). Gary A. Rends-
burg [p. c.] notes that this root is a by-form of /h ly/
become weak (Koehler and Baumgartner 300) =
Middle Hebrew and Jewish Aramaic /h l/.
11. /h m(h)/ sweat; butter
Ugaritic /xmt/; Jewish Aramaic /xm()/ (Koeh-
ler and Baumgartner 308)
12. /yre/ fear
Ugaritic /yr/ (Koehler and Baumgartner 399); Avi
Shivtiel [p. c.] adds Classical Arabic waraa to repel,
listed in Lane Vol. 8 (1893, 2933).
13. /kl/ restrain; keep from
classical and colloquial arabic archaisms 601
Classical Arabic /kalaa/, Egyptian Aramaic , Jewish Ara-
maic , Syriac , Geez kl (Koehler and Baumgartner
436)
14. /kisse/ seat
Akkadian /kuss/, Ugaritic /ks/, Phoenician /ks/,
Aramaic /krse/, Syriac /kursy/ with plene written
/u/, Classical Arabic /kurs/ < */kursi/ (Koehler and
Baumgartner 446)
15. /ks/ headdress of the moongod at the time of the full moon;
full moon
Akkadian /kusu/, Phoenician /ks/, Syriac /ks/ ~
/kas/ (Koehler and Baumgartner 446)
16. /lbo/ East of Jordan (Gary A. Rendsburg [p. c.] translates
entrance; on the way to.)
Assyrian /labu/, Egyptian /rbw/ (Koehler and Baum-
gartner 470)
17. /ys / go out
Phoenician , Ugaritic /ys /, Old South Arabian /wz/,
Geez ws ~ wd (Koehler and Baumgartner 393)
18. /lb/ lion
Akkadian */labu/ (Koehler and Baumgartner 472 state
that they follow Benno Landsberger ), Ugaritic /lbt/, Old
South Arabian /lb(t)/. Avi Shivtiel [p. c.] adds Classical
Arabic /labwa/ lioness, not in Koehler and Baumgartner,
but listed in Wehr (1974, 857). Vit Bubenik [p. c.] alerts
me to the alternant /labua/ lioness (Madina 1973, 591).
Te Proto-Semitic root is lb. Te form with the semi-
vowel is secondary.
19. /mle/ be full
Classical Arabic /malaa/ fll; /malia/ be full, Old Ara-
maic , Biblical Aramaic , Ugaritic /ml/, Akkadian /mal/,
(Koehler and Baumgartner 523); Geez /mala/, not in
Koehler and Baumgartner, added courtesy of Aaron D.
Rubin , [p. c.].
20. /ms / meet; fnd
Classical Arabic /ant/ < /amt/, Jewish Aramaic /ms/,
Biblical Aramaic /mt /, Syriac /mt / ~ /mt ay/, Old
South Arabian /mz/, Geez ms (Koehler and Baum-
gartner 553)
602 alan s. kaye
21. /mr/ feed
Classical Arabic /maria/ agree with (of food), Ugaritic /
mr/ (Koehler and Baumgartner 563)
22. /mass/ Masu
= /masu/ name of a North Arabian tribe (Koehler and
Baumgartner 570)
23. /n/ half done
Classical Arabic /na/ be raw, uncooked (especially of
meat) (Wehr 1974, 1014) (Koehler and Baumgartner 570)
24. nb denominative of /n/ prophet
Classical Arabic /nuba/ prophecy, /tanabbaa/ (Form
V) to prophecy (Wehr 1974, 937) (Koehler and Baum-
gartner 586)
25. nd drive (cattle); detach, remove from
Geez nd drive cattle
26. /n/ lif up
Classical Arabic /naaa/, Phoenician , Moabite , Ugaritic
/n/, Old South Arabian /n/, Geez ns (Koeh-
ler and Baumgartner 638) Biblical Hebrew /nm/
clouds; damp; fog, Classical Arabic /na/ hovering
clouds (Koehler and Baumgartner 638)
27. /n/ lay claim; lend
Middle Hebrew ny, Classical Arabic /nasaa/ grant
credit (Wehr 1974, 959) (Koehler and Baumgartner
638)
28. /sl(h)/ participle /msullm/ be paid pointing to sl
Classical Arabic /salaa/ pay promptly (Koehler and
Baumgartner 658)
29. /pl/ miracle
Classical Arabic /fal/ good omen (Koehler and Baum-
gartner 759)
30. /pr/ zebra
16
Classical Araic /fara/ wild ass; onager (Wehr 1974, 701).
Te Akkadian cognate is purmu. (Koehler and Baum-
gartner 775). Vit Bubenik [p. c.] states that Akkadian
par mule is closer phonologically than is purmu.
16
Although Koehler and Baumgartner (1998, 775) translate zebra, this is erroneous
for onager (thanks to Wolfart Heinrichs and Gary A. Rendsburg for this correction).
classical and colloquial arabic archaisms 603
31. /s b/ army
Akkadian /s abu/, Old South Arabian /db/, Geez /
s b/. Loanword in Egyptian /d bi/ (Koehler and Baum-
gartner 790)
32. /s me/ be thirsty
Classical Arabic /z amia/ (Wehr 1974, 583)
(Koehler and Baumgartner /zamiya/ seem to be errone-
ous.)
Ugaritic /zm/, Akkadian /s amu/, Old South Arabian
/zm/, Geez s m (Koehler and Baumgartner 806)
33. qy /q(h)/, impf. /yq/ vomit
Classical Arabic /qa/, Geez qy vomit, Akkadian
/qu/ [sic] for /qu/ excrete (Egyptian q vomit)
(Koehler and Baumgartner 83637) (also Koehler and
Baumgartner 1958, 83637)
34. qn /qinne/ be envious of
Classical Arabic /qanaa/ become intensely red, /qni/
blood-red, deep-red (Wehr 1974, 791) (Koehler and
Baumgartner 84243)
35. /qr/ read
Classical Arabic /qaraa/ read, Ugaritic /qr/ read
Koehler and Baumgartner (851) relate Classical Arabic
read with /qr/ II encounter, befall and not with /qr/
I call, recite > read (Koehler and Baumgartner 849).
In my opinion, this view is erroneous. Teir theory is
also expressed in Koehler and Baumgartner (1958,
849). Tey further note that Biblical Hebrew qry = /
qr(h)/ encounter (Koehler and Baumgartner 1958,
853) = Classical Arabic /qaraa/ to go about (not in
Wehr 1974) and Classical Arabic /qar/ receive as guest
(Wehr /qar, yaqr/ receive hospitably; entertain (1974,
761)
36. /rp/ heal
Classical Arabic /rafaa/ mend; darn, Old South Ara-
bian /rf/ repair, Geez rf stitch together (Koehler
and Baumgartner 903)
37. /ne/ hate
Classical Arabic /anaa/ hate (Wehr 1974, 487)
604 alan s. kaye
(Koehler and Baumgartner /aniya/ for Classical Arabic
appears to be erroneous). Ugaritic /n/, Old South Ara-
bian /n/ (Koehler and Baumgartner 925)
38. /b/ Sheba
Classical Arabic /saba/, Sheba Akkadian /sabe/ man-
kind (?), Old South Arabian /sb/ Sheba (Koehler and
Baumgartner 939)
39. /w/ a bad thing
Classical Arabic /sa/ be evil; bad (Koehler and Baum-
gartner 951)
40. /t/ room
Akkadian /tu/ room; guard chamber (Koehler and
Baumgartner 1015)
3. Conclusion
Te basic thesis of this work is to afrm that Classical Arabic should
stop being referred to as the preserver of all, or nearly all, original Proto-
Semitic phenomena. Tis paper suggests that certain Classical Arabic
forms are indeed secondary, as e.g., /l/ no discussed above. Consider
also Classical Arabic qalb heart < * lbb heart. Te latter root is also fully
preserved in Classical Arabic lubb heart = Hebrew leb and Akkadian
libbu < Proto-Semitic *libbu (= Proto Afro-Asiatic */lib-/ ~ */lub-/ (Orel
and Stolbova 1995, 362). Moreover, Egyptian ib supports Proto-Semitic
*libbu, while the vocalism in Classical Arabic lubb can be explained as
the result of regressive labial assimilation . Tus, Classical Arabic q- in
qalb is a remnant of a prefx of some kind or root determinative
17
used
in an old Arabic dialect or an earlier Semitic language antedating Clas-
sical Arabic (cf. Classical Arabic qadima or qadama to gnaw; compress
the lips (Wehr 1974, 544 < damma bring together further suggestive of
q- as a root determinative of some kind). Additional comparisons along
the lines of the present investigation of Classical Arabic and colloquial
Arabic dialects with other Semitic languages will undoubtedly further
17
Harsusi /h elbb/ and Mehri /h ewbb/ have /h -/ before the root /lb/ heart, which is
apparently lost in Soqotri /elbeb/ and Sh eri /b/. Mokilko (East Chadic) /ulbo/ heart
looks as though it displays */q/ > //, as occurs in many eastern Arabic dialects .
classical and colloquial arabic archaisms 605
demonstrate that spoken Arabic dialects occasionally preserve more
original forms than does Classical Arabic .
4. References
Hinds, Martin and El-Said Badawi . 1986. Dictionary of Egyptian Arabic: Arabic-English.
Beirut: Librairie du Liban.
Kaye , Alan S. 1972. Arabic /iim/: Arabic Synchronic and Diachronic Study. Linguistics
79, pp. 3172.
. 1976. Chadian and Sudanese Colloquial Arabic in the Light of Comparative Arabic
Dialectology . Te Hague: Mouton and Co.
Koehler, Ludwig and Walter Baumgartner . 1958. Lexicon in Veteris Testamenti Libros.
Leiden: E.J. Brill. (Abbreviated Koehler and Baumgartner 1958.)
. 1998. A Bilingual Dictionary of the Hebrew and Aramaic Old Testament . Leiden:
Brill. (Abbreviated Koehler and Baumgartner 1998.)
Lane , Edward William . 1893. An Arabic -English Lexicon . London: Williams and Nor-
gate. (Reprinted 1968, Beirut: Librairie du Liban.)
Madina , Maan Z. 1973. Arabic -English Dictionary of the Modern Literary Language. New
York: Pocket Books.
Moscati , Sabatino, Anton Spitaler , Edward Ullendorf , Wolfram von Soden, eds . 1964.
An Introduction to the Comparative Grammar of the Semitic Languages (Porta Lin-
guarum Orientalism). Wiesbaden: Otto Harrassowitz.
Orel, Vladimir E. and Olga V. Stolbova . 1995. Hamito-Semitic Etymological Dictionary .
Leiden: E.J. Brill.
Sobelman, Harvey and Richard S. Harrell , eds. 1963. Arabic Dictionary of Moroccan
Arabic . Washington, D.C.: Georgetown University Press.
Wehr , Hans. 1974. A Dictionary of Modern Written Arabic . 3rd ed. by J. Milton Cowan.
Ithaca, N.Y.: Spoken Language Services.
DO THEY SPEAK THE SAME LANGUAGE? LANGUAGE USE
IN JUBA LOCAL COURTS
Catherine Miller
IREMAM-CNRS of Aix en Provence, France
1. Introduction
1.1 Language Description and Language Uses: the unavoidable
discrepency
K. Versteegh s book (1984) on Pidginization and Creolization in Arabic
was one of the frst attempts to link two previously separated domains,
Arabic studies and Pidgin /Creole studies. Te controversy that followed
publication of his book contributed to difusion of information about the
main contemporary Arabic based Pidgin-Creole varieties of the South
Sudanese basin, namely Ki-Nubi and Juba Arabic (and former Turku ).
No serious Arabicist can now ignore the three above mentioned names,
and a number of publications have been devoted to the description of
these varieties. To recall but a few: B. Heine (1982), X. Lufn (2005),
U. Mahmud (1979), C. Miller (1984, 1992), J. Owens (1991, 1997),
M. Tosco (1993, 1995), R. Watson (1984), I. Wellens (2005), E. Yokwe
(1995).
Most of these studies provide a systemic description of Ki-Nubi or
Juba Arabic , and compare them with other Pidgin -Creole languages,
with other Arabic vernaculars, or with local African languages in order
to establish their specifcity and autonomy and to postulate hypothe-
ses about their genesis and development. Although still considered as
the orphan of the orphans (Owens 2001), marginalized in both the
Creole Studies and the Arabic studies, Juba Arabic and Ki-Nubi start
to be rather well described, even if many more need to be done. One of
the impacts of all these studies is the acknowledgement that Ki-Nubi
and Juba-Arabic are indeed autonomous and specifc varieties that can-
not be confused with any other Arabic vernaculars . Tey are identifed
through their specifc name; as it is well known, to name something is
to provide it with a specifc identifcation.
An interesting development of this linguistic achievement is the fact
that, since Ki-Nubi and Juba Arabic have been identifed as specifc
608 catherine miller
languages, the local speakers (Nubi people from Uganda or Kenya and
all Sudanese Southerners ) are expected to speak what the linguists have
described as Juba Arabic and Ki-Nubi . Tis is what is happening for
example in the case of language tests applied to asylum seekers in a
number of European countries. If somebody claims to be a Southern
Sudanese but speaks an Arabic variety close to Northern Sudanese col-
loquial varieties instead of the Juba Arabic described by the linguists,
he cannot be a true Southerner. Asked to provide some counter-exper-
tises, I realized that the way of speaking of these doubtful Southerners
was sometimes rather similar to some cases I recorded long ago in some
local courts of Juba , with well-attested Southerners!
In many contexts, most Sudanese Southerners constantly shif from
a more Creole level of Juba Arabic (or basilectal level) to a level more
infuenced by Northern Sudanese Colloquial Arabic (mesolectal level).
Te coexistence of diferent levels of Juba Arabic was frst highlighted by
U. Mahmud (1979), who applied the Creolistic concepts of continuum
and basilectal and mesolectal varieties to describe the variation found
in the verbal system. I dedicated a number of papers to emphasize the
degree of variation and diversity included under the generic label of
Juba Arabic (Miller 1984, 1987, 1989).
In the continuum approach, each pole (basilectal /mesolectal ) is char-
acterized by a number of specifc features (isoglosses ), while the inter-
mediate levels of the continuum are characterized by diferent degrees
of occurrence and melting of the various features. It appears, however,
that it is extremely difcult to draw an implicational scale that will be
characterized by a regular acquisition of dialectal features along the
continuum scale. In fact, each speaker tends to have its own way of mix-
ing the various features, some focusing more on phonological features,
others on lexicon or morphological features.
I myself, insisted on the fact that the evolution of Juba Arabic was far
from linear, and was not automatically leading to a process of decreoliza-
tion . Diferent trends of change and restructuration were simultaneously
recorded in a city like Juba and diferent infuences were operating on
the daily language; one leading towards a rapprochement to Khartoum
Arabic , the other toward what I have called a process of vernaculariza-
tion (Miller 1987).
Tis diversity and variability of Juba-Arabic raise the following ques-
tions: where are the boundaries of Juba Arabic ? On which criteria can
we decide that a person speaks or does not speak Juba Arabic ? Are the
language use in juba local courts 609
speakers conscious that they are mixing diferent features and variables
in their speech? Do the speakers and auditors have the same perception
than an outsider linguist ?
1.2 Language Uses in Juba Local Courts
In order to illustrate the diversity prevailing under the generic label of
Juba -Arabic , I choose to present complete samples of interaction, instead
of short extracts given as examples. Tese samples were recorded in
1981 and 1984 in two local Courts of Juba , the capital-city of Southern
Sudan . Local courts were of particular interest because very diferent
kinds of people were coming to present and defend their cases. Taking
into account that more than twenty years have passed since the mid
1980s, and that almost nothing is known about the present linguistic
reality of Juba , such samples could be considered as a kind of linguistic
testimony of what speaking Juba Arabic meant in the 1980s.
Te term local court and its Arabic equivalent makma ahliyya ~
makama ahliya was applied to Southern Sudanese A and B Courts at the
levels of the village, the district or the city. Various languages were used
in these local Courts.
Outside Juba , the A Court was the villages Court headed by the
villages chief, assisted by two or three elders. Te language of communi-
cation was, most of the time, the local African vernacular (Bari , Kakwa ,
Zande , Moro , etc.). Te B Court was the district Court headed by the
paramount Chief assisted by the village chiefs. It dealt with matters not
solved in the A Court. In multilingual districts of Equatoria , such as Yei
B Court, the languages spoken were local vernaculars (Kakwa , Moro ,
Madi , Avokaya ), Juba Arabic , Bangala (the local variety of Lingala ) and
a few English .
In Juba , the only A court was the Garawiyya , located in the center
of Juba, near the main market, and dealing with low-level criminal cases
(robbery, neighbors quarrels, etc.) involving any person living in Juba,
whatever his ethnic afliation and period of stay. Te court was headed
by an appointed local judge (Ramadan , 49 y. old) and two assistants. Te
main language of communication was Arabic (Juba Arabic up to North-
ern Sudanese Arabic (NSA ). A few speakers used an African vernacular
and were assisted by a translator. Te summary of the case was always
pronounced in Arabic but written in English .
Te Kator B court was the Bari local court (the Bari being the main
local group of the Juba-Rejaf area), located in Kator district. It was
610 catherine miller
dealing with personal status cases involving at least one Bari participant.
Te President of Kator B court was a Bari (Stephen , a retired police-
man), assisted by two or three Bari elders. Te language of communica-
tion was mainly Juba Arabic , Bari and a few English .
Te localization and specialization of each court infuenced the
language use. In Juba , the Garawiyya court was far more exposed to
the infuence of NSA than the Kator B court. In the Garawiyya, only 4
speakers, not resident in Juba, spoke in their native vernaculars , while
35 speakers spoke only in Arabic (with 16 speaking a more mesolectal
variety and fve speaking NSA). In Kator, 10 speakers spoke mainly in
Bari against 24 in Arabic (with none speaking NSA and 14 more or less
mesolectal):
FIGURE 1 LANGUAGE USES IN KATOR AND GARAWIYYA
LOCAL COURTS
Courts Speakers* Non-AR Vernaculars Juba Arabic NSA English
Kator B 34 10 Bari 24
Garawiyya 39 3 Bari
1 Dinka
30 5
Total 73 14 54 5 0
* Number of speakers does not include the Judges
Te above fgure classifes each speaker according to the use of one
dominant language (English , NSA , JA and Non Arabic Vernaculars ). In
fact, the delimitation between each language was not that clear-cut and
there were many cases of language mixing, as will be evident in the Cor-
pus presented in 5. Appendices, the texts: 5.1, 5.2, and 5.3.
2. Presentation and Brief Analyse of the Corpus
Te samples presented in the Appendix are taken from three cases, con-
sidered to be representative of three types of speaker. Te participants
involved in Case 1, belong to the important Nilotic Dinka -speaking
group, a non-local Juba ethnic group. Not well-educated, they speak
Juba-Arabic as a second language. In Case 2, the participants belong to
a small ethnic group from Western Southern Sudan (Bay from Bahr al
Ghazal Region ), are educated, work in the police or the army, and speak
a level of Arabic very much infuenced by NSA and WSA (Western
language use in juba local courts 611
Sudanic Arabic ). In Case 3, speakers are all Bari Juba citizens and speak
Bari or a vernacularized Juba Arabic , very much infuenced by Bari
features.
In both courts, the procedure was rather similar. Te case starts with
an oath (on the Bible , the Qurn, or a spear according to the partici-
pants religion),see (s.9) in Case 1 and (s.14) in Case 2 . Ten the plain-
tif tells his/her story; afer some questions, the judge summarizes the
story and asks the defendant to present his/her own version. Additional
questions follow; sometimes, additional testimonies are required. Afer
listening to all parties, the Court deliberates and announces its verdict.
While most cases took between 30 to 40 minutes in Juba, they could
take hours in the villages. Cases tended to be shorter in the Garawiyya
than in Kator .
Speakers are presented as follow: J.= Judge S. = Courts Secretary. T. =
Translator, P. = Plaintif, D. = Defendant and W. = Witness. Elements
in italic are considered to be average basilectal Juba Arabic . Elements
in bold are considered to be more mesolectal . Underlined elements
are borrowings from the local languages, mainly Bari . Plain words are
English borrowings. Many elements are common to both basilectal and
mesolectal Juba Arabic . Tey have been put in both italic and bold, but
the distinction between more basilectal or more mesolectal remains
very fuid. Te transcription of the Bari sentences have been done in
1987 by a Bari student, who followed the standard Bari written system
used in Equatoria .
Mesolectal features taken into consideration include a) phonologi-
cal features (consonants , x, , t , d, s , , h , long vowels and geminates ),
b) morphological features (object sufx pronouns , defnite article , plu-
ral and gender afxes , TMA and person verbal afxes ) c) lexical items.
It must be noted that the insertion of mesolectal features lead ofen to
mix forms , which are neither Northern Sudanese Arabic (NSA ), nor
Creole Juba-Arabic . African vernacular features taken into consider-
ation include:
a) phonological features (p, , , , , , , j strong palatalization and
africation );
b) lexical items and idiomatic expressions . Syntactic features will not be
discussed here.
Te brief analysis that follows intends to higlight the individual level as
well as main examples of mixing with focus on mesolectal insertion. But
612 catherine miller
readers are invited to go to Appendix 1, 2 and 3 in order to evaluate the
degree of mixing within each individual speech.
2.1 Text 1, Te Dinka Case, Garawiyya Juba , 1981. (Transcript in
Appendix 1)
Summary: A Dinka woman coming from a village accuses a Dinka man
living in Juba to have rob her a sac of sorghum when she came down
from the ferry. She speaks only Dinka and her speech is translated by
another Dinka from Juba. Te case started the day before but was post-
poned in order to look for an additional witness who could prove the
identity of the defendant. Te woman afrms that he is called Maluk
while the defendant claims to be called Deng . At the beginning of the
second audience, the secretary summarizes the facts (s.1) and then the
Judge is looking for the witness (s.2 to s.9) who fnally arrives and sup-
ports the plaintif against the defendant that he knows since 1975 (s.10
to s.20); the later is fnally recognized culprit by the Court (s.21).
Tis case illustrates a type of language use rather common in the
Garawiyya Court, that I will defne as a formal basilectal non vernacular-
ized type of Juba-Arabic . Tis type of speech is very common among non
L1 speakers of JA (rural people, older generation) in formal situations
and lead to many mix forms . Although it contains most of JA basilectal
features (such as verbal invariable forms), it is characterized by numer-
ous instances of phonological variation (s/; k/x; j/z; a/, etc.) and by
the frequent dropping of the verb s fnal epenthetic vowel (kelem vs. JA
kelemu speak, gal vs. JA gale say, baaref vs. JA bi-arifu know, etc.). A
number of lexical items are taken from NSA like der want (NSA dyir/
JA aoju), bet house (NSA bt/JA jua), particle aw or (JA wela), etc.
Some old terms inherited from the Ottoman military language remain
like dosoman troubles, fght (cf. s.8) and appear to be part of the court
vocabulary. One notes the irregular insertion of a number of mesolectal
morphological features borrowed from NSA such as defnite article al-
(l/el) the, verbal sufxes and prefxes as well as afx pronouns . In Case
1, all speakers speak this formal non vernacularized JA , with various
degree of mesolectal insertions.
Te two Dinka participants (the translator and the witness) use inde-
pendent personal pronouns and invariable verbal stems like Translator
(T.) in sentence (s.3) nina m dr mi we dont want to go or sentence
language use in juba local courts 613
(s.5) ya na klem le ede I talked to him like this. Te pronuncia-
tion shows the infuence of the Dinka Mother Tongue , particularly con-
cerning the realization of the phonemes // /s/ and /z/ realized as [] (a
frequent pronunciation among Nilotic speakers ) like in en what (JA
senu /NSA en); ol somebody (NSA zl/ JA jol), etc. Vernacular infu-
ence is also noticeable in the realization of the vowels , with a tendency
to realize /e/ like an open [] and /o/ like an open []: k (s.3), le (s.5),
bls (s.12), barf (s.13), nugasr (s.12), b and ka (s.16), etc.
Te Dinka witness (W) realizes a number of infected verbal forms
and afx pronouns :
(s.12) nugasr (JA ana bigeseru/NSA agaar)to limit, shorten, (s.20) axk
(JA aku btak) your brother, gutta lek (JA ana gale le-ita) I told you,
negabel (JA . ana bigabalu/NSA agbil) I meet, nesalem (JA ana biwodi
salam to/NSA asallam) I great, neddu (JA ana wodi leo/NSA nadd) I
gave him). It may be noted that 1st sg. imperfective form is n- + verb
(similar to WSA ) and not a- + verb (= NSA).
Te secretary (S) has a JA pronunciation: (s.1) asrin twenty (NSA
arn), asan because (NSA an), jl person (NSA zl) but izebu he
brings (JA jibu/ NSA ijib).
Te Judge displays many instances of phonological variation: z/j (s.2)
win zl el-brefu jl da (with dominance of [j] cf. (s.4); x/k (s.6) axu
brother, (s.15) kemsa fve, (s.17) inta bikutu you put; /s (s.4) aan/
asan because, (s.17) musu isnt it.
He keeps many invariable verbal forms : cf. (s.4) nina il we take, (s.8)
nina balasu we stop, (s.21) inta bigul sehi you say right, etc. and many
independent personal and possessive pronouns : (s.9) el-mahkma nadi
eta the Court calls you; (s.15) inta baarfo aglaK bitao you know his
behavior? (s.21) weled btak suker winu where is your small boy?
Te mesolectal variables include the 2nd sg. personal pronoun inta
you (JA ita), pl. deictic del this (JA de), defnite article l (JA de) and
a number of conjugated verbal forms : (s.9) twori you show, (s.21)
kizibta you lied, sarkta you stole, etc. Te conjugation of verbs leads
to some cases of malapropism such as (s.9) nasaltak we ask you with
the coexistence of n- 1st pers. imperfective and -t 1st or 2nd pers. per-
fective . Another mix form is the verb gul in (s.4) bas iji wri gl he
just come to tell that (JA bas bija wori gale/ NSA iji igl or iji iworri
innu). Te insertion of mesolectal features increases at the end of the
case (s.22), when the Judge is enouncing his verdict and is using a num-
ber of technical terms such as tazwir falsifcation, jiza faults/penalty.
614 catherine miller
Note however that the Judge always alternates basilectal and mesolectal
realizations like the alternation sarkta/ita sereG in (s.22).
A number of mix JA /NSA realizations recorded in Case 1, were almost
permanent in the Judges speech of the Garawiyya and appear also in
Case 2. Tey seem to indicate that, indeed, this type of mix level was
considered as the appropriate level in the Garawiyya Court.
2.2 Texte 2, Te Bay Case, Garawiyya Juba , 1981. (Transcript in
Appendix 2)
Summary: a young Bay police-lady (Hawa ) from Bahr al Ghazal accuses
two Bay guys to have tried to attack her the night before. She recalls
that she was going back to her home when the two young men started
to insult her and beat her. Fortunately she was able to call for help and
a soldier came to rescue her and, afer many events, managed to bring
the two boys to the police ofce. Te two defendants contest the ladys
version and claim to be victims of a confusion, i.e. the soldier took them
instead of the real culprit. Te lady and the two boys agree that they
never met before, in spite of the fact that they all belong to the same
small tribe (Bay). Te two Judges argue between them concerning the
importance of this common ethnic background. Te soldier is called as
a witness and makes a very long testimony, which supports the police-
womans story. Afer many discussions and controversies, the two boys
are recognized culprits and have to pay a fne of 36 Sudanese pounds.
Tis case illustrates a more mesolectal level, mastered by the persons
who have been educated in Arabic and particularly those coming from
the region of Bahr al-Ghazal. However, each speaker has his specifcity,
and while the young lady is the most infuenced by colloquial features ,
the men alternate between basilectal and mesolectal features and cre-
ate many mix forms . A number of lexemes appear to be shared by all
the participants such as der (JA auju) want, bet (JA jua) house, gul (JA
gale) say, fat (JA futu) pass, darab (JA dugu) heat, suf/uf (JA ainu) see,
le-raet-ma (JA lakadi) until, etc.. At the level of the verbal system, one
notes many occurrences of participial forms such as jay coming, gaad
staying, el working and the frequent use of TMA auxiliaries/verbal
particles like kan, bikun, gam/gum, gaad, lisa, biga/baga, ja, aoz.
Hawa (s.15 and s.17), the young policewoman speaks a level very close
to NSA or rather to Western Sudanese Arabic (WSA ). She pronounces
many long vowels , as well as velar fricative /x/ and sibilant // but she
realizes few pharyngeal and emphatic consonants : i.e. (s.15) saa watch
language use in juba local courts 615
(NSA saa), saba morning NSA ab), talta I came (NSA t alat) but
end of (s.17) sa
'
a fekka the watch dropped. She conjugates almost all
verbal forms : cf. (s.15) jit, saketta, zitta, hum zdu and (s.17) gumta,
wagefu, gulta, tetfahim, netfahim, maet, etc. Object pronouns and
possessive pronouns are systematically sufxed : (s.15) garas-ni f-s ulb-i
he pinched me in my buttocks; (s.17) dagga-ni, axwn-i, masak-ni,
xam-ak, bedugg-k, etc. She uses a number of TMA auxiliaries, like
gam and ja to mark inchoative:
(s.15) ana gumta zitta katwa btay I accelerated my walk
(s.17) da tawli masakn ja garasni He immediately sized me and
start to beat me
waled de ja gam darabni bunya Te boys started to beat me
strongly.
Hawa masters the morphological rules of NSA and shows very few
occurrences of malapropism or mix form . Gender and plural agreement
however remains irregular: (s.15) ana jay I came-masc., ana mya
I went-fem.; (s.17) arabiya aba yagif the car (f.) refused (masc.) to
stop (masc.), awld der iyamal mayi mukel boys (pl.) want (sg.) he-
makes with me problems/ the boys want to make me problems.
Defendant 1 (John Gabriel , s.28, s.30, s.34) starts answering both Judges
with mesolectal features :
(s.7) J2 sgol btk ? Your work D1: m el/ t leb Not
working/student
(s. 24) samtu I heard-it
(s.28) ana kunt jay min hay kmryal li-l-bT I was coming from district
Commercial to the house.
Ten he alternates between few infected verbal forms and many
invariable verbal forms as well as between afxed and independent
pronouns :
(s.30): ma blis gaadn isugg zl de / uman ft giddm/ ana ja wara/ nama
ana wosol mbil/
/he and police stay-pl they-drive person this/ they pass (sg.) in front/ I
came (inv.) behind/ when I arrive (inv.) Mobil/
He and the police ofcer were driving this person, they passed in front,
I was coming behind when I arrived at the Mobil Station.
(s. 38) bass darab bitoman ana ja katal /only fght of-them I came (inv.) kill
(inv.)/
I just came to stop their fght.
616 catherine miller
Defendant 2 (William Peter , s.45, s.47, s.51) realizes many infected ver-
bal forms and afx pronouns :
(s.52) gabodni zamn gale nemi-l-nukta niuf/ min wisolta unak an
nera mukila bta dn kif
/they-size-me (3rd perf. ) time say I-go (1st imp. ) the post I-see (1st
imp)/from I-arrived (1st perf.) there for I-see (1st imp) problem of
John how/
Tey captured me when I went to the police station to see/since I
arrived there to assess the problem of John . . .
Like the Dinka witness in Case 1, William uses the WSA n- 1st sg. imper-
fect pers. pronoun instead of NSA a-. Note also the use of verb gale say
to introduce an embedded sentence (zamn gale time that).
Te witness (W. the soldier) starts with a formal mesolectal level:
(s.75) ana kunta aal f-l-jawazt, ana sken f Muluk I was work-
ing in the passports, I lived in Muluk. But very quickly he alternates
between basilectal and mesolectal features : ana jit ja wosolu f medresa
/ana ft/ falan fatet.. I arrive near the school/ I passed/ of course I
passed; baadin ana jt ana gul when I arrived I say, ana gulta tayib
ya axwna matakum sakal bit f tarka zede I said good oh my broth-
ers, dont quarrel a girl in this way (note JA negative imperative marker
matakum dont!).
He ends up speaking almost only Juba-Arabic , including at the pho-
nological level ( >s, x > k). Unlike the speakers of Case 1, his level of
Juba Arabic is more grammaticalized . It includes many verbal TMA
markers as well as embedded sentences:
(s.75bis) d tawli gum amol musakl fo mbili li raat ma aoz kaser mrat bta
arabiya zatu
He immediatly started making problems in the Mobil station until
he was almost breaking the mirror of the car
ana bija ft f jawazt ya neselem kelem kabr li jamaa f taakir el ana
akr line zaman
I was going to the passeport (ofce) in order to bring the news to
the people about the delay that I delayed before.
Compared to Case 1, Judge 1 (Ramadan ) shows the same level of pho-
nological variation (s/, x/k, z/j) but uses more lexical and morphologi-
cal mesolectal features (see s.25, s.37, s.52, s.74, s.79). It may be noted
that in (s.25) when Judge1 is summarizing the testimony of the police-
language use in juba local courts 617
woman, he makes more gender agreement than she did and employs the
2nd and 3rd fem. personal pronouns inti and iya:
(s. 25) el-bint de gl iya maya Te girl said (masc.) that she was going
(fem.);
inti jya min wn you (fem.) coming (fem.) from where?;
saa btao fakkat waga watch (fem.) of-him drop (fem.) fall (mas.).
But he also mixes with more basilectal JA and consistently uses the
invariable form gul say:
(s. 25) intakum gul inti ermuta sakeT you-pl say you (fem.) prostitute only
(s. 52) nker u gl woket dak kalam inti gul ma hasal klu-klu
deny and say time-that word you (f.) say not happen all-all/
He denied and said (that) at that time what you say never happened.
As in Case 1, the conjugation of verbs and auxiliaries leads to cases of
malapropism or mix forms .:
(s.52) asa nafarn de badi sumit kalamu /now person-Dual this afer I-lis-
tened word-his/
now afer I listened the word of these two persons (sumit vs. JA asuma
and NSA samit)
f tarka gmo daribtu bi daraba adid /in way they-stand you-beat
with beat strong/ On the way you started to beat her strongly (gmo
vs. NSA gumtu).
Compared to Judge 1, Judge 2 (see s.66, s.70, s.76, s.78), sticks to a more
basilectal pronunciation ( > s, x > k, etc) and grammar :
(s. 66) musu barau rtan bta bay keda wonos /Neg alone language of Bay
like-this spoken/
Isnt it a specifc Bay language which is spoken? (note the impersonal
structure rendered by shif of stress to the fnal syllable of wonosu).
Case 2 highlights the coexistence of diferent levels (types?) of (Juba)
Arabic and diferent degrees of mixing. Te level of the police-lady and
the intermediate level of Judge 1 and D2 could be an indication that the
mesolectal level is indeed the prestigious urban formal level that people
speak or try to speak in formal context like the Court. Tis hypothesis is
however not sustained by the data from Kator B Court, which indicates
another type of urban use.
618 catherine miller
2.3 Texte 3, Kator B Court, Judge Stephen , 1984. (Transcript in
Appendix 3)
Summary: Tis case starts in a noisy atmosphere and the Judge Stephen
asks for silence from the audience (s.1). Te secretary and the police-
ofcer try then to stop a person who wants to talk (s.2 to s.5). Te case
concerns the non-payment of a dowry. Te plaintif Gabriel (from s.10
to s.30) accuses the defendant (Santino ) to have married his sister dur-
ing the frst civil war and to have never paid the dowry. Te sister had
died and Santino did not take care of the four children, and one died in
an accident. Gabriel now takes care of the children and is asking for the
payment of the dowry. From (s.31) to (s.33) there is another interrup-
tion from a woman. In (s.35) the defendant explains why he was in the
incapacity of paying the dowry. A number of relatives come to give their
testimony (parts not reproduced in the Appendix). At the end, the fam-
ily of the defendant agrees to pay the remaining of the dowry and the
Judge Stephen (s.36) asks everybody to reconcile.
Te plaintif is speaking in Juba Arabic while the Defendant is speak-
ing in Bari . Te Judge Stephen , as well as the police and the secretary
shif between Juba-Arabic , Bari and a few English (see s.30 for an exam-
ple of code mixing Bari-English).
Tis Case illustrates a type of language use very common among
the Bari population of Juba and surrounding areas, with a deep inter-
penetration between Bari and Juba-Arabic , which certainly helped the
vernacularization of JA .
When the Judge Stephen and the plaintif are speaking in JA , one
notes the phonological infuence of the Bari languages with realizations:
f > p pi in (NSA fi); pogu on (NSA fgu),
z > j julumin spoiled (NSA mazlmn)
> aat banaat girls (JA banat/NSA bant), maat die (JA mutu/
NSA mt)
e > gdiyat case (JA gediya/NSA gadya) (// is noted in Bari
script and by Bari speakers).
A number of words are common to both JA and Bari like mali dowry
from Arabic maal money, or kurju cultivate from Bari kuruju. Some
expressions are idiomatic translations from Bari expressions such as
(s.38) tusu bujak ~tufu buzaK spit saliva, i.e. give benediction.
language use in juba local courts 619
When the defendant is speaking in Bari , one notes the presence of a
number of Arabic (sometimes from Ottoman military Arabic) words like
temporal markers kan was/before, badin then/at that time, asa now,
discourse markers taban of course kalas OK, numerals (all numbers
are in Arabic ) and words linked to administration or political events
such as hawadis civil war, hukum judgement, korbaat whip.
Inter-penetration between the two languages goes over phonological
and lexical transfers. It includes also syntactic similarities like the use of
verb say (gale in JA , adi in Bari ) to introduce embedded clauses (Miller
2001).
(s.7) taban ana sufu zede gal kan uwo ro ptisu mal . . . .
/naturally I saw like-this say if him go look dowry . . ./
I realized that if he was going to look for the dowry
(s. 36) kirut adi monye sarji adi ukum nagwon kju monye a ukumbe adi asa
man dek pitn kwe
Ten her father sent a message that the judgement passed before was
that now I want the remainder of my money.
Note that almost none of the mesolectal features , so common in the
Garawiyya , were recorded in this case. I noticed that the Judge Stephen
never tried to imitate a mesolectal level, even if some of the young par-
ticipants were talking in mesolectal JA . If he had to alternate and adapt
to the speaker, it was between JA and Bari and not between JA and
mesolectal. Stephens way of speaking is still a rather formal and juridi-
cal way of speaking and cannot be associated with an informal level of
JA . However his speech (particularly s.37), as well as that of Plaintif
Gabriel , was rather similar to the level of JA broadcasted by the radio of
the Sudan Council of Churches and is rather similar to what linguists
will consider as Juba Arabic (see Watson (1984) in particular).
3. Conclusion
Te three cases illustrate diferent levels/types of Juba-Arabic . Of
particular interest was the attitude of the two Judges, Ramadan in
the Garawiyya and Stephen in Kator . Ramadan realized a number of
mesolectal features and used some Arabic Juridical terms. However, he
was adaptating his speech level to his interlocutors (cf. compared degree
of mesolectal insertion between Case 1 and Case 2) and he never tried
to assert his Authority through the exclusive use of a high linguistic
620 catherine miller
norm. He also never commented about the language use of the other
participants. As for Stephen, he always spoke a basilectal vernacularized
Juba Arabic and was playing between Bari and Juba Arabic . He ofen
made some comments about the Southern identity (ena f januub here
in the South, kalam bta januub words/languages of the South, arabi
bitana bta januub our Southern Arabic) and once criticized a young
man talking like a Northerner.
Te language use of the Judges indicate that while the Garawiyya
Court was infuenced by its surrounding Arabized Malekiyya neighbor-
hood, the Kator B Court was infuenced by the Bari surrounding, even
if most speakers prefer to speak Arabic rather than Bari. Two types of
urban models were present here. On the one hand, the Northern Suda-
nese Arabized urban model symbolized by the merchant community
(both Arabs and non Arabs living near-by), which still had an infuence
upon part of the Southern population (old settlers as well as newcomers).
On the other hand, the East African urban model (Nairobi , Kampala )
brought back by the returnees and supported by the local Churches and
the political Southern activists. In 1984, at the breaking of the second
civil war, these two urban models were still coexisting together with
more rural traditional ways of life. Language diversity, as well as religious
and ethnic diversity, were considered natural components of the city.
How far did somebody like Hawa , the young Bay policewoman,
and Stephen , the former policeman or Gabriel consider that they were
speaking the same language (Juba-Arabic ?), or that they were speaking
two diferent languages? Tis was a question that I did not ask at the
time of recording and that I cant answer. I later worked with some Bari
informants in Khartoum . Tey could very easily reproduce Creole /basi-
lectal Juba Arabic features, when asked to do so. For those who knew
NSA , they were perfectly able to distinguish between the two systems.
Tey had therefore a clear consciousness of what JA was, as a distinctive
linguistic system. I noticed, however, that when I asked some of them
(students trained in linguistics ) to transcribe some of the tapes recorded
in Juba local Courts , they tended to systematically transcribe them in a
basilectal Juba Arabic phonology , without reproducing the mesolectal
variations. It was as if, for them, people from Juba were speaking one
language. I did not discuss with them the reasons of their attitudes and
I dont know if they were not aware of these variations.
Te recording of natural corpora help to better grasp the natural
diversity. But it makes the linguistic analyses more complex and renders
the concept of autonomous linguistic system rather problematic. As it
language use in juba local courts 621
is also the case with Standard and Colloquial Arabic , it is possible to
describe the two poles of the continuum as two diferent systems. But
it is hardly possible to decide where the boundary is between these two
languages within the continuum.
4. References
Heine , Bernd . 1982. Te Nubi Language of Kiberia. An Arabic Creole . Berlin: Dietrich
Reimer.
Lufn , Xavier . 2005. Un crole arabe : le kinubi de Mombasa , Kenya . Muenchen: Lin-
com.
Mahmud , Ushari . 1979. Variation and Change in the Aspectual System of Juba Arabic .
Ph.D. thesis, Georgetown University.
Miller , Catherine . 1984. tude socio-linguistique du dveloppement de larabe au Sud
Soudan . Tesis, University of Sorbonne Nouvelle-Paris III.
. 1987. De la campagne la ville. volution fonctionnelle de larabe vhiculaire
en Equatoria (Sud Soudan ). Bulletin du Centre dEtude des Plurilinguismes (Nice) 9.
126.
. 1989. Kelem kalam bitak: langues et tribunaux urbains en Equatoria . Matriaux
Arabes et Sudarabiques (Paris) 2. 2358.
. 1993. Restructuration morpho syntaxique en Juba-Arabic et Ki-Nubi : propos du
dbat universaux/superstrat/substrat dans les tudes croles. MAS-GELLAS Nouvelle
Srie 5. 137174.
. 2001. Grammaticalisation du verbe dire et subordination en Juba Arabic . Leons
dAfrique. Filiation, rupture et reconstitution des langues: un hommage G. Manessy,
ed. by Robert Nicola , 455482. Leuven: Peeters.
Owens , Jonathan . 1991. Nubi , Genetic Linguistics and Language Classifcation.
Anthropological Linguistics 33, 130.
. 1997. Arabic -based Pidgins and Creole . Contact Languages. A Wider Perspec-
tive, ed. by Sarah G. Tomason , 125172. Amsterdam-Philadelphia: John Benjamins
Publishing.
. 2001. Creole Arabic : the Orphan of all Orphans. Anthropological Studies 43,
3. 348378.
Tosco , Mauro and Jonathan Owens . 1993. Turku : A Descriptive and Comparative
Study. SUGIA 14. 177268.
Tosco , Mauro. 1995. A Pidgin Verbal System: Te case of Juba Arabic. Anthropological
Linguistics 37, 4. 423459.
Versteegh , Kees . 1984. Pidginization and Creolization : the Case of Arabic . Amsterdam-
Philadelphia: John Benjamins.
Watson , Richard . 1984. Juba Arabic for Beginners. Juba: Summer Institute of Linguistics .
Wellens , Ineke . 2005. Te Nubi Language of Uganda . An Arabic Creole in Africa . Leiden:
E.J. Brill.
Yokwe , Elisai M. 1985. Te Diversity of Juba -Arabic . Studies in African Linguistics ,
Supplement 9. 323328.
622 catherine miller
5. Appendices
5.1 Te Dinka s case: Garawiyya Juba , 1981
(1) S. ombre gedya f mantr tultu mya tnn u asrn / tabn ombre azim
asan mra de ibu ns/ jl el ga-ref asn zdu bta jl de / gal yani
de wa gl mlk/
Yesterday the case in the register (was recorded as) three hundred
twenty two. Of course it has been reported so as the woman could bring
people, somebody who knows him [i.e. the defendant] and could tes-
tify. Because the man said he is called Deng and she said he is called
Maluk .
(2) J. win sud/ zl el-brefu jl da / inta
Where is the witness? Te one who knows this person? You?
(3) T. ama ol igl en ya wktu / gl ns tmon f k k / lkan mai
lomon / oman gl zl de armi / mumkin f jerma tni nina m dr mi
/ umon kfu b / m dr msi/
Listen she says what Wakotu (??) /she says that their people are in
Konyo Konyo/ if she goes to them / they say this person is a thief / maybe
there are other crimes / we dont want to go / they are afraid of him, they
dont want to come.
(4) J. Nna bas l haj-e-en/ nna ndin jl /wo biji an ije-wri isim ta jl de
bs / ms kalm tni/ bas iji wri gl ya mahkama na brefu jol de / sim
t fln / asn wo indi tnn isim ena / jl de biwri yat el-numro biyref
fg / bes kalm tni m f/
We take him for what? / we called the person / (so that) he comes to
show the name of the person only / Nothing else / He just comes and
tells Court I know this person / his name is so / because he has two
names here / Te person tells which number (name) he knows / there
is nothing else.
(5) T. Ya na klem le ede bt nagfa / gl l/ lkn intkum dr kalam/
klita ol ir wodna ma bl f k k
Indeed I talked to him like this but he refused / he says to me / if you want
words / lets somebody go take us with the police in Konyo Konyo.
(6) J. La la / tyeb hwa indu xu / jl bita f / rajel bita wn ?
No, no / OK does she have a brother / does she have somebody / where
is her husband?
(7) T. ay f f bra
Yes he is outside.
language use in juba local courts 623
(8) J. Tyeb / kan uwo ra ma axu bitao / ma rajel bitao / uwa birh wri nas
del / yajeb nas del / ma f kalam dosoman / keli nasma blis lokan f blis
Good / she should go with her brother / her husband / she goes to tell
this people / she brings this people / there will be no troubles / lets listen
to the police if there is a policeman.
[interruption] Te policeman refuses to go with the woman and says:
/ you can not approve a sick man / you cannot approve it a all / because
we blis ma negder/. ./
(8bis J.) its up to us / lakin de ya ma besma min kalm tay aja zede / de fkra ana
kelem ta sakeT maales / kan keda kwayis / izakan u ma jebu nas de / jol
/ nas el beld el ja wri isem / nina balasu-l-gedya /
Its up to us / but this one does not listen to me / this was a simple
idea / no matter / if so ok / if she does not bring the people / the per-
son / people of the village who come to show the name / we will end
the case.
[the case is suspended, the woman goes outside and comes back with a
witness]
(9) J. inta bahalef harba wala bk / bibl / asma el-mahkma ndi eta aan
haja / nasaltek aja / twori kalam mosbt inta baaref / eta ma tddisu
/ mara de gl aslu inta baarfu jl de / jl de gabel keda besuf ?
You swear on the spear or on the Book? / the Bible ? / Listen the Court
call you for something / we ask you something / you tell the right thing
you know / you dont hide / this woman say that you know this person /
this person did you see her before?
(10) W. jol de / ana baaref yisim bita / isem t maluk / maluk aywen
Tis person / I know his name / his name is Maluk / Maluk Aywen
(11) J. inta arfu da mitin
Since when do you know him?
(12) W. Ya zl taban ana bigul neya /asa jina asa al ma ksel ydi / fa wd
de ma al u bls masa alna / u wd de taban u barefu zol de taban /
lakin u mubleK / badelma ana baref zol de / ana ma mumken nugasr
gediya bita l / interruption /
O naturally I will say the truth / actually I work as a cleaner / and
the boy came to me and the police came to me / and this boy of course
he knows this person / but he is the plaintif / since I know him (not
clear if it is the plaintif or the defendant) I cannot shorten(?) his case
if
624 catherine miller
(13) J. tegriban jo kem sanawt eta barfo?
Approximately, for how many years do you know him?
(14) W. aywa s / min kema u abain
Yes true / since 1975.
(15) J. inta mudda de / min kemsa u sabain li haddi nahar de / jl de min de
inta brfo uwo / hl inta baarfo aglaK bitao / u jl kwayis wala kan
indu jerima bta nas biil haja bta nas
You, during this period / since 1975 until now / this person since you
know him / do you know his behavior? / Is he a good person or someone
who has crime of people stealing things of people?
(16) T. s kalm ya ayed el-reys / inta tab
y
an ta-aref zol wa inta yikun
kunta g
y
ad f bet wd / inta ma botaaref tani jol de / bref b u binum
wn / inta ainu f ka / teselem ydu u bref taban / u lekin inta m
ref baaml enu aw baaml enu
You are right Mr. the President / Of course you know a person if you
are staying with him / you dont know else / this person you just know
where he sleeps / you see him on the way / you great him you I know
him of course / but you dont know what he is exactly doing.
(17) J. m batl / Deng Mayen inta bikutu isomu bitak musu Deng Aywen /
lakin dk bigul isom btk Malk Aywen / indak sual le / asal /
Not bad / Deng Mayen didnt you put your name as Deng Aywen?
/ But this one says that your name is Maluk Aywen / Do you have a
question for him? / Ask!
(18) D. uwa d / ana indu sual wa gidamu / uw akyuT / ana nam seta yom
/ uwa ma adana akel wala mya / de lujal jr / [laughter in the room]
/ mara de dr fgri ana bi bi b-l-awanta
Tis one / I have one question for him / Is he a brother? / I slept six days
/ he did not give me food nor water / Hes a ?! / Tis woman wants to
make me poor by her tricks!
[Te witness and the defendant start to argue between themselves in Dinka ]
(19) J. mutarjem winu where is the translator?
(20) W. nerja le kalam de / uw gal enu / izakan ana axk ita addni futur aw
ita adni ay / u gal keda / lekin ya assi ana gutta lek jebel / l-zol de
ana baaref u-binum wn / wa baamol enu f-l-yom u baamol enu ana ma
baaref /negabel f ska nesalem yeda / fa neddu l -akel/ m indi akwn
rabto kulu
Shall I return to you the words? / What did he say? / if I am your brother
you give me breakfast or tea / He said like this / but I told you before /
language use in juba local courts 625
this person I know where does he sleep / and what does he do during the
day I dont know / I meet him on the way, I great him / and I would give
him food? / Didnt I have brothers that I all raised?
(21) J. to W. izakan ita kan kazb / inta begul sehi lama kan l-mahkma sala
/ lokan gal ana aarfa kalam btau uwa jol batl / lakin inta bet-
kelem ez-zn / l-haG / inta gul ma baaref el aja uwo baamol /
bes inta ligo f seka bes / de mus kwayis? / tb ya mara / weled btak
suker winu
If you were a liar / you would have say true when the Court ask you / if
you had say I know hos words he is a bad man / but you speak the right /
the truth / you say you dont know what is he doing / you just meet him
on the way / isnt it fne? / Ok woman / where is your small boy?
(22) J. to D. delwokti maakama ligak bi galata galatatn / alla inta kizibta
/ inta gulta isma bitaki Deng / wa hini akwana btk igulu l de
ms isema btak / da tazwir / isem btak Maluk Aywen / de barau
jiza / jiza numro itnin / sarkta dura bta mara de / inta sarkta
dura li
"
ana inta gr isma bitak / lokan ita ma serG inta ma mum-
kin bitager isma btk /. . . [the story continues again and again..] /
fa keda /nina binsuf nas / daiman mujrimin ketr / nama gabod
bikr isem /. . ./ an keda kalam de nihna gul inta muznb ala-
t-tnin / fa kalam el-mahkama / inta biyemsi sita suur f sijin /
izakan inta ma jibtu talta u taltin gine bta mara de / w jiza
btk an inta xin l-aman / inta bmsoku talata suur / kulu juma
s-sijin btak tesa sar / izakan ma dafa grs / kan dafa grus inta
tla bara min sijin
Now the Court fnd you with two faults / today you lied / you said
your name is Deng / and your brothers say no its not your name / this
is falsifcation / your name is Maluk Aywen / this alone is a fault / the
fault number two / you stole the sorghum of this woman / you stole
the sorghum because you changed your name / if you did not steal you
would not change your name / [the story continues] and like this / we
see people / many criminals always / when they are taken they change
their names / because of this we say that you are culprit for two things
/ and the decision of the Court / you go six months in jail / if you dont
bring the 33 pounds of this woman / and your fault that you are a per-
jurer / you take three months / the total of the jail is nine months / if you
dont pay the money / if you pay you will got out from the jail.
626 catherine miller
5.2 Te Bay s case
/. . ./ indicates a cut.
/. . ./
(4) J2. smu mnu D1. John Gabriel
Your name?
(5) J2. abla ? D1. Bay
Tribe
(6) J2. sken wn D1. Hay is-sinema
Living where?
(7) J2. sgol btk D1. ma el/t leb
Your work? Im not working/
student
(8) J2. mrk D1. tnn u irn
Your age? Twenty two
(9) J2. mutm numro tnin smu mnu D2. welyam pitr
(10) J2. abla D2. bay
(11) J2. sgol btk D2. ma el/t leb
(12) J2. student / ah yr ed D2. omr/ sabaater sna
(13) J2. yur ed by ingliz/ eta mus asa tleb wela kf
Your age in English / arent you student or what?
(14) J1. ya askri inta bitalfa / ah an iglu l-hG / kut yidek na / tabn
mesya mu kda / gl wl al-azm / kitb el-mukaddes / agl el-hG
/ kl-el-haG / wel y gr el-hG / kda wor el-mdu li-l-mahkma
/ nafern da mal lk nu
soldier you take an oath / ah in order to say the truth / put your hand
here / of course Christian isnt it ? / say Lord the Great / Sacred book
/ I say the truth/ all the truth / nothing else than the truth / so tell the
matter to the Court / what did these two-persons make to you?
(15) P. wellay ombreh hawli saa saba f-l-misa ana jay bi tark bta-l-m-
drasa tijriyya de / ana jt / f askri mi giddmi lbes rasmi /
baadin el-awld bard kaman f giddmi / ana jt talta bi jambum
/ el-be henk gl le ya axi salm m f / el bi-jay bard u kaman gl le
salm m f / ana saktta / aslu m gutta hja / ana mya / baadin
ana gumta zitta katwa btay / hum bard zdu katwa waray / el-be
henk ja garasn fs ulbi / badin el-bi jay ja daggan
By God, yesterday / around seven oclock in the evening I was coming
by the road of the Commercial School / I came / Tere is a soldier walk-
ing before me, wearing his ofcial clothes / then the boys also are before
me / I came near them / this one told me of my brother, no greetings? /
the other also said no greetings? / I kept silent / I did not say any thing /
language use in juba local courts 627
I was walking / then I accelerated my walk / they also accelerated their
walk / this one started to pinch my buttocks / and then the other started
to beat me . . .
(16) J1. b-l-yedd wela bi kur
With the hand or with foot?
(17) P. bi-yddu garasn f sulb bi-ydd / baadin el-bi-jay ja daggan
keda bi-ketfa / ana gumta wagfu / gulta mlkum ya axwni / gl le
mlkum kf / ma titfahim ma
'
an / ana gult nitfahim ma
'
akum
agl makum en / gl le inta askut / inta m ermta u bass / ana
gulta lem ukrn / ana met / el-bi jay da tawli masakn ja garasni
/gulta le inta mlek asa / gl li axlaK xamak asa ana beduggk
/. . ./ tawli gumta nadt askri / gutta ya askari taal haj f el-ja-
maa del al-ajamn min / tawli hu / waled de ja gam darabni
bunya / askari jire ja / ja gl intum mlkum ya awld /. . ./ gulta hum
aj
y
omni / an keda ana korokta lek / askri da gl xls xls de
mawdua bast /
"
arahkum / woled de lisa der amol maaya mukel
/. . ./ sukna maan l-merkez tawli /. . ./ askari da / f arabiya tli
keda / askri tawli wgef arabiya / arabiya aba yagif / mai wgef
musfa keda /. . ./ lamm- jere mi li-l-arabiya henka / kullum
itnn tawli jmu foK / de gm bdrd fK / gaadn nesru / baa-
din de gm bdrbni bi wara kamn /. . ./ sa
'
fekka min yddi / ma
ref kn wa ya lu / wela kan wg wn / na ma rfu /. . ./ ya
gumna jna lahadd-l-nkta / sbbo le / inta en / nahl d nek wa
inta ermta u bass / inta mi riya skeT /. . . . /
With his hands he pinched my buttocks with his hand / then this one
slapped me / I stopped / I said whats the matter my brother / he said
what the matter how? / let you understand us ! / I said I understand you
I tell you what? / He said to me you shut up / you are just a pimp / I told
him thank you / I went / this one sized me directly and pinched me / I
told him whats the matter / he said close your mouth now I will beat you
/. . ./ I immediately called the soldier / I said soldier come and see this
people who are attacking me / immediately he / the boy started beating
me badly / the soldier run /asked what the matter boys / I said they
attacked me / this is why I shout afer you / the soldier said ok ok this
is a simple matter / lets go ! / the boy still wanted to make me problems
/. . ./ we brought them to the post directly / / the soldier / there was a
car coming like this / the soldier stopped the car immediately / the car
refused to stop / it went stopping at some distance / when he run afer
the car there / both of them immediately attacked me again / this one
beated me again / we were fghting / then this one hit me from behind /
my watch came of from my arm / I dont know if he took it / or if it fall
628 catherine miller
somewhere / I dont know /. . ./ so we came up to the police-station / they
insulted me / what are you / Curse on your religion you are just a pimp
/ you go in the street for nothing . . .
/. . ./
(24) J1. intum samt kalm da D1. samtu
Did you hear these words? I heard
(25) J1. el-bint de gl iya maya w intakum ta-azmtlu skeT / gl dik ya
ermta / u de gerres fslbu hene / w-inta bardu daggest bi ketf
/ u badin gl mlkum / intakum gul inti ermuta skeT / inti jya
min wn / f tarka gmo daribtu bi daraba adid / wa gam korrk
l-askri ja / w bardu baad el-asker / kamn intu g
w
um darbtu
baad / asn askri mi an ijib / yemsi kelem lel-l-sowwaK el-kan
waggofu / fat giddm / enta azemtolu darobtu li-raet ma saa btao
fakkat waga / wa wahed halaK kamn / da amn / entu gabelt f
ska / inta /. . ./
Te girl said she was walking and you attacked her for nothing / this
one said pimp / and that one pinched her buttock here / and you also
slapped her / and then she said whats the matter / and you said you are
just a pimp / from where are you coming / on the way you start to beat
her badly / and she cried the soldier came / and also afer the soldier /
you start to beat each other / during the time the soldier went to bring /
went to talk to the driver who stopped / he went in front / you attacked
her and beated her until her watch came of / and a hear-ring also / this
is sure / you met in the way / you /. . . ./
(28) D1. el-bit-de / ana kunt jay min hay kmryal li-l-bT / fi wad gidmi
/ ana ma baarf gbel / ma bit de wara / towones m bit de k /
umon gam fi akla / an na zeyde / ana askate umon /
Te girl / I was coming from the Commercial District / there is one
in front of me / I dont know him before / with the girl behind / he dis-
cusses with the girl / they start quarreling / because Im like this / I make
them silent.
(29) J1. da awli s kem
This about what time?
(30) D1. awli s sbaa / fi jay bls / fi jay / bit krk le bls / fialn el-blis
j / woddit len mukla klu / ana mi le-beT / yalla ez-zl el-
daraba-l-bit de / ma blis gaadn isugg zl de li-nukta / uman
ft giddm / ana ja wara / nama ana wosol mbil / uman lisa / jl
el-dagga bit u blis / ana ja ligm lsa fi mukla / dl fi mukla unk
/ min kede nama blis ainu na / el-bit gl aywa na gibel hink / ana
language use in juba local courts 629
gl filn ana gibel hink / akala hsel ana f / ya al-amal min ene
ana ma rif / blis masako jl de u jl de ft / ya ja bas masakan
whed be-rejlan /. . ./
Around seven oclock / a policeman came / the girl called afer the police
/ of course the police came / and brought us the all problem / I was going
home / Ten the person who hit the girl / he and the police they were
driving this person to the police-station / they went in front / I was com-
ing behind / when I arrived to the Mobil station / they were still, the one
who beat the girl and the police / I found them still quarrelling / from
this when the police saw me the girl said yes I was there before / I said
of course I was there before / the quarrel happened I was there / then I
dont know what happened / the police sized the person and the person
went / then one took me by the legs /. . ./
/. . ./
(34) D1. away / badaln talta / na / blis u bt da / sillna fi nkta / amdulla
na lig waed min az ma mbil enk / baaref ns bT / na gl
kwayis / kan keda ami klem ns bt / gl na fi nkta / il-ana fi
nkta / minnak axuy de / nama nsma kede / r fi nkta enk / nama
anu axuya ja / gl ya tni f / kutt tawli jowa / u badin sab / bit
de gam j len / gl saa woddur u en / u filan ana ma uftu saa
klu klu /
Yes / we stayed three / me, the police and the girl / they took us to the
police station / Amdullah I found one I wanted in the Mobil station /
he knows the house / I said good / if its so go and tell the people of the
house that I am in the police-station / they took me in the station / from
there my brother / when he heard like this / he went to the post / when
he saw my brother coming / he said the other came / they put him inside
immediately / and then in the morning / the girl came and told us that
her watch was lost and what / and of course I did not see her watch at
all
(35) J1. tyib / lama kn wled de /el-biyamml makil maa bt de / inta
ma mait / inta ajiju / inta kan barak wela ma wald da /
Good / when there was this boy / who was doing the trouble with the
girl / you did not go / you support / you were alone or with this boy?
(36) D1. na bara
I was alone
630 catherine miller
(37) J1. inta bara / tyeb badin hinay de / lama blis gabdu weled dk
mi bi / u gamaa kal da / inta ajijtu maa el-bint-da / blis
lamma j / hl giddm el-blis eta amoltu makil maa bint da
You were alone / good and then / when the police captured that boy
and went with him / and the group quarrel / you supported the girl / the
police when he came / did you in front of the police make quarrel with
this girl?
(38) D1. ana ma a
w
mol makel / bass darab bitman na ja katal
I did not make quarrels / I stopped their fght
/. . ./
(45) D2. el-zl de ombare ma biill nihna / ana gaad fi-l-bet / ya rasel wdd
da min el-mbl / wodd de maa unk / gl ya ns el-bet John gabad
/ ya ma kan nemi yisf li-l nkta
Te person of yesterday did not take us / I was at home / and he sent
the boy from the Mobil station / the boy went there / he says people of
the house John has been captured/ so I had to go and see at the police
station.
/. . ./
(50) J1. nta ma fs / u gabodk mittn
You were not present / where did they capture you?
(51) D2. gabodni zamn gale nemi-l-nukta niuf / min wisolta unak an
nera mukila bta djn kif / bass waed askri gl ya lzem bikn
wahed minhum / bas ja fi hene / tni ma barja bt / u fialan jabni
fi-l-arsa / ya sab bit de maa unk / djn gal ya bit bta ombare
enay jabni fi arsa unk / /
Tey captured me when I went to the police station to see / when I
arrived there in order to assess the problem of John / one soldier said he
must be one of them / he came here / I didnt come back to the house /
and of course they took me in jail / and in the morning the girl came /
John said this is the girl from yesterday who brought me in jail /. . ./
(52) J1. ya bit / asa nafarn de badi sumit kalmu / da numro wahed de
nker / u gl woket dak kalam inti gul ma hasal klu-klu / gl uwa
mi liga inta kunta akal maa wahed tni / w uwo mi aan iyaziz-
kum / iyazizkum bas inta maa zl dak / el-wokT inta ndit askri
ja / w-el-askri gam sgo / sG el-waled maaki / inta mi u bardu
waratum kaman / kddam giddm henk intum wageftu kaman bit-
kelem kalmt keda / lama u masa wosolu lekum henk / bls gl xls kan
izan keda ta
'
ali inta kaman maa el-jamaa del / gam xala dk u alu uwa
language use in juba local courts 631
girl / these two persons afer I heard their words / number one this
one denied / and said at that time, the story you tell did not happen at
all / he was going and met you quarreling with somebody else / and he
came to support you / support you against this person / when you called
the soldier came / and the soldier took the boy with you / you go and he
is behing you also / in front in front there you stopped you talked / when
he arrived near you / the police say ok if it is like this come you also with
the people / he let that one and sized him
/. . ./
(66) J2. /. . ./ keda ita indu rtn / musu barau rtan bta bay keda wonos
Do you have a language / isnt it a specific Bay language which is
spoken?
(67) D1. ana borton ndogo lakin ma f rtan bay
I speak Ndogo but there is no language Bay
(68) J2. kulu ndgo intum trtn ndgo
You all speak Ndogo?
(69) P. ana borton ay / norton kalam enu lakin
Yes I speak / but what shall I say?
(70) J2. de bigl uman beroton enay de rutn / de bigul l / ma boroton ndgo
/ beroton bay / nina azin rfu uman sei bay aw /. . ./ :
Tis one says that they speak their tribal language / this one says no /
he doesnt speak Ndogo / he speaks Bay / we want to know if there are
really Bay!
(71) J1. tayyb / izakan bay / el-fakka bikun f enu / lokn ma bay bikun f
enu / nesma el-jarma bass
Ok / if Bay / the diference would be in what? / If not Bay what would
happen?/ we listen to the crime only!
(72) J1. hata kan abla tni bara ma tam / jol el-arfu jol amel ma ajam /
biwori gul ya de / kan gabil wela ma bay biwori / bas el muhim inta
jeb el-blis el-kn ahall al-mawdu /. . ./
Even if it is another tribe it doesnt matter / the person who knows the
person who attacked him / he shows that it is him / if from the Bay or
not he shows / the important you bring the policeman who solved the
matter!
/. . ./
632 catherine miller
(74) J1. asma / al-sakya da ixtarrat gl inta shid / lama kan hsel beinakum
maa nafarn del maskal / eta keda gul s el eta bitaarefi bi zabti / eta
jt kif / eta legt maamal enu keda / gul len / le-l-mahkma
Listen the lady (?) decided that you are the witness / when the prob-
lems happened between you and this two persons / say what you know
exactly / how did you come/ what did you fnd? / tell us / the Court!
(75) W. taban siytu / ana kunta saal fi kart / fi-l-jawazt / mal el-jensi-
yat / badin ana sken fi muluk / lama ana jit ja wosolu fi medresa
Komeriyl / ana ft / taban al-bit de askri ana ma baaref / lbis
maleki / ana ma biyaref askari aw ma askri / faaln ftt / giddm
swiya / baad-ma fatt / al-bit de gum bikor bi-isma-l-blis / gal ya
blis ya blis tal agaod / fialan f gann btatna / bta al-blis / ay jol
bikrek bi-isim el-bls ma mafrd tafgo / lajem ita-ajri suf fi senu fi
senu / tekusu / fialan ana jay / ana jay legtum del itnn klu biajem
el-bit de / baadin ana jt / ana gul ya jamaa mlkum fi sen / gal
el-bit de sottemuni u ft keda / ana gulta tayib ya axwna matakum
sakal bit fi tarka / bi tarika zede / keda tosf el-jisim btna w-el jisim
bitkum wa jisim el-bit / taajem el-bit fi tarka bi-l-ll zede m sh /
wa lisa bikroko / lsa bikroko / fa ana asala-l-bit / hasala enu maak
/ gul nas del ya fat keda / ns del ajemu fi tarik / agru sulbu bitao /
u nama kelem umon setemu / fialan uman setemu /. . ./ bad-ma ana
rajo enka / ana raja / ana ja legitum lisa bisakal ala bit / nama ana
raja min mahal taban ana ft keda / baad-ma ana asma korokoro
waray b-ism el-blis / ana jire tawli ja wara w ana ja legitum l-itnin
del kullu / uman yau / fa filan kelemtum liw / ana gul ya akwnna
nina janbiyin ma mafrd neskel maa badna / hajt zede ma kwayis
/ el-muskila de best tkelwokT // el bit de je gum kelem gl ya askri
il tawasolna fi nukta //
Of course Sir / I was working in the cards / in the passports / the place
of nationalities / then I am living in Muluk / when I arrived at the Com-
mercial school / I passed / of course I dont know that the girl is a soldier
/ she wore normal clothes / I dont know if she is a soldier or not / of
course I passed / a little bit in front / afer I passed / the girl started to call
by the name of the police / she said police police come in / of course
in our laws / of the police / anyone who call afer the police you cant
let him / you must run see what is happening / you search / of course I
came / I came and found them both attacking the girl / I came / I said
people whats the matter / the girl said they insulted me and went / I said,
good my brothers dont quarrel a girl in this way / by this way / you see
my body and your body and the body of the girl / you attack the girl by
night like this its not correct / and they still shout / they still shout / and
language use in juba local courts 633
I asked the girl / what happened with you / she said this people passed
like this / this people attacked her on the way / they pinched her buttock
/ and when she spoke they insulted her / of course they insulted her /. . ./
when I came back / I came back / I found that they still quarrel on the
girl / when I came back from the place I went there / afer I listened the
cries behind by the name of the police / I run directly back and found
them both / there they are / and I spoke to them / I told them brother
we are southerners no need to quarrel between us / things like this are
not good / the problem is small until now /. . ./ the girl spoke and said
soldier you must bring us to the police ofce /. . . ./
(75bis) W. ma f maal hukuma keda gerb fogo /. ./ fi arabiya gaad j /. . ./ ana
gum jire bi-wara arabiya de bi-ztu /. . ./ min ana jire henak / ana
asma baga el-bit de biga bikore / uman tni keda bidugu el-bit bi-
wara /. . ./ filan ana jere kede kede kede / ana ja gabodtu da / xalas
amdullah / fi fi mujra waga fi jua mujra /. . ./ negedem kede kede
kede likaat nina wosolu mobli /. . ./ d tawli gum amol musakl
fo mbili li-raat-ma aoz kaser mrat bta arabiya zatu / aoz amol
muskil fi jua arabiya ztu / baadu sd el-arabiya uf kede m f /
da gum maragu bara / ana bardu gum maragu /.../ tni arabiya
ja / rakebna fog le-raat wosoluna fi nokta / baad-ma wosoluna
fi nokta / ana gul jol da jere kals lakin axu da biyaref mahal de /
d nker gul m biyaref da / d maal btao barao u d barao / u ma
biyaref klu klu / ana gul kf / lajem de aku btao li
"
anu uman hader
al-muskel giddmi ana /. . ./ el-hamdulilay lam fatna balK /
mowdu negla askri de masi mustafa / ana biga ft fi-jawazt ya
neselem kelem kabr li jamaa fi taakir el ana akr lina zaman / ana
msi wodi kabar li-hum enay / lama fi roj
y
u bitay / el-hamdullay d
kamn gum wasal / ana gal ya de bs ya de bs / uman l-itnn el
ajam ala bit ya de bs /
Tere is no governmental place nearby /. . ./ a car came / I run afer the
car /. . ./ when I run there / I heard the girl shouting / they were again
beating her from behind /. . ./ of course I run like this like this . . ./ I
captured this one / ok thanks to God / there is a sewer canal / he fall in
the sewer /. . ./ we accompany like this like this until we arrived at the
Mobil station /. . ./ this one started to make problem in the station until
he was near to break the mirror of the car / he almost make problem
inside the car / then the owner of the car see its not possible / this one
went out / I also went out /. . ./ another car came / we went in until we
arrived at the police station / afer we arrived in the station / I said this
one run away but his brother must know the place / he denied he said
he doesnt know him / this one has his own place and this one also / and
634 catherine miller
he doesnt know him at all / I said how? / It must be his brother because
they attended the troubles in front of me /. . ./ Tanks to God when we
opened the case / the matter of the transport of the soldier to the hospi-
tal / I went to the passports to bring the news to the people there about
my delay before / I was going to bring the news to them there / when I
was coming back / Praise God this one also arrived / I said this is him
this is him / the two who attacked the girl, this is him /. . ./
/ story continues . . . ./
(76) J2. Tayib intakum ref nefsu kalatnin
Good do you recognize that you are faulty?
(77) D1. la la la kalm ke ana wosol dak ana baray / kalam de kadb / ana ma
jay maa jol / ana jay baray
No no this words I arrived there I was alone / these words are lie / I
didnt come with somebody / I came alone
(78) J2. ya de suhud kelem kalam bta / keda wonosu ma suhud / asalo /
indk sual yeslo
But the witness spoke / so discuss with the witness / ask / do you have
a question for him?
[Discussion between the defendant and the soldier /. . ./]
(79) J1. intum kalatanin / mahkama gul
"
annu intu aemtu bit da fi-t-terk
bidun ay munasaba / u atemtu u gul ermuta / u into garestu-ha
li-raet-ma sa
"
a bitwdda / tamenu xemsa u arbain gine / w indu
wahed fardeK halla wodur / tomon sabaa gine u nus / de klu bi
kuss maskel btakm maao fi-t-teriK /. . . continue /. . ./
You are faulty / the Court says that you attacked this girl on the way
without any reason / and you insulted her and said pimp / and you pinch
her until her watch got lost / its price was 45 pounds / and she had ear-
rings which got lost / their price seven pounds and half / all because of
your quarrel with her on the way/. . . ./
5.3 Kator B Court 1984, Record 3/ 1984
Judge Stephen to the public
(1) J. . . ./ numro tnin koreraK / izakan nasi bikore bi-ataku ktir ma bikli
makama ymsi / izakan nas biasma kalam de bmsi bi-r / u gdiyat
bemsi bi sra / gowm / mumkin nihina baamolu isrin gediya pi yom
/ izakan ma pi pogu ijaat / ajaat bi sikil de /. . ./ arjuk sabab wa intum
language use in juba local courts 635
nas mutaimin mutamain / el ja asan asma makama / intum ma bija
ena sakiT / eta jet asan testepiT f makama / asan kul wa baadin
bukura bisala kalam pi bet / aw mara kaman banaat kan ja asma
asan ma baamalu kalaT pi biyutum / sukran / istamir gediya tni /
Number one, shoutings / if people shout laugh a lot it does not let the
Court going / if people listen the words go quietly / and the case goes
quickly / fast / we can do twenty cases in one day / if there is no troubles /
things like this /. . ./ I praise young people and you the defendants / those
who came to listen the Court / you didnt come here for nothing / you
came to beneft in the Court / because each one later tomorrow solve
his problem at home / or women, also girls they come to listen in order
not to make mistake in their house / thank you / continue next case!
(2) S. Taban Sabastian ../ Alexi Lado
S. to the J. lisa ma ja
Not yet arrived!
to a man nearby) kede belay atla awal haja / ihina f gdiya / wa mata
lakbat / ita yau ga-asma kalam ini / ita ja bikelem / istena lakad bad
gdiya / msi
Please go away from here frst thing / we are in a Case / dont interfere
/ you have heard the words here / you will speak / wait until afer the
case / go!
(3) J. to policeman kli jol de yeji hina / korbaaT btak win ? /
Lets this person going here / where is your whip?
(4) Police yau le Wani Its with Wany
(5) J. ilu min Wany / u de jatu barfu urkali ? / masi tow gene-gene bita enay /
Take it from Wani / and this one does he know the Courts police? /
go and sit there!
(6) J. mublek minu / Mogga lo munu D. ana
Who is the Plantif? / Whos Mogga? I
(7) J. a isim bitak D. Gabriel Morbe
Ah, your name?
(8) J. Barinit wen D. lobunuk
Bari from where?
(9) J. u eta majlum knyo u knyo.
From what are you victim?
(10) P. de taban kan pi ukti keda / kan juju ukti f jaman hawadis / badin ana
ruwa nadi ml / / gal mal maap / taban jaman de hawadis / taban ana
sufu zede gal kan uwo ro ptisu mal / mumkin boro ligo harib / taban pi
jaman de pi aja de /
Of course there was my sister / he married my sister during the civil
war / then I went to ask for the dowry / he said there is no dowry / of
course it was during the civil war / I realized that if he was going to look
636 catherine miller
for the dowry he would maybe fnd the war / of course at that time there
were such things
(11) J. hawdis pi sanatu km (9) P. tamanya u sitin
Troubles of which years? Sixty eight
(12) J. badin ada ta awadis mitin
Ten the end of the troubles when?
(13) P. tnen u sabain / badin kida pi itnin sebeyin / taban ina kaman nadi mal
/ gal mal maap / ila nina amul aja de / dgit mrisa / anina aju wodi
le uwo / gali kasara ita jibu / ana ma indu haja tani / badin uwo amulu
karama / badin dgig mrisa de nina wodi le aja de / le uwo / ma amul
lena haja tani / badi swiya yaba ja mat /anina amulu karama ta yaba
wkit maat / wa asa ukti de / ukti de mat / wokan ma wdi le ana ml
/ bes teletin jine eli kan wdi le nina
Seventy-two / then in seventy-two / we asked again for the dowry / he
said that there is no dowry / only we do this thing / beer four / we want
to bring it to him / he said you bring it for nothing / I have nothing else /
then he made a ceremony / then this beer-four we brought it to him / he
did not do anything else for us / later on our father died / we made a cer-
emony for our father when he died / and now my sister / my sister died /
and he did not give us the dowry / only 30 pounds that he gave us.
(14) J. mt mitin When did (s)he died?
(15) P. mat aja de f tnin u tamanin / saar idaser
(S)he died in seventy-two / in November
(16) J. indu iyal Does (s)he have children?
P. indi iyl He has children
(17) J. kam How many?
(18) P. f tegriban arba / waiT kan ge dowru ma rokuba / uwo ma wdi baal
pogo / ja waga f nr / gam ja mat
Approximately four / one was walking on his knees / he did not take
care / he fall in the fre / he died
(19) J. pi talta P. pi talta ay
Tere is three? Yes there is three.
(20) J. talta de f bant fogo P. f wae bit u badin tnin yl / tnin woled
Among the three Tere is a girl and two
there are girls? boys/two boy.
(21) J. eli mutu de woled Te one who died was a boy?
(22) P. la / bineya / wa asa itnin iyal de uwo ma biwdi baal fogo / asa itnin
de ge geni le ana
No / a girl / and now the two boys he does not take care of them / now
these two are staying with me.
language use in juba local courts 637
(23) J. ge gene le eta / kalam kwes / aah /
Tey stay with you, good, aah?
(24) P. wa min de taban ina julumin / bineya tanina ma mijawaz / bes ina
akudu teletin jine fakat / u bineya kaman maat / iyal kaman ana lisa
yau ge akilu
And from there of course we have been spoliated / our daughter is
not married / we just took 30 pounds only / and the girl died / and the
children I am still feeding them.
(25) J. itakum yau ge raba
You (pl.) are the ones who raise them
(26) P. nina yau ge raba / yau ana majlum ma kalam de
We are the one who raises them / I am spoliated by this story
(27) J. ita masi leo pi bet / le bet btomon gale malu ita keli iyal de kede . . .
Did you went to him in the house / to their house to ask why you lef
the children like this
(28) P. walay insala ana masi aslu ma bijibu kabar
By god, even if I go he will never inform me
(29) J. ma ja kurju
He did not cultivate (for you)?
(30) P. bikurju lakin oslu ma bowodi le iyal de
He cultivates, but he never gives to the children
/. . . . . ./ Interruption with another case
(31) a woman : nan kan a tu i diit na kora na torobo na a ilu utu ti mrri
mrri yi ko wate
I went at the time of the distribution of the furniture and then he
fought with me and his wifes
(32) J. ina kju kdyo kiyang baligga / kede kiyang sina / maybe nan kan a
mistake / ado yie gwodam nyo Wani ? ma tini wuni ilo korobat / ti
ko yege o kunen ko doggu kito kadi / bonggwat ban le /
She had reported the case earlier / wait frst of all like this / I might have
been mistaken / why do you stand still Wani ? you go receive that whip /
let him carry these, carry these furniture / are there no clothes?
(33) Woman: bonggwat ma a kekeren Clothes! He torn them
(34) J. Santino makme ko Santino there it is
(35) D. Walayi / a hal diri bijab i yemba nio nagwon nan yembi kiyasir nanyit
na i diit na hawadis / a de i inu diit ni kan ukum a itinasar jine
badin kayayu a saba jine / badin kjulu anan medde sona anan di min
taab kulu anan kan lunggi luaser baba uba a ko doggi male kwe kunu
uba nyu / kirut ni a kine mali kirut taban jur kaman ktir bayn a nan
/ kalas utu adi kalas lo utu taban bubul yemba makune mali lep-
eng aje dippa salet / loutu gwe a komonit ti poki ibang / nyena diri i
gwe i dida ko ina uro / taban ko yaba ko atu a nan bubul tindu o
638 catherine miller
nagwon kata / wu hal kajelu ni a baba lo twane / badin ne a monye lo
twane / kirut adi monye sarji adi ukum nagwon kju monye a ukumbe
adi asa man dek pitn kwe ti gurut kune ko na utu nan dek sitin
jine /
By God / really during the time of my marriage when I married his sis-
ter during the civil war / by then the Judgement here was twelve pounds
and ours was seven pounds / At that time, on seeing this, and in order to
avoid these troubles, I decided to call my fathers brother from Juba so
that he could go to collect my dowry from Juba there / Ten from there
this dowry, of course the village is large, not me alone / ok, people said
ok, this man can marry, here is the dowry, he has prepared a cooking
place / He has become an in-law, let him be in the house / at this point
I started to stay with this woman / of course, if her father comes, I can
give him what ever possible / and at that time my father died / then
her father died too / then her father (i.e. an uncle) sent a message that
the judgement passed before was that now I want the remainder of my
money of this woman I want sixty pound /. . . . ./
[Te case continues in Bari and JA with many diferent participants. At
the end the Defendant and his family agree to pay the remaining of the
dowry]
Conclusion from Judge Stephen :
(37) S. dgiga / dgiga / ah bagi mal el kede eta bi silu lau pi bet le nsibat inak
/ ah / u wdium kabar gal nina jain / asan bisilu iyl de kli ruwa tusu
bujak / uman kaman biptisu kruP kwes kida / u jama biji yesrub / wa
nas umu yani makasutin / nas klu / kulu jene beji intum bestenu / jebu
gdiya tani
Minute / minute / ah the rest of the dowry you take it to them at home
to your in-law there / and give them news that we are coming / so that
they take the children for spitting saliva (benediction) / they also look
for a good sheep like this / and the people come to drink / and the moth-
ers people are happy / the maternal uncles people / every child you wait
for (??) / bring another case!.
PARADIGMATIC STABILITY AND FINAL LARYNGEALS IN
NIGERIAN ARABIC , OR WHY HISTORY REPEATS ITSELF,
WITHOUT ACTUALLY DOING SO
Jonathan Owens
CASL, University of Maryland
1. Introduction
Arabic presents a rich, if underutilized, domain for examining processes
of linguistic change. At one extreme are the challenges presented by the
rapid and dramatic restructuring witnessed in the Creole Arabic variet-
ies of the southern Sudan and East Africa (Versteegh 1984, 2004). At the
other are the mundane, day-to-day, decade to decade, century to cen-
tury processes which move slowly through the diferent varieties, some-
times leading to marked changes, sometimes not. In this paper I would
like to document one such change, which has happened not once, but a
number of times in the history of Arabic , and relate the phenomenon to
a more global interpretation of Arabic language history .
At issue is the conceptualization of Arabic language history . Te
reigning paradigm today, indeed one established some 150 years ago,
is that a classical language , or Old Arabic , by various processes of
simplifcation passed into the modern dialects , or Neo Arabic . One of a
number of problems with this model is that it leads one to collapse what
are ofen internally diferentiated developments into a common mold,
which are then assumed to substantiate the Old Arabic /Neo Arabic
dichotomy . Why this militates against a properly nuanced reading of
Arabic language history is that what may lie behind linguistic changes
are a combination of diverse local developments and fundamental
linguistic principles. Tere have indeed been many changes in the many
varieties of Arabic spoken throughout Asia , Africa and in former times,
Spain . Te fact of change in one variety, however, does not justify an
historical model based on the Old/New split.
Tis point will be illustrated on the basis of verbs ending in a glottal
stop , or in some cases, a fnal voiceless glottal fricative /h/. Classical
Arabic , of course, has a glottal stop, whereas nearly all modern dialects
do not.
640 jonathan owens
Tus, in Classical Arabic forms such as bada begin and ja come
appear in most dialects as bada and j. For purposes of this paper I
will assume that there has been a change of , whereby verbs such
as bada fell together with verbs like ban he built.
1
In contemporary
dialects these will typically belong to the same class of weak-fnal verbs ,
as the following sampling indicates:
(1) Algerian bd-t, bn-t I began, I built, Nigerian bad-t/ban-t,
Eastern Libyan bid-t/bin-t, Mardin (Turkey ) baday-t/banay-t, etc.
2
A related development is attested in contemporary Western Sudanic
Arabic comprising NE Nigeria , northern Cameroon , Chad and the
western part of the Sudan (see Owens 2006, chapter 5). In the rest of
the paper I will describe the development in the western Sudanic area
on the basis of a relatively detailed corpus-based analysis of Arabic from
Maiduguri in NE Nigeria, and then address the question of its signif-
cance for the history of Arabic.
2. Nigerian Arabic
3
Nigerian Arabic is unremarkable vis vis other Arabic dialects in its
verbal structure. Like all other varieties of Arabic it has two basic verb
forms , perfect and imperfect . Te perfect stem is marked by sufxes
indicating person, number and gender, in the imperfect these being
indicated by sufxes and prefxes . Both perfect and imperfect stems
have two conjugations based on whether the stem vowel is high or low.
In (2) strong verbs are illustrated, with the verbs katab write and libis
1
Te proviso for purposes . . . leaves open the possibility that the modern glottal
(stop) -less dialects go back to original glottal-less dialects in Old Arabic (see Rabin
1951). Certainly the glottal-stop-less varieties go back to pre-diasporic Arabic , as
attested in their widespread distribution across the Arabic -speaking world, and their
attestation in Old Arabic sources. If there are proto-forms of Arabic without the glottal
stop , the thrust of the paper would need to be reorientated. A few Yemeni dialects have
a glottal stop (Behnstedt 1985, 43).
Even in classical times, there was a variety without the glottal stop that a glottal stop-
less variant which was prominent enough that in the Koranic reading tradition (qirt)
recitations are fully allowed without the phoneme (Ibn Mujahid ).
2
Te value of the sufx , -t, -t, -ayt is an issue independent of the status of the glot-
tal stop .
3
Research support for this work was provided by the German Research Council
(DFG).
paradigmatic stability and final laryngeals 641
wear, which belong to opposite conjugations in both the perfect and
imperfect:
(2) Low stem vowels: perfect /a/ katab; imperfect /a/ bi-lbas he dresses
High stem vowels: perfect /i/ libis; imperfect /u/ bu-ktub he writes
4
By and large the assignment of verb stems in imperfect and perfect to
the two classes is lexically determined, and as seen in the example, one
and the same verb stem can, indeed as a rule is, assigned to a diferent
class in the perfect and imperfect tenses.
Final weak verbs in Nigerian Arabic also follow the same classifcatory
schema, as can be seen in the following examples in (3):
(3) Low stem vowels : bana he built, bi-lga he fnds
High stem vowels : ligi he found, bi-bni he builds
2.1 Final laryngeals //, /h/ in Nigerian Arabic , some data
So far as the basic paradigmatic facts go, Nigerian Arabic is fundamen-
tally identical to other Arabic dialects , and broadly similar to classical
Arabic . One phonological change involving two sounds has, however,
complicated the distribution of lexical forms .
Te two sounds are h and . Historically speaking, NA h has two sources,
OA (or pre-diasporic ), /h/ as in ahar month < OA *ahar, and OA *h
as in hilim dream, < h ilm. NA // derives from //, as in irif he knew,
< OA arifa. Te change of OA /h / and // to /h/ and // is one common
to Arabic in Chad , Nigeria and parts of the western Sudan . It happens,
however, that /h/ and // themselves are weak sounds, and liable to
variation of diferent kinds. Te range of variants includes the following:
Tey may be kept: biarif he knows, ahar month, whid one and ahamar
red.
5
Tey may change to the semivowel /y/ next to an /i/: biyarif, wyid.
Lastly, they may be deleted altogether: biarif, shr, wid.
4
As in many dialects , there is no phonemic contrast between short high front and
back vowels. In this dialect , however, /i/ and /u/ must be lexically specified, as they are
unpredictable. As far as verbs go, given a lexical stem specification, the pre-formative
vowel is usually determined by vowel harmony rules: if the stem vowel is /u/, the prefix
vowel is /u/. Otherwise it is /i/.
5
Via the so-called gahawa-complex, whereby an /a/ is inserted in the sequence of
guttural C + C, in this case < *ahmar < ah mar.
642 jonathan owens
Te variation spoken of here is at the individual level, determined by
contextual and socio-linguistic factors (see Owens 2006, 242 for statisti-
cal treatment); one and the same speaker may in one place for month
give ahar, and in another r.
Te question I will investigate in this paper is what happens to
verbs like simi hear, gaa
6
cut and karah hate, verbs with fnal
/h . Tere are two reasons for concentrating on the laryngeals in this
position. First it is only in this position that a change in the laryngeals
is nearly categorical, as will be explained.
7
Secondly, it is here that the
phonological change has direct morphological consequences. If they
are deleted in fnal position, what efects, if any, the deletion has on the
morphological structure of the language need to be specifed. Given a
form such as tismao you.M.PL hear, if the fnal glottal stop is deleted,
the form tismao would automatically arise. Tisma-o, with fnal V-V is
not a paradigm otherwise attested in the dialect , however. Alternatively,
it could collapse with the already existent weak fnal paradigm , tilga you
get, tilg-o they get (see (3) above), which would yield tism-o. Which
alternative emerges is described in this section.
Te data for addressing this question is of two types. One is textual
data, which will be summarized in 2.2 below. First I present the results
of a test in which 8 Arabs from Maiduguri were presented orally with a
series of test sentences in which one word was lef out. Te respondents
had to complete the sentence with the correct form. All of the answers
involved verbs with a fnal // or /h/ , the purpose of the test being to see
in which contexts the fnal laryngeals were kept or deleted. A sample
question was the following, involving the verb simi hear, the frst
sentence requiring a perfect verb, the second an imperfect :
(4) hu simi an-ndim haw ana kula .
he heard the man and I also
hi tisma l-kalm haw hinna kula .
she hears the sentence/matter and they F also
6
The // is emphatic and implosive.
7
It is my impression that the deletion in this position is statistically more common
than in word initial or medial position, though I have not checked this.
paradigmatic stability and final laryngeals 643
Te most obvious response for the frst was heard him, the Arabic for this
being either sim-t-a heard-I-him if the fnal // is deleted, or simi-t-a
if it is kept, while for the second either b-isma-ann-a 3-hear-FPL-it
(if kept) or bism-ann-a (if deleted). Note that in the imperfect a fnal
V-initial sufx (-an FPL in the second example) replaces the fnal stem
vowel if the fnal laryngeal is deleted.
Te respondents were all under 30 and most are educated. It was found
that those older than 30 had trouble concentrating on the questions, so
unfortunately age diferential could not be taken into account.
A total of 16 sentences were asked. Te anticipated responses were
distributed among diferent morphological classes of verbs , both basic
and derived verbs for example, and among diferent infectional contexts,
though only subject sufxes were tested for.
Questions were asked for both perfect and imperfect verbs, as
illustrated in (4) above. Two types of sufxes can be distinguished here,
those that begin with a C (e.g. 1SG -t) and those that begin with a V
(e.g. FPL -an). In the 16 test sentences, the answers divided into forms
involving the following sufxes:
(5) Perfect
verbs with subject person sufx -t N = 5
8
3MPL -o N = 3
3FSG -at N = 1
Imperfect
3 MPL -u/o
9
N = 5 (3FPL -an, N = 2)
total N = 16
Tere were 128 total responses (16 u 8). In all but 2 cases the responses
conformed to the anticipated answers. Te two deviant answers are
ignored here, so in all there are 126 responses. Te results are summa-
rized in (6). In (7) three typical responses are given.
(6) Stems appearing with:
8
Tis stands for any subject sufx that begins with a -t, including for example 1 sg,
sim-t I heard, 2 FSG sim-ti you F heard and 2 MPL man-tu you MPL prevented.
9
In the imperfect the MPL sufx is -u afer a high stem vowel , bim-u they go from
the stem bimi, and -o afer a low stem vowel , bilg-o they fnd, from the stem bilga,.
644 jonathan owens
(6) Stems appearing with
(a) (b) (c)
/h = , stem vowel kept /h kept
(= weak verbs ) (mixed stems ) (= strong verbs )
Perfect
-t 36 0 2
-o 11 1 12
-at 2 0 6
Imperfect
-u/o 14 1 25
-an 8 8 8
Total 71 2 53
before C-initial sufxes :
36 0 2
before V-initial sufxes :
35 2 51
(7) (a) (b) (c)
/h = mixed stem /h kept
Perfect
-t ga-t I cut gaa-t
Cf. weak verb ma-t I went cf. strong verb katab-t
-o sim-o they heard daba-o they killed sim-o they heard
cf. ma-o they went
Imperfect
-u/o bism-o they hear bisma-o bisam-o
cf. bilg-o they get
Te verbs in column (a) behave like weak-fnal verbs , examples of which
are given in brackets in (7). Te verbs in column (c) behave like strong
verbs , verbs with 3 consonantal roots . Te mixed stem in column (b)
has attributes both of stems with deleted fnal laryngeal and of those
with the laryngeal maintained. Like the former the laryngeal is deleted;
like the latter the fnal stem vowel is maintained.
Tere is a fundamental contrast defned by the variable C- or V-initial
sufx .
Final /h disappear almost categorically before a C-initial subject sufx
(36 cases where /h = (column a), 2 where they are kept (column c)).
Te verbs are then conjugated like weak-fnal verbs , that is, verbs with a
CVCV stem , e.g. ga-t I cut < *gaa-t , cf. ma-t I walked. Otherwise
the maintenance of /h is slightly dominant statistically, though only in
one instance is the dominance overwhelming, namely in the form b-at
she sold with the 3 FSG sufx -at . Whether the conditioning factor here
paradigmatic stability and final laryngeals 645
is the sufx itself or the verb stem (CC-, which I suspect is the case) is
not apparent in the data since only one -at frame was used.
As with C-initial sufxes , usually before a V-initial sufx if /h are
not used, the stem is shifed to the weak-fnal class, e.g. ga-o they
cut, budb-o they M slaughter. In two cases, however the stem vowel
was kept: daba-o they m slaughtered, bitba-o they M follow. Also, in
three cases where /h are not used, stress was irregularly shifed to the
penultimate syllable , bukr-u they M hate.
None of the respondents categorically used or disposed of the fnal
/h , even in a discrete sub-class of forms (e.g. perfect stems , excepting
the C-initial conditioning factor). Table 1 summarizes the global scores
for individuals.
TABLE 1 INDIVIDUAL RESPONSES
Respondent 1 2 3 4 5 6 7 8
/h kept 4 2 9 2 9 13 9 4
/h dropped 12 14 7 14 6 3 7 11
For 6 of the respondents there is textual material against which their test
scores can be compared. Tese texts include standard interviews as well
as less formal situations. Of the six, one has no tokens of fnal /h in his
text, and one has only 1. Te total scores from the texts, classifed into
morphological context, are given in Table 2.
TABLE 2 TEXT SCORES OF SIX INDIVIDUALS
Perfect Imperfect AP Total
t/n -o at u/o an V,
/h 0 6 5 10 3 0 0 24
10 7 5 5 0 6 2 35
Te contexts are the following; in the perfect : sufx -t/n 1 SG, 2, 1PL,
MPL -o, FSG -at, in imperfect MPL -u/o, FPL -an, 3 MSG with object
sufx , AP with plural sufx . An example of each (taken from texts) is as
follows, where relevant giving one example with the fnal laryngeal kept,
one with it absent.
dabt I slaughtered, daboh-o they slaughtered, dab-a they M
slaughtered it F (these 2 tokens from same speaker); wagaat she fell;
simata she heard him, bism-o they M hear; tugo you M fall; bimba-an
646 jonathan owens
they F are sold; basem I hear him; tbn (AP, active participle )
following-MPL (<tbi). Note that the AP and object-sufx contexts
were not used in the test frames.
In two cases the sample in the textual material is not very representative.
Te FSG -at sufx is represented only on two lexemes , one speaker with
fve tokens of wagaat she fell, another with fve of sim-at she heard
(etc.). Also, the FPL -an occurs only in one speakers text.
In most respects the textual and test material agree: in texts h/ are
categorically dropped before the subject sufxes beginning with -t/n,
and they are similarly nearly categorically absent in the test. In both sets
of data the stems shif to the weak-fnal class. Similarly, in both there is a
greater degree of variability before the MPL sufx (-u/o) in both perfect
and imperfect verbs , the ratios of kept/dropped tokens being roughly
the same. In Table 3 the percentages are calculated by dividing the total
tokens of stems where /h are kept by the total of laryngeal stems with a
MPL sufx . Te token count is given before the percentage.
TABLE 3 TEXT COUNT: /H KEPT BEFORE MPL
SUFFIX /DROPPED, % KEPT
perfect imperfect
kept dropped % kept dropped %
test 11 12 48 14 25 64
text 7 6 55 10 5 66
Te texts also indicate that before object sufxes there is a categorical
shif to the weak-fnal paradigm , regardless of whether or not the sufx
begins with a V or C; thus daba-hin he slaughtered them F, budb he
slaughters it M, rather than dabah-hin/budbaha.
108
Tere is also a fair degree of agreement between the test scores of
individual speakers and their treatment of /h in natural speech. Speaker
4, for instance has the lowest percentage retention of /h in the test, and is
also lowest in the text count, while speaker 6 has the highest percentage
retention in both. Only speaker 7 has a lower percentage retention in the
text than in the test.
10
Afer strong verbs object sufxes are sufxed directly to the fnal -C, with no other
changes occurring, e.g. katab-a he wrote-it, katab-hin he wrote-them. Afer verbs end-
ing in a fnal -V the fnal vowel lengthens before a sufx , ligi-hin lig-hin he found
them. Before object sufxes (which were not tested in the frames like (4)) the laryngeal -
fnal verbs shif to the weak-fnal class.
paradigmatic stability and final laryngeals 647
TABLE 4 TOTAL RETENTION/DELETION OF FINAL /H
IN TEXTS OF 5 SPEAKERS
Speaker 2 4 6 7 8
/h kept 0 4 9 6 5
deleted 1 11 2 11 8
For the present sample of speakers it can then be said that verbs with
fnal /h belong to a mixed paradigm : before C-initial subject sufxes
they belong to the paradigm of weak-fnal verbs ; before V-initial subject
sufxes they are treated variably, in cases even by the same speaker,
sometimes as CVCVC stems , sometimes as CVCV, weak- fnal stems .
2.2 Larger sample
Looking at the larger sample of texts, as summarized in Owens (1998),
the extent of deletion in the pre-vocalic context was indicized over three
sample groups, Maiduguri interviews (N = 58), Maiduguri group con-
versations (N = 61) and village interviews (N = 52). Aspects of the cor-
pus can be briefy summarized here. In the following, an index of 0%
means that the laryngeal is never deleted, of 100% that it always is. Table
5 groups all speakers together, while Table 6 divides them according to
place and sex. Te following indices pertain only to the context V /h-V,
i.e. V-initial sufxes .
TABLE 5 TOTAL SAMPLE
Table 5a: Place Table 5b: Sex Table 5c: Age
Village 71% Male 79% <32 83%
Maiduguri 79% Female 71% 3249 83%
>49 61%
TABLE 6 PLACE X SEX
Male Female
Maid 81% 75%
Villages 74% 54%
More so than the data examined thus far, the overall scores point to a loss
of fnal laryngeals in pre-vocalic position , the loss being most pronounced
648 jonathan owens
among younger, male residents of Maiduguri , with Maiduguri well ahead
of the village overall among both males and females. Te text scores
show a higher degree of deletion overall than do the test scores described
above. Tis is interesting, if the test scores indicate a higher degree of
monitoring (the reading list efect in the classical, early sociolinguistic
studies), it would indicate that the laryngeals are somehow there to a
higher degree than the test scores would show. Te lower female scores
would deserve comment, if adequate comparative data were available,
which it is not. In western studies, females have consistently been shown
to lead changes. However, even if western sociolinguistics attempts to
draw far-reaching conclusions from such tendencies (Chambers 1995,
139), the present data cautions against overgeneralizing, to say the least.
Looking at the data overall, a change can be said to have occurred
in the apportionment of fnal verbal /h verbs : before C-initial sufxes
they merge with weak fnal verbs whereas before V-initial sufxes
they variably merge with weak-fnal verbs . Factors speak both for and
against the variable merger going to completion. In favor thereof, urban
dwellers, especially younger ones, have the highest degree of merger.
Rural dwellers, however, have a lower degree, and at this point in the
history of Arabic in Nigeria , they constitute the overwhelming majority.
Moreover, there are structural factors favoring retention: in explicit tests
the fnal laryngeal was retained more ofen than in spontaneous speech.
Moreover, in other positions, V-/h-V sequences, such as word-internally
(ahar month), are generally maintained, so at this point at least, the
high degree of variability is most marked at the right morphological edge.
Moving to the next section, the current situation will be represented as
having a split paradigm , with the possibility, that in the future laryngeal -
fnal verbs will merge completely with weak-fnal ones .
Before proceeding to the fnal section, it is relevant to note that
an identical treatment of laryngeal -fnal verbs is found in Abbeche
Arabic in eastern Chad (Roth-Laly 1979, 810). Tis shows that the
split of laryngeal -fnal verbs into two morpho-phonological classes is
old enough to be a pan-western Sudanic Arabic trait. Again, however,
without more detailed study of the situation in eastern Chad it is not
possible to say more than that. If the loss of laryngeals is more advanced
in Chad than in Nigeria , one would predict that they would also be lost
eventually in Nigeria. If, on the other hand, the same complementary
treatment of fnal laryngeals is found as in Nigeria (governed by form of
sufx ), one would rather see the split as stabilized.
paradigmatic stability and final laryngeals 649
3. Morphological Regularity, paradigmatic stability
Before continuing, it is relevant to examine the manner in which the
laryngeal -fnal verbs have split into two classes. As seen in (7), the laryn-
geal-fnal verbs either end in a laryngeal (column c in (7)) or in a vowel
(column a). Tey thus either remain in the class of C- fnal verbs or they
shif to the class of weak-fnal verbs :
(8) strong verb (C-fnal) weak verb (V-fnal)
gaaan they F cut gati you FSG cut
cf. katab-an they F write cf. ma-ti you FSG went
What happens only rarely is that a new morphological class is created,
one with properties of both the strong and weak verbs . In (6) there are
two such tokens, attesting to the fact that such forms do in fact occur.
11
Tey are given in column b; both tokens occur before V-initial sufxes .
A form like daba-o they M killed derives from dabah-o with deletion
of the fnal laryngeal . Rather than switch to another stem class, which
is what usually happens when the fnal laryngeal is deleted (dab-o), the
fnal stem vowel has been maintained. In essence, a new morphological
class has been created. Tis class, however, has not become widespread
or established in the speech community.
Looking at the development structurally, the deletion of the fnal
laryngeal leads to the splitting of /h-fnal verb stems into two pre-
existing morphological classes, the distribution of these being broadly
defned by morpho-phonological context. What did not happen is that
the loss of the fnal laryngeal in one context led to the development of a
new morphological class based on the stem CVCV-sufx . Morphological
stability takes precedence over forms, which would be derived by the
logic of phonological rules.
Te WSA developments would ostensibly appear to support the idea
of linguistic history repeating itself: fnal laryngeals in Arabic verbs
tend to drif into weak-fnal ones . Tis repeats a similar process, which
occurred in pre-diasporic Arabic , as illustrated in (1) in section one
(bada-t/bad-t). Under one interpretation of Arabic language history ,
one could this add drif to the catalogue of features characteristic of New
Arabic (allowing for the cautionary note 1 above).
11
In the texts from speakers from the present sample no mixed forms occur. In other
texts, however, they are attested.
650 jonathan owens
However, the data allows for a more nuanced description. Te change
described for WSA can be broken down into two parts. On the one hand,
what may be termed a principle of paradigmatic stability, one aspect of
morphological stability, can be invoked:
(9) Principle of paradigmatic stability: do not create new paradigms
In its categoricality this statement is clearly too strong, and one can imag-
ine adding many conditions to it, but it serves present purposes. Both
the Old Arabic and the WSA data obey the principle: the loss of a fnal
laryngeal in both cases did not lead to the creation of a new paradigm .
Rather, the laryngeal -less forms simply collapsed, or in the case of WSA ,
are still in the process of collapsing, into already-existing paradigms .
Te other part is the phonological change that creates the condition
for the collapse into pre-existing paradigms . Tis change is one of
happenstance. In WSA , /h / became /h/ and // became //. Tis type of
change is not unique among varieties of Arabic . // has moved to // in the
Tihama , /q/ appears as // in Cairene , Damascene , as well as elsewhere.
Maltese presents a complicated picture of its own. and // merged in
//, which in turn was lost, leading, as in WSA , to the merger of *//-fnal
verbs with weak fnals.
129
(10) sm-ayt I heard
tf-ayna we threw (< df )
Why the changes occurred in WSA is, frankly, not clear at this point, as
is the question, why in some dialects and/or h continue on to . One
12
In Maltese the final -a of the suffix is conditioned by the historical pharyngeal , and
hence contrasts with, say, bn-eyt I built, where the suffix goes back to the diphthong
*ay-t.
In fact, the historical phonology of Maltese remains to be worked out in detail. Not
least is the problem of a not inconsiderable dialect variability, with its potential impor-
tance for historical reconstruction (cf. Owens 2006, chapter 7 on imla in Maltese).
Mifsud (1995, 308-9) explains the final /y/ in sm-ayt etc. as a change of the historical
pharyngeal trace to /y/. This analysis is interesting in and of itself, but probably deserves
an article of its own. Briefly, while Mifsuds analysis still maintains the paradigmatic
stability principle, it is on an a priori basis more complicated than the treatment offered
here. Mifsud notes that in general verbs with historical final voiced pharyngeals merge
with weak-final verbs (e.g. nitfa we throw, like ninsa we forget). The current analy-
sis sees the merger as having occurred throughout all inflectional paradigm members,
allowing for the underlying phonetic conditioning of [a] due to the pharyngeal trace.
Mifsuds analysis would split the paradigm of voiced pharyngeal-final verbs . In the pro-
cess this creates an otherwise unattested CCay-C stem , where the /y/ represents C
3
of the
root , and hence would contrast with the split paradigms -analysis offered above for WSA ,
since in the present analysis laryngeal -final verbs split into two existent paradigms.
paradigmatic stability and final laryngeals 651
can, of course, appeal to sub-stratal infuence. Tis is quite plausible in
the case of WSA . However, why should the shif have occurred in WSA ,
but not in another notable Sprachinsel where Arabic is a remarkable
minority language, Uzbekistan ?
13
In short, the phonological shif to /h/
and // and the further loss of these sounds, is history in its truest sense,
contingent on particular developments in particular places, in particu-
lar times. Tis contingency is in evidence in WSA in the very variability
of the process. It is unfortunate that longer term trend studies cannot
be followed through on to trace the further development of this phe-
nomenon. At what point, if ever, does the change go to completion, and
why?
In any case, once these sounds are given up, the paradigmatic stability
principle takes over. Tis obtains in all varieties of Arabic that have (1)
above, in WSA , as described here, and in Maltese as well, as suggested
around the discussion of (10) above.
In short then, the sequence of developments is as follows:
(11) loss of fnal laryngeal
(12) obeisance of the paradigmatic stability principle (see (9) above)
In other words, given the loss of the laryngeal , collapsing into the weak-
fnal paradigm follows automatically. In the current framework, only
(11) is history proper. (12) is suggested to be the instantiation of a gen-
eral linguistic principle in Arabic verbal morphology . In a sense, so long
as Arabic exists, (12) will be operative in the way described here. Its
reappearance at diferent points in the history of Arabic , however, is not
compatible with a conceptualization of Arabic as having changed from
and Old type to a New type . To the contrary, because the same struc-
tural forces are at work in pre-diasporic times as are at work in 2006, no
change has occurred.
4. Appendix
Verbs used in test frames (given in the form of a possible answer): dbaho
they slaughtered; simt I heard; wjt I faced; karaho they hate; gat
I cut; garti you FSG stopped; mantu you MPL prevented; binfaan
they F are useful; budbaho they M slaughter; bisaman they F hear;
13
Moreover, in WSA emphatic sounds are fully maintained, but they are lost in
Uzbekistan Arabic .
652 jonathan owens
tuwjuhu you MPL face; bikrahu they M hate; bitbao they M follow;
bat she sold; bibu they M sell; gaao they M cut.
5. References
Behnstedt, Peter . 1985. Die nordjemenitischen Dialekte . Teil 1: Atlas. Wiesbaden:
Reichert.
Chambers , Jack . 1995. Sociolinguistics Teory. Oxford: Blackwell.
Ibn Mujhid , Ah mad b. Ms. 1972. As-Saba f l-qirt. Ah mad awqi D ayf, ed . Cairo:
Dr al-Marif, 1972.
Mifsud , Manwel . 1995. Loan Verbs in Maltese . Leiden: E.J. Brill.
Owens , Jonathan . 1998. Neighborhood and Ancestry: Variation in the Spoken Arabic of
Maiduguri , Nigeria . Amsterdam: Benjamins.
. 2006. A Linguistic History of Arabic . Oxford: Oxford University Press.
Rabin , Chaim . 1951. Ancient West Arabian . London: Taylors Foreign Press.
Roth-Laly , Arlette . 1979. Esquisse grammaticale du parler arabe dAbbeche . Paris:
Geuthner.
Versteegh , Kees . 1984. Pidginization and Creolization : the Case of Arabic . In Amsterdam:
Benjamins.
. 2004. Pidginization and Creolization Revisited: Te Case of Arabic. In Haak ,
Martine, R udolf de Jong, and Kees Versteegh , eds. Approaches to Arabic Dialects:
A collection of articles presented to Manfred Woidich on the occasion of his sixtieth
birthday . Leiden: Brill, 343358.
SOME ASPECTS OF DIGLOSSIA AS REFLECTED IN THE
VOCABULARY OF LITERARY AND COLLOQUIAL ARABIC
Judith Rosenhouse
Swantech Ltd., Haifa, Israel
1. Introduction
1.1 Preliminary remarks
Tis paper focuses on diglossic diferences between literary and collo-
quial Arabic mainly from the morpho-phonological and lexico-semantic
aspects .
Te lexical part of the language is probably the area that most clearly
distinguishes between these varieties of Arabic. Our survey deals with
phonology , morphology and lexicon/semantics .
Afer the introduction, the second part deals with phonological
features that distinguish literary Arabic from colloquial Arabic (e.g.,
consonants such as /q, j/ and /t /, the vowel of word-fnal imla, and the
emphatics ). Te main diferences between the two varieties of Arabic
refect the well-known typical segmental features of colloquial Arabic.
Next, morphological or morpho-phonological comparison between the
phonological system of cognate lexical items in colloquial Arabic (as
spoken in Israel ) and literary Arabic is undertaken.
Literary Arabic uses a considerable number of lexemes in their
foreign forms ofen adapted to Arabic morpho-phonology ; in such cases
colloquial and literary Arabic lexemes are ofen cognate . We look briefy
into the lexical areas where these phenomena occur.
Tese features can be described within language interference or
language contact as terms of one linguistic approach, or as code mixture
according to another. From this point of view, these features do not difer
much from phenomena in other Arabic dialects (Boucherit , 2000, Heath
1989, Owens , 1998, Taine-Cheich , 2000) or languages (e.g., Ritchie and
Bhatia , 2004).
654 judith rosenhouse
1.2 Introductory remarks
Te centuries-long development of modern literary Arabic , elsewhere
also Modern Standard Arabic, and colloquial Arabic , the two faces
of the Arabic language , and the diferences between literary Arabic
and colloquial Arabic , have been the topic of much linguistic debate
especially since Ferguson s (1959) well-known paper on diglossia (e.g.,
Blau , 1977, Blanc , 1960, Bousafara-Omar , 2005, Diem , 1974, Fernandez ,
1993, Fck , 1950, Haeri , 2003, Holes , 1995, Kaye , 1994, 2002, Mejdell ,
2002, Meiseles , 1975, 1980, Monteil , 1960, Stetkevych , 1970, Versteegh ,
1997). As this area is also of much interest for Professor Kees Versteegh
I dedicate this paper to him.
Literary and colloquial Arabic reveal diferences on all linguistic
levels, including phonology , morphology and syntax , in addition to
vocabulary , as described in Ferguson (1959). Our recent work on
trilingual dictionaries (Rosenhouse , 2001, 2004) has given us a genuine
and tangible feel for this situation from the lexical aspect of Arabic
diglossia in Israel , and the present paper is partly based on items in
those volumes.
Some random examples of such diferences are shows in Table 1
below. We take into account the relevant literature, and rely on a lexical
search of several Arabic newspapers from which we gathered examples
for the studied points.
TABLE 1 EXAMPLES OF LITERARY AND COLLOQUIAL ARABIC
NON-COMPATIBLE LEXEMES
Literary Arabic Colloquial Arabic Gloss
dumya lube Puppet
ihd t arh Abortion
ih tad a libes el-kundara To put shoes on
zujj qazz Glass
kura t be Ball
zawraq fka Boat
Te situation of the Arabic language has recently been described as
follows in Ayyoub : In societies where a substantial segment of the
population is literate, and two variants of language usage exist, the gap
between the literary/written and spoken Arabic tends to diminish. With
the gap disappearing, a continuum emerges, ofering the speaker a rich
array of possibilities (2002, see also Kaye 2002). Such a mixed language
diglossia as reflected in the vocabulary of arabic 655
can be examined at present from oral material, or from written texts
dating from the 7th century to the present (e.g., Blau 1977,
1
and see
Chapter 1 in Mazraani 1997, for a review of relevant literature). For
Arabic, this mixture is known also as middle Arabic (al-lua al-wust ,
Badawi 1973).
2
Te importance of the lexical aspect in the distinction between literary
and colloquial Arabic is partly due to the sheer number of the lexical
items , which naturally far exceed the number of grammatical structures
and phonological categories .
Our research hypothesis and therefore starting points are based on
two assumptions: 1. More literary Arabic lexemes (and more loanwords
from various languages ) are used in colloquial Arabic, than colloquial
Arabic lexemes are used in literary Arabic (cf., e.g., Al-Falay 1996,
chapter four). Tis apparently is because literary Arabic ofen uses loan
translation (calque ) or new terminology coined by ofcial authorities
such as academies of Arabic language in Egypt and elsewhere (e.g.,
Shraybom-Shivtiel 2005) instead of loanwords . 2. When foreign lexemes
are used in literary Arabic or colloquial Arabic, they usually adapt to the
Arabic system phonologically, and also morphologically when necessary
and possible.
1.3 Literary and colloquial Arabic cognate words
Literary and colloquial Arabic cognate words vary mainly in phonologi-
cal and/or morphological features and not semantically (Cadora 1979).
Most of the colloquial Arabic examples referred to below refect col-
loquial Arabic as spoken in Israel . Te picture we get from this coun-
try may difer somewhat from that in other countries. In fact, literary
Arabic is not uniform either, and literary Arabic lexemes vary in diferent
1
Several views of middle Arabic do not consider it a separate variety or a variant of
the language, for it lacks a grammatical coherence : we simultaneously fnd correct forms
of the literary Arabic and deviations from these forms . We even fnd hybrid forms that
are neither literary nor dialectal (Ayyoub , 2002. Cf. e.g., Kaye , 2002, Mejdell , 2002).
2
Diglossia is also related to bilingualism , since two language systems are dealt with
(cf. Fasold , 1984, Rosenhouse and Goral , 2004). Thus, when speakers use colloquial
Arabic lexemes in a literary Arabic context, or vice versa, they may do it either for spe-
cific functional (stylistic) purposes or as an unintentional process of code switching .
Although we do not deal with code switching in this paper, I believe that the use of liter-
ary Arabic words in a colloquial Arabic context is not to be considered code switching
or code mixing if the literary Arabic word is already integrated in colloquial Arabic and
used in it as the only accepted lexeme for that specific notion.
656 judith rosenhouse
Arabic-speaking countries , in particular in the felds of technical
and scientifc vocabulary (Holes 1995, Badawi 1997). For simplicity,
we refer to literary and colloquial Arabic as distinct and independent
systems.
Te following sections of the paper focus on a comparison of
contemporary literary and colloquial Arabic vocabulary as refecting
Arabic diglossia , and not on their diachronic development . Afer a short
description of phonological and morphological features of literary and
colloquial Arabic vocabulary ( 2) , three aspects relating to semantics
are examined:
(1) the use of foreign lexemes in literary and colloquial Arabic ( 3.1) ;
(2) the use of colloquial Arabic words in literary Arabic and literary
Arabic words in colloquial Arabic ( 3.2) ;
(3) semantic diferences between cognate literary Arabic / colloquial
Arabic words ( 3.3). Section 4 concludes this essay.
2. Phonological and morphological diferences between literary
and colloquial Arabi c
Te brief description of phonological and morphological diferences
between literary and colloquial Arabic lexical pairs in this section is
intended to demonstrate processes that distinguish literary from col-
loquial Arabic . Tese processes appear in the oral articulation of literary
and colloquial Arabic and mark them clearly. Ferguson (1959) consid-
ered such lexical literary / colloquial Arabic pairs as an important fea-
ture of diglossia . For a detailed phonological and morphological analysis
of colloquial / literary Arabic lexical items in the discourse of speakers
from three countries (Egypt , Iraq and Libya ) see Mazraani (1997).
2.1 Phonology
Phonological diferences between literary and colloquial Arabic in Israel
and elsewhere are among the most studied and noted features in the
literature of Arabic diglossia (e.g., Altoma 1969, Blanc 1960, Meiseles
1980, Fischer and Jastrow 1980, Holes 1995, Mazraani 1997, Rosenhouse
1984, Versteegh 1997). Usually, speakers of a colloquial Arabic dialect
use the same consonants for literary Arabic speech as for colloquial
Arabic . Te emphatics /s , d, t ,
/ in certain (rural)
dialects. Sometimes such dialect speakers do not merge these phonemes
in literary Arabic . Te vowel system of many colloquial Arabic dialects
(in Israel ) includes two vowels that do not exist in literary Arabic : /e, o/,
and the parallel long monophthongs /, / which usually refect the liter-
ary Arabic diphthongs /ai, au/ respectively. When speaking in literary
Arabic , the traditional literary Arabic articulation of these diphthong s,
as diphthongs, is usually retained in Israel .
Let us discuss now the problematic phonemes of the system as they
occur in Israel . Mainly Muslim , Christian , and Druze speakers of rural
dialects pronounce the phoneme q as uvular . In urban dialects this
phoneme is articulated as glottal / / and in Bedouin dialects and some
rural dialects its pronunciation varies, ofen under certain conditions,
between /g/ and /j/. However, Bedouins and urban speakers do
articulate the phoneme /q/ as uvular when reading the Qurn or when
certain literary Arabic lexemes are used in words such as al-qurn the
Koran, qadiyye problem, issue, qawmiyye nationalism, or dimuqrt iyye
democracy
3
(cf. Mazraani 1997, Haeri 2003). A similar process refers to
the other phonemes under discussion here: the colloquial Arabic dialects
vary in their articulation between the urban /d-d-z / for /z / and /s, t/ for /
t /, and the respective traditional literary Arabic articulation /d/ and /d/
which simultaneously characterizes rural and Bedouin dialects . Te four
basic emphatics /s , d, t , z / (especially at word-fnal position ) are ofen
3
In non-phonetic dictionaries the transcription usually hides such dialect difer-
ences . Also in our dictionary only one letter transcribes q in colloquial Arabic .
658 judith rosenhouse
pronounced as non-emphatic phonemes in some colloquial Arabic
dialects (in Israel ), mainly urban ones, as in /s/ for /s / in colloquial Arabic
/sand / for literary Arab ic/s undq / box (see Rosenhouse 2002). Tese
phonemes are, however, transcribed in colloquial Arabic dictionaries
(in Israel ) without consideration of the option of this non-emphatic
articulation . Colloquial Arabic (in Israel) has, on the other hand, a
few emphatic consonants that do not exist in literary Arabic , and are
usually phonetic variants of non-emphatic phonemes . Te phenomenon
is marginal in rural and urban colloquial Arabic in Israel and is more
prevalent in Bedouin dialects (in Israel and elsewhere). But it may form
phonemic distinctions in certain words such as /bb a / pope vs. /b
ba /
her door, or /aa / God vs. /al-la / he said to her.
Te vowel /a/ of the feminine ending in noun and adjective forms,
which in modern literary Arabic is pronounced /a/, may be /a, e/ or
/i/ in colloquial Arabic (in Israel ), with diferences depending on
dialect features and phonetic factors. Te phonetic condition of urban
and rural dialects in this region is that if the consonant preceding
the sufx is pharyngeal , laryngeal , emphatic , or uvular , the vowel
remains /a/; cf. /sa / hour, clock; otherwise it bends upward (imla
of classical Arabic ), as in e.g. /sane, sini / year. Te /a/ is also a feature
characterizing (colloquial Arabic ) Bedouin dialects . Tus this sufx
uses the vowel /a/ in formal (literary Arabic ) and Bedouin speech , but
also in (non-Bedouin) colloquial Arabic when preceded by a back or
guttural consonant . Te diference in the use of /e/ or /i/ in this ending is
less easy to defne, because although it is sometimes apparently dialect-
dependent (see examples in, e.g., Geva-Kleinberger 2004) it ofen refects
free variation and is sometime also a more central /i/. Consequently,
four contexts of articulation of the t marbt a as /a/ (rather than /e/
or /i/) can exist: (1) formal, literary Arabic ; (2) a Bedouin dialect ; (3)
phonetically-conditioned urban/rural colloquial Arabic (in Israel); or
(4) a mixed/switched utterance using literary Arabic pronunciation
within a colloquial Arabic structure (in Israel).
Te features briefy described here for colloquial Arabic (in Israel ) are
similar in principle to those in many other colloquial Arabic dialects ,
including Mesopotamian qltu and glt dialects (Blanc 1964, Jastrow
1978).
Unlike consonants, the vowel system difers in colloquial Arabic
from literary Arabic almost in all the Arabic dialects . Tis is expressed
by expansion of the phonetic vowel system on the one hand and the
merging of several vowels to a smaller set of phonemes on the other
diglossia as reflected in the vocabulary of arabic 659
hand (the latter is conspicuous in North African dialects ). Te wider the
diferences between the traditional literary Arabic phonological system
and that of daily colloquial Arabic , the more literary Arabic articulation
errors may be found in the speech of such colloquial Arabic speakers .
Tis also is a typical feature of the code-mixing phenomenon . Examples
abound in any recorded colloquial or literary Arabic text (e.g., Blanc
1960), and even presidential speeches (see Mazraani 1997).
2.2 Morphology
Morphological diferences between literary and colloquial Arabic (in
Israel ) exist in various categories. Te pronominal system exhibits well-
known diferences in the free and bound pronouns in the nominal and
verbal systems . See examples in Table 2.
TABLE 2 EXAMPLES OF MORPHOLOGICAL DIFFERENCES
BETWEEN LITERARY AND COLLOQUIAL ARABIC
Literary Arabic Colloquial Arabic Gloss
nah nu nih na, ih na we
antum intu(m) you, pl. m.
druki drek your sg. f. home
drukunna drku/drkun/drken your pl. f. home
katabtu katabt I wrote
taktubna (b)tukutbi/(b)tikitbi you sg. f. write/will write
Te sufxed object bound pronouns also difer: literary Arabic t alabtuka
vs. colloquial Arabic (in Israel ) t alabtak I requested you, m. sg., literary
Arabic fatah ahu vs. colloquial Arabic (in Israel) fatah o he opened it, etc.
In the nominal system we note literary Arabic af al vs. if al in several
local dialects of colloquial Arabic (in Israel ), e.g., literary Arabic aswad
colloquial Arabic (in Israel) iswad black in the colors groups. Another
example is the female form of some adjectives, which in literary Arabic
are of the fal pattern whereas in colloquial Arabic they are regular,
i.e., with the t marbt a , e.g. literary Arabic sakrn colloquial Arabic
(in Israel) sakrne drunk. Diferences between literary and colloquial
Arabic (in Israel ) also occur in plural patterns
4
as the examples in
Table 3 show.
4
We refer to those lexemes where the singular is identical or at least cognate in both
literary and colloquial Arabic .
660 judith rosenhouse
TABLE 3 EXAMPLES OF DIFFERENCES BETWEEN PLURAL FORMS
IN LITERARY AND COLLOQUIAL ARABIC (IN ISRAEL )
Literary Arabic Colloquial Arabic (Israel ) Gloss
br byra wells
ahur, uhr (u)hr months
numr nmura tigers
nuqs nawqes defects, faults
bah h rna bah h ra seamen
Such diferences are, of course, in addition to basically phonetic
diferences of cognates , existing in singular as well as in plural forms
see Table 4.
TABLE 4 PHONETIC DIFFERENCES BETWEEN COGNATE LITERARY
AND COLLOQUIAL ARABIC LEXEMES
Literary Arabic Colloquial Arabic (Israel ) Gloss
ism usum name
far fr mouse
mint aqa mant iqa region
ibar ubar needles, injections
asnn snn teeth
daqiq daqyeq minutes
astid a astze gentlemen, professors
In the verb system also certain verb patterns vary between literary and
colloquial Arabic (in Israel ). See regular verb patterns such as literary
Arabic qadara colloquial Arabic (in Israel) idir was able, literary
Arabic arafa colloquial Arabic (in Israel) iref knew, and their
respective non-past patterns: literary Arabic yaqdiru colloquial Arabic
(in Israel) yidar will be/is able, literary Arabic yarifu colloquial
Arabic (in Israel) yiraf will know/knows. Diferences are also found in
verbs of C1//, e.g., literary Arabic yaxud u - colloquial Arabic (in Israel)
yxud/yxud /yxed will take/he takes/, C2w/y/ pattern, e.g., literary
Arabic xifu colloquial Arabic (in Israel) xuf I was afraid, C3w/y
pattern, e.g., literary Arabic baqiya colloquial Arabic (in Israel) baqa /
biqi he remained, literary Arabic qara a colloquial Arabic (in Israel)
qara, qiri he read, and many more.
Diferences of these kinds are well known from the literature, and in
a dictionary presenting both literary Arabic and colloquial Arabic (in
Israel ) (such as Rosenhouse 2001, 2004) one at once sees the abundance
diglossia as reflected in the vocabulary of arabic 661
of these forms. However, the main goal of a dictionary study is word
meanings, and to this issue we now turn.
3. Semantic considerations of the diferences between literary Arabic
and colloquial Arabic lexemes
3.1 Te use of foreign lexemes
Like other languages, Arabic has absorbed many lexical items from
foreign languages from its classical past to the present. Since the sec-
ond half of the 20th century the major source of foreign loanwords in
Arabic has been the American English language , which has become the
modern lingua franca of the world. Before that, French , Italian , Spanish ,
Turkish , Persian , Latin , Greek and others made lasting impacts on Ara-
bic vocabulary (Rosenhouse , to appear).
Literary Arabic was revived in the 19th century through the diverse
eforts of writers, journalists, scientists and linguists. Tese eforts led
to the establishment of academies for the Arabic language in Damascus
(1919), Cairo (1934), Baghdad (1947) and Amman (1976). Coining new
terminology for modern notions is one of the declared tasks of these
academies, as it is of many language academies elsewhere. Te linguistic
methods applied include borrowing terms and integrating them into
the language phonologically and morphologically; extending meanings
of existing (Arabic) words; analogical creations based on existing
roots; translation of foreign words into Arabic (calque ); and blending
(Stetkevych 1970, also quoted in Suleiman 2005, Shraybom-Shivtiel
2005). Such Arabicizing of foreign terminology goes on in other Arabic-
speaking countries too (e.g., Al-Qahtani 2002, for Saudi Arabia , and
Kharbush 2002, for Jordan see Badawi 1997). Tis activity has added
thousands of new words to the literary Arabic vocabulary , although
there is no full agreement between the various language institutes
(whatever their titles) in Arab countries and although many of these new
words remain buried in dictionaries without enjoying general public
use (Badawi , 1997). Whatever the distribution of lexical innovations
in literary Arabic vocabulary, this process clearly contributes to the
decrease in the number of foreign terms in literary Arabic , particularly
in the semantic felds of modern technology and science .
Unlike literary Arabic , colloquial Arabic is not subject to any ofcial
language policy and it develops spontaneously, albeit under the efect of
internal and external circumstances. Accordingly, foreign terminology
662 judith rosenhouse
fows freely into colloquial Arabic and is integrated in keeping with
the needs of its speakers (see Rosenbaum 2000/2). As noted, this
terminology now comes mainly from English , but residues of the efects
of other languages from earlier periods can still be found.
In Israel the situation is somewhat diferent from Egypt (as Rosenbaum
ibid. describes it), since in Israel Hebrew is the dominant language
(Rosenhouse forthcoming). Since the establishment of the state of Israel
in 1948 native speakers of Arabic in Israel have been in contact with
Hebrew ofcially and spontaneously. Since they use Hebrew daily, the
latter has a strong impact on their Arabic mother tongue . Te fact that
they study Hebrew at school from grade three to the end of high school
reinforces their Hebrew profciency . Hebrew has been the major source
of borrowing of foreign lexemes for colloquial Arabic in Israel for the
last fve decades or so. English lexemes (including American-English)
also penetrate colloquial Arabic directly, or through Hebrew that ofen
serves as mediator (see, e.g., Abdeen 2004, Amara and Spolsky 1986,
Koplewitz 1989). A simple dictionary search soon reveals many cases
where literary Arabic uses Arabic words (ofen modern innovations)
while colloquial Arabic (in Israel) uses foreign words for the same items
(see Table 5). Literary and colloquial Arabic also share identical foreign
loanwords . Some of these words in colloquial Arabic (in Israel) may be
transferred directly from the foreign language , or indirectly via literary
Arabic , in particular when they refer to higher and modern lexical
domains such as politics , religion or other cultural areas , e.g., ittis l
communication, ofen by phone, ijr:t procedures (in court), asm
muddda antigens, is t iniyy artifcial, ih tiyt reserves, ih s iyyt
statistics, awz safar passport (see more examples in Table 6). Te
best sources for such examples are newspapers and weekly magazines,
which usually use the literary Arabic variety nearest to middle Arabic .
On the other hand, a large share of lexemes of basic daily colloquial
Arabic (in Israel) vocabulary refers to proper names of products and
place names, which are not usually translated into Arabic, e.g., jeans,
coca cola, jeep.
5
5
Usually, such words are not translated into any other language, though they may be
adapted to them phonologically and morphologically, as described above.
diglossia as reflected in the vocabulary of arabic 663
TABLE 5 EXAMPLES OF ARABIC LEXEMES IN LITERARY ARABIC
AND FOREIGN LEXEMES IN COLLOQUIAL ARABIC
(FOUND IN ROSENHOUSE 2001)
Literary Arabic words Foreign colloquial Translation
Arabic words
art an-nr fyz (electric) fuse
at ra sandw sandwich
amat al-itil bgiyye plug (in a car)
t abt ba rakt racket (in tennis game)
ajala ih tiyt iyya spr spare (tiyre in a car)
miqwad strin steering (in a car)
muh arrik matr motor
mujhir mikroskb microscope
funduq otl hotel
TABLE 6 EXAMPLES OF IDENTICAL FOREIGN LEXEMES IN
LITERARY AND COLLOQUIAL ARABIC (FROM ROSENHOUSE 2001,
AND AL-AHRM, 4/3/2006)
Literary Arabic Colloquial Arabic Gloss
al-aksada al-aksade oxidation
influenza at -t uyr Id. bird influenza
al-interbank ad-dlriyy Id. The Dollar Interbank
arf istit l ar-ray Id. archives of public
opinion polls
aristuqrt iyya Id. aristocracy
bank Id. bank
bra Id. beer
brtn Id. protein
btagz Id. gas stove, oven
daktr Id. doctor
ad-diktafn Id. dictaphone
dmuqrt iyy Id. democratic
elektrniyya elektrniyya electronic
mashariyya fza vase
anarl Id. general
hmglbn Id. hemoglobin
istd ist d stadium
ihz al-mbil belefn
6
mobile phone
6
Based on the Colloquial Arabic term for this device.
664 judith rosenhouse
malyn Id. million
malynr Id. millionaire
mawqi gugl Id. Google site
milyr Id. milliard/billion
mda mda fashion
miknkiyy mekanki mechanic
msq klsk msq klskiyye classical music
tmtk tmtki automatic
al-tizm Id. autism
ar-rtr Id. Rotary (club)
s ln br Id. saloon bar
sekretra Id. secretary
sigr Id. cigar
simfnia Id. symphony
sinryu Id. scenario
slkn silikon silicon
sirmik Id. ceramics
as-s undq al-aswad es-sandq el-iswad the black box in a vehicle
amb Id. shampoo
slaidt Id. slides
fniyya Id. Chauvinism
klta uklt a chocolate
talifn (also htif) talafn telephone
zarfa zarfe girafe
3.2 Literary Arabic words in colloquial Arabic and colloquial Arabic
words in literary Arabic
Diglossia has been ofen described in rather idealistic terms, as com-
prising two varieties, literary Arabic and colloquial Arabic . However,
more and more voices (in the 20th century) describe Arabic on a con-
tinuum of various degrees and types of intermingling elements of these
two poles or varieties (see Ayyoub 2002, Badawi 1973, Blanc 1960, Diem
1974, Ferguson 1996, Fischer and Jastrow 1980, Meiseles 1975, 1980,
Mitchell 1986, Kaye 1994, Holes 1995, Versteegh 1997, etc.).
Since literary Arabic is the prestigious variety of Arabic , while collo-
quial Arabic is hardly deemed (by its native speakers, even today) to
merit any ofcial attention or reference, colloquial Arabic speakers will
naturally tend to borrow from literary Arabic more than literary Arabic
TABLE 6 (CONT.)
Literary Arabic Colloquial Arabic Gloss
diglossia as reflected in the vocabulary of arabic 665
(writers or speakers) would tend to borrow from colloquial Arabic .
7
Tis
prestige factor is also well known in language communities that borrow
from other languages (e.g. Owens 1988, Tomason and Kaufman
1988/1991, Tomason 2001, Weinreich 1967).
Moreover, since communicating in literary Arabic is obviously more
difcult for colloquial Arabic speakers than communicating in colloquial
Arabic , both lexical and grammatical elements from colloquial Arabic
slip into these speakers literary Arabic utterances by code-mixing and
codeswitching processes (cf. Haeri 2003, Meiseles 1975, 1980).
Te varieties developed in Arabic (typical of the 20th century) are
entitled al-lua al-wust middle Arabic (Meiseles 1975, 1980, Holes
1995, Kaye 2002, Versteegh 1997), and educated spoken Arabic
(Badawi 1973, El-Hassan 1978, Mitchell 1986), among others. It has
been observed that the modern version of Middle Arabic establishes
bridges between literary Arabic and dialects (Ayyoub 2002). Moreover,
the deliberate use of dialect within the written text has resulted not in
the dialect as a distant variant but as a level of literary Arabic (Ayyoub
2002; see also Rosenbaum 2000/2, 2004, and Somekh 1991, 1993).
A major feld of diferences between literary and colloquial Arabic
vocabularies concerns the semantic felds of the items. literary Arabic
has been used for high scholarly felds since classical times, and still
today it is the carrier of modern scholarly terminology in the realms
of religion , culture , science and technology , namely areas not typically
considered part of the colloquial Arabic world (see examples in section
3.1). However, with the integration of these modern topics and concepts
in the speakers daily life, their lexical elements (from literary Arabic)
penetrate the daily discourse , via the mediating stage of code mixing
until they integrate in colloquial Arabic and become an inseparable part
of its vocabulary .
8
An important part of colloquial Arabic oral communication and
vocabulary is ofen described as dealing with emotions and personal
needs (also in Holes 1995, Versteegh 1997). Tis is complemented by the
fact that personal topics are not normally discussed in literary Arabic .
7
That colloquial Arabic borrows and uses literary Arabic items more than literary
Arabic borrows from colloquial Arabic is readily seen in any dictionary.
8
For more examples of such lexemes see Diem 1974, 4647, section 9.54, and Abd
Al-l s introduction (1971, 7) to his dictionary of Colloquial Egyptian Expressions
Whose Origins are Arabic (i.e., cognates).
666 judith rosenhouse
Although literary Arabic certainly has ways of emotional expression, as
written literary Arabic prose and poetry show, the general attitude to
emotion in literary Arabic, even in modern writers, is still generally not
as warm as their attitude to the fowing and rich colloquial Arabic style.
Te diference between the serious, preaching style of literary Arabic
and the spontaneous colloquial Arabic speech has been demonstrated
and explained, for example, through the simultaneous use of literary and
colloquial Arabic in three Arab leaders speeches (Mazraani 1997). Arab
writers ofen complain that literary Arabic vocabulary and structure
need to be facilitated and made more natural, supple and personal,
like colloquial Arabic. For example, Tawfq al-Hakim and Mahmud
Taymur write in the beginning of the 20th century about their struggle
to make their works natural by using colloquial Arabic words and
expressions (see for example Diem 1974, 117 f.),
9
and Mikhail Nuayma
(among others) puts in quotation marks the colloquial Arabic words
he uses in his early stories (e.g., sanatuha al-jadda, sat d ahab), thus
marking their diference from the main story style. As a result of this
distinction, however, many Arab writers end up using colloquial Arabic
in their worksin drama, in novels and short stories, and always in
caricaturesto make their expressions more realistic as well as lively. In
Egypt this spontaneous development has reached the level of a writing
system (Rosenbaum 2000/ 2002, 2004).
Literary Arabic has developed and is now much more fuent and
easy than it was at the turn of 19th century, when the revival movement
was gaining power (Haeri 2003). Still, the general attitude to colloquial
Arabic and literary Arabic has remained largely unchanged.
3.3 Cognate literary /colloquial Arabic words and semantic diferences
between them
Te basic and largest part of the vocabulary in Arabic is shared by liter-
ary and colloquial Arabic , and includes pronouns , kinship terms , body
parts , social structure , animals , geographical objects , basic activities
(verbs), etc. Tese lexemes are cognate items, namely the same words
(expressed in the same consonantal roots), used in diferent forms
9
From another angle, Frayha (1955) describes his sons frustration and problems in
answering a simple question in literary Arabic . Frayha (1959) later suggested a method
for accomplishing the simplified language (although later on he stopped publicizing this
position).
diglossia as reflected in the vocabulary of arabic 667
(infections, derivations) in literary and colloquial Arabic . Te main dif-
ferences between literary and colloquial Arabic cognates seem to be not
semantic but phonological or morphological (see section 2 above), and
lexemes where the semantic notion is not shared by literary and collo-
quial Arabic are relatively few. Cadora (1979) analyzes the lexicon in the
Syro-Lebanese region , to which colloquial Arabic (in Israel ) belongs,
and we therefore refer to it here. Cadora realizes that Swadesh s list
(1952), on which his work is based, includes words whose meanings are
so general that cultural innovation or discontinuation does not afect
them greatly ( 26), which reduces their validity. Still, he fnds (27) in the
Syro-Lebanese varieties checked against this vocabulary list, that 151
(75.5%) of the 200 lexical items have non-contrastive compatibility (i.e.,
more or less, lexical similarity) Tis rate is even higher (165 or 82.5%) if
the dialect of Dr Ez-Zr is removed from the analysis. Te comparison
of literary Arabic and Syro-Lebanese colloquial Arabic varieties reveals
an even higher compatibility, which on the average reaches 91% (29).
Comparing the studied Syro-Lebanese dialects (except Dr Ez-Zr )
among themselves, Cadora fnds 96% non-contrastive compatibility
between each pair of dialects (32). Te diferences between these dialects
and more distant ones (Casablanca, Cairo, Jidda, Baghdad ) are larger, of
course, but lets keep to our region.
Cadora prepared another list, based on Swadesh s list (1952), and an
unpublished list by Ferguson and Said (1958), with his modifcations of
them. Tis new list fnally had 101 contrastive compatible items. From
the results of this work, on the whole, he fnds that Arabic tends to use
cognate words in literary and colloquial Arabic , while diferent words
(rather than cognates) occur in cases where the meanings difer between
literary Arabic and colloquial Arabic.
Diferences between literary and colloquial Arabic cognates may
arise when the literary Arabic items have several semantic felds , not
all of which are used in modern colloquial Arabic lexical items . Te
colloquial Arabic items may refer to one of the several semantic felds of
the literary Arabic item. Such cases may be due to the fact that modern
literary Arabic draws on a huge dialectal inventory from the past, which
is not known in all the colloquial Arabic dialect regions. Among the
most famous examples, is f, Literary Arabic (rarely used) to polish vs.
colloquial Arabic he saw (it is probably famous also because it appears
in Ferguson 1959). Other examples more specifc to the region of
colloquial Arabic in Israel are ujra, in literary Arabic wages, colloquial
Arabic taxes (wages in colloquial Arabic in Israel is more ofen ajar,
668 judith rosenhouse
ma or rteb , which are also used in literary Arabic); muallim in literary
Arabic means a teacher or instructor, but in colloquial Arabic teacher,
as well as a boss in a workshop, an ofce etc.
10
Kabs(a) is in literary
Arabic pressure, attack, colloquial Arabic heavy (rain) (parallel to the
English expression its raining cats and dogs); fat ina (literary Arabic)
/ ft en (colloquial Arabic)literary Arabic be clever, smart, colloquial
Arabic remember; qalbliterary Arabic heart, colloquial Arabic ones
internal part of the body, used also for the heart and entrails.
4. Conclusion
Tis chapter has focused on literary and colloquial Arab ic vocabulary .
An important feature of colloquial Arabic vocabulary is that it is more
afected by literary Arabic lexemes than literary Arabic is by colloquial
Arabic lexemes (as expected). Te reasons for this have been studied
elsewhere and we did not analyze them here. Sufce it to say that this
picture is largely due to the prestigious status of literary Arabic com-
pared to colloquial Arabic (see Rosenhouse forthcoming).
We have also seen that phonological and morphological features mark
many auditory / articulatory diferences between literary and colloquial
Arabic . Our semantic comparisons between literary and colloquial
Arabic lexemes have shown both diverging and converging trends of
development in each of these varieties.
A relatively small part of the sum total of Arabic lexical items is
composed of foreign loanwords that penetrate Arabic due to the
cultural trends of modernization / Westernization. In this section of
the vocabulary we also note that colloquial Arabic uses more foreign
loanwords (in Israel sometimes borrowed via Hebrew ) than literary
Arabic does. Tis state is apparently due to the fact that in many cases
literary Arabic uses translations (calque ) and newly coined terms rather
than the foreign ones used in colloquial Arabic (see Abu-Haidar 1992,
Shraybom-Shivtiel 2005).
We did not fnd many colloquial Arabic items in Israeli newspapers.
Tose found occurred mainly in advertisements and caricatures
where foreign loanwords relating to politics, shopping and economics,
10
Wehr (1971) gives also master of trade as one of the meanings of this word.
diglossia as reflected in the vocabulary of arabic 669
sports and entertainment, etc., are used, as in many other language
communities in the world. On many hoardings (especially in Arabic-
speaking townships or villages in Israel ), however, colloquial Arabic (of
Israel) vocabulary and structures play an important role, even if they
are not the norm. A considerable measure of linguistic infuence on the
general public comes from mass media such as radio and TV, where
advertisements are orally and visually published. Tese recurring ads
afect the public subconsciously so that this source of mixture between
literary and colloquial Arabic , which was on the increase all through
the 20th century, has been going on up to now. Abu-Haidar (1992, 104)
sums up the efect of modern standard Arabic on dialect convergence
(in Baghdad ) by saying that certain lexemes (in their MSA forms )
are now part of Muslim Baghdadi and Christian Baghdadi speakers
everyday experiences (105). She adds: salient features of Muslim
Baghdadi and Christian Baghdadi have for a long time constituted
closely defned boundaries separating the two dialects from each other
and from Modern Standard Arabic . At the present time, however,
these boundaries are shifing as some salient features become obso-
lete and more unassimilated Modern Standard Arabic forms are difused
into both dialects (p. 105). In her conclusion Ayyoub (2002) states:
From its journey across the 20th century, the Arabic language carries two
remarkable traits: never in the past have the written and spoken languages
been intricately intertwined; never before have the centrifugal forces of
standardization affected the dialects as much as now . . .
A separate question is, however,
what will become of the Arabic language in the future? If the past century
has sensibly modifed the modalities of functions between literary and
dialectal Arabic, present for more than one and a half millennia, multiple
paths remain possible (Ayyoub 2002).
Literary Arabic is a source of fascination (for researchers and native
speakers alike, if we may add) because of its relation to both the secular
and the sacred (Haeri 2003). Moreover, economic constraints and
globalization alike favor literary Arabic , as a common language for a vast
geographic area, more advantageous than multiple languages (Ayyoub
2002). Ayyoub (2002) poses additional questions about the future of
literary Arabic and colloquial Arabic in the 21st century: Will a form
of Middle Arabic be the future? Will the dialects become the national
languages of tomorrow?
670 judith rosenhouse
We see that the diglossic makeup of Arabic is changing and evolving
all the time, including the structure and vocabulary of this language.
In spite of the generally unfavorable attitude to code mixing /switching
between some foreign language and colloquial Arabic , these processes are
apparently inevitable under the modern global village circumstances.
Tese foreign lexical elements penetrate colloquial Arabic, the receiving
language (or matrix language , according to Myers -Scotton 1993), and
usually become an integral part of it. A similar process takes place in
literary Arabic , although apparently at a smaller (slower?) rate. Since
this process existed also in the past (see, for example, Al-Falay , 1996,
chapter four) without lessening the vitality of the Arabic language ,
we assume that the similar process of modern times will not lead to
an entirely diferent future development. In spite of the huge number
of lexical items (including borrowed and innovated words) and their
semantic felds in a language like Arabic, lexemes are an open class and
may be added, exchanged, renewed or discarded according to need.
When thinking of structural changes of Arabic due to code mixing or
code switching phenomena, we do not consider the lexical developments
of literary and colloquial Arabic as described here to be the straw that
breaks the camels back. Moreover, the features and examples in this
chapter do not difer much, in principle, from those found in other
Arabic dialects (Boucherit 2000, Heath 1989, Owens 1998, Taine-Cheich
2000). We therefore assume that the Arabic language will continue its
basic path sweeping along all the side trimmings noted in the changes
of its vocabulary. We also support the view that the relation between
the studied linguistic varieties in Arabic (Arabic dialects and literary
Arabic) is similar in many respects to the relations between diferent
languages in bilingual situations (Fasold 1984).
Te above analysis may be seen within the framework of phenomena
of language interference or language contact according to one linguistic
approach, in addition to bilingual code mixing according to another.
We hope the examination of the intertwining vocabulary of literary
and colloquial Arabic will contribute to the attempts to improve
our understanding of the Arabic language. Since the paths that the
Arabic language will take do not have to be identical to the paths of
the past, further attention to the issues raised here should prove to be
interesting.
diglossia as reflected in the vocabulary of arabic 671
5. References
Abd al-l , Abd al-Munim Sayyid. 1971. Mujam al-alfz al-mmiyya d t al-us l
al-arabiyya. Cairo: Maktabat Nahd a al-Mis riyya.
Abu-Haidar , Farida. 1992. Shifing Boundaries: Te efect of modern standard Arabic
on dialect convergence in Baghdad. In Broselow et al., Perspectives on Arabic
Linguistics IV, 91106.
al-Falay , Ibrhm S alh . 1996. Izdiwjiyyat at-lua: An-naz ariyya wa-t-tat biq. Ar-Riyd
King Sad University.
Al-Hassan , S.A. 1978. Educated spoken Arabic in Egypt and the Levant: A critical
review of diglossia and related concepts. Archivum Linguisticum, 8(2):112132.
Altoma , Salah J. 1969. Te Problem of Diglossia in Arabic: A comparative study of classical
and Iraqi Arabic. Cambridge: Harvard University Press.
Arnold , Werner, and Hartmut Bobzin, eds. 2002. Sprich doch mit deinen Knechten
Aramisch, wir verstehen es! 60 Beitrge zur Semitistik Festschrift fr Otto Jastrow zum
60. Geburtstag. Wiesbaden: O. Harrassowitz.
Ayyoub , Georgine. 2002. An Odyssey of Words: Evolution of the Arabic language in the
20th century. Al-Jadid Magazine 8:40 (Summer).
Badaw, Sad Mus t af. 1973. Mustawayt al-lua al-arabiyya al-mus ira f mis r. Cairo:
Dr al-Marif.
Badawi , Mohamed. 1997. Probleme des Fachwortschatzes im Arabischen dargestellt
insbesondere an der Terminologie der Teleinformatik. Hildesheim: Georg Olms Verlag.
Bhatia , Tej. K., and William. C. Ritchie, eds. 2004. Te Handbook of Bilingualism. New
York: Blackwell.
Blanc , Haim. 1960. Stylistic Variation in Spoken Arabic: A sample of interdialectal
educated conversation. In C.A. Ferguson , ed., Contributions to Arabic Linguistics,
81156.
. 1964. Communal Dialects in Baghdad , Cambridge, Mass.: Harvard University
Press.
Blau , Joshua. 1977. Te Beginnings of the Arabic Diglossia . A study of the origins of
Neoarabic. Afroasiatic Linguistics 4:428.
Boucherit , Aziza. 2000. Refexions sur le contact de langues partir du cas dAlger. In
Manwel Mifsud, ed., Proceedings of the Tird International Conference AIDA, Malta,
8388.
Bousafara-Omar , Naima. 2005. Diglossia. In Kees Versteegh et al., eds., Encyclopedia
of Arabic Language and Linguistics, vol. 1, 62937.
Broselow , Ellen, Mushira Eid , and John McCarthy, eds. 1992. Perspectives on Arabic
Linguistics IV, Papers from the Fourth Annual Symposium on Arabic Linguistics.
Amsterdam/Philadelphia: J. Benjamins.
Cadora , Frederic. 1979. Interdialectal Lexical Compatibility in Arabic: An analytical study
of the lexical relationships among the major Syro-Lebanese varieties. Leiden: E.J. Brill.
Diem , Werner. 1974. Hochsprache und Dialekt im Arabischen: Untersuchungen zur
heutigen arabischen Zweisprachigkeit. Wiesbaden: F. Steiner.
Elgibali , Alaa, ed. 1996. Understanding Arabic: Essays in Contemporary Arabic Linguistics
in Honour of El-Said Badawi . Cairo: American University of Cairo Press.
Fasold , Ralph, W. 1984. Te Sociolinguistics of Society, Cambridge: Blackwell.
Ferguson , Charles Albert. 1959. Diglossia. Word, 15: 325340.
. 1996. Epilogue: Diglossia revisited. In Alaa Elgibali, ed., Understanding Arabic: Essays
in Contemporary Arabic Linguistics in Honour of El-Said Badawi , 4967.
. (ed.). 1960. Contributions to Arabic Linguistics. Cambridge, Mass.: Harvard
University Press.
, and Majid Said. 1958. Lexical variants in Arabic dialects (unpublished).
672 judith rosenhouse
Fernandez , Mauro. 1993. Diglossia : A Comprehensive Bibliography, 19601990 and
Supplements. Amsterdam: J. Benjamins.
Fischer , Wolfdietrich, and Otto Jastrow . 1980. Handbuch der arabischen Dialekte.
Wiesbaden: O. Harrassowitz.
Furayha , Ans. 1955. Nawa arabiyya muyassara. Beirut: Dr a-aqfa.
. 1959. Tabst qawid al-lua al-arabiyya al us l jadda: iqtir wa-namd ij.
Beirut: American University of Beirut.
Fck , Johann. 1950. Arabiyya: Untersuchungen zur arabischen Sprach- und Stilgeschichte.
Berlin: Akademie-Verlag.
Geva-Kleinberger , Aharon. 2004. Die arabische Dialekte der Stadt Haifa in der ersten
Hlfte des 20. Jahrhunderts. Wiesbaden: O. Harrassowitz.
Haeri , Niloofar. 2003. Sacred Language, Ordinary People: Dilemmas of culture and politics
in Egypt . New York: Palgrave Macmillan.
Hary , Benjamin. 1996. Te importance of the language continuum in Arabic
multiglossia. In Alaa Elgibali, ed., Understanding Arabic: Essays in Contemporary
Arabic Linguistics in Honor of El-Said Badawi, 6990.
Heath , Jefrey. 1989. From Code Switching to Borrowing: Foreign and diglossic mixing in
Moroccan Arabic. London: Kegan Paul International.
Hetzron , Robert, ed. 1997.Te Semitic Languages. London: Routledge.
Holes , Clive. 1995. Modern Arabic: Structures, functions and varieties. London: Longman.
Jastrow , Otto. 1978. Die Mesopotamisch-arabischen qeltu Dialekte. Vol. 1: Phonologie und
Morphologie. Wiesbaden: O. Harrassowitz.
Kaye , Alan S. 1994. Formal vs. Informal Arabic: Diglossia , triglossia, tetraglossia, etc.,
polyglossia- multiglossia viewed as a continuum. Zeitschrif fr Arabische Linguistik,
27:4766.
. 2002. Diglossia . Te state of the art for the new millennium. In W. Arnold and
H. Bobzin eds., Sprich doch mit deinen Knechten Aramisch, 37988.
, and Rosenhouse, Judith . 1997. Arabic dialects and Maltese. In R. Hetzron, ed.,
Te Semitic languages, 263311.
Koplewitz , Emanuel. 1989. Te use and integration of Hebrew lexemes in Israeli spoken
Arabic. Multilingual Matters 71.181195.
Levin , Arieh. 1995. Grammar of Spoken Arabic in Jerusalem. Jerusalem: Magnes Press
(in Hebrew ).
, and Friedmann, Yohanan, eds. 2004. Studies in Honor of Moshe Piamenta.
Jerusalem Studies in Arabic and Islam 29, Jerusalem: Magnes Press.
Mazraani , Nathalie. 1997. Aspects of Language Variation in Arabic Political Speech
Making. Richmond, UK: Curzon.
Mejdell , Gunvor. 2002. Features of lua wust mixed discourse in spoken Arabic of
Egypt . In A. Youssi et al., eds. Aspects of the Dialects of Arabic Today, 31728.
Meiseles , Gustav. 1975. Oral Literary Arabic : Its main characteristics in speech and
reading, Ph.D. thesis, Te Hebrew University in Jerusalem (in Hebrew).
. 1980. Educated spoken Arabic and the Arabic language continuum. Archivum
Linguisticum 11:118143.
Mifsud , Manwel (ed.). 2000. Proceedings of the Tird International Conference AIDA,
Malta, 29 March2 April 1998. Paris: Association International de Dialectologie
Arabe.
Mitchell , Terrence F. 1986. What is educated spoken Arabic? International Journal of
the Sociology of Language 61, 732.
Monteil , Vincent. 1960. Larabe moderne. Paris: C. Klincksieck.
Myers-Scotton , C. 1993. Duelling Languages: Grammatical structures in code-switching.
Oxford: Clarendon Press.
Nuayma , Mxl. 1961.
6
Kna y m kna: Short Stories, Beirut.
diglossia as reflected in the vocabulary of arabic 673
Owens , Jonathan. 1998. Neighbourhood and Ancestry: Variation in the spoken Arabic of
Maiduguri, Nigeria. Amsterdam/Philadelphia: John Benjamins.
al-Qahtani , S. 2002. Arabization and the Teory of Language Planning: An Applied Study
of Terminology Arabization in Saudi Arabia . Beirut: Center for Studies of the Arabic
Union.
Ritchie , W.C., and Tej K. Bhatia ,. 2004. Social and psychological factors in language
mixing. In T.K. Bhatia and W.C. Ritchie, eds. Te Handbook of Bilingualism, 336352.
Rosenbaum , Gabriel M. 2000/2. Do you parler Arabi? Mixing colloquial Arabic
and European languages in Egyptian literature. Materiaux Arabes et Sudarabiques
(Nouvelle Serie), 10:1147.
. 2004. Egyptian Arabic as a written language. In A. Levin and Y. Friedmann, eds.
Studies in Honor of Moshe Piamenta, 281340.
Rosenhouse , Judith. 1984. Te Arabic Bedouin Dialects: General problems and a close
analysis of North Israel Bedouin Dialects. Wiesbaden: O. Harrassowitz.
. 2001. Trilingual Practical Dictionary: Hebrew literary Arabic -colloquial Arabic .
Rosh Ha-ayin: Prolog.
. 2002. Phonetic trends of colloquial Arabic dialects in Israel. In W. Arnold and H.
Bobzin, eds., Sprich doch mit deinen Knechten Aramisch, 599611.
. 2004. Trilingual Practical Dictionary: Literary Arabic -Hebrew colloquial Arabic ,
Rosh Ha-ayin: Prolog.
. Forthcoming. Te English language and its impact on Arabic in Israel. In. J.
Rosenhouse and R. Kowner, eds., Globally Speaking.
, and Mira Goral . 2004. Bilingualism in the Middle East and North Africa: A focus
on the Arabic-speaking world. In T.K. Bhatia and W.C. Ritchie, eds. Te Handbook
of Bilingualism, 835868.
, and Rotem Kowner, eds. Forthcoming. Globally Speaking: Motives for Borrowing
English Loanwords in World Languages. Clevedon, UK: Multilingual Matters.
Shraybom-Shivtiel , Shlomit. 2005. Te Revival of the Arabic Language as a Mission of the
National Ideology in Egypt . Jerusalem: Magnes Press (in Hebrew ).
Somekh , Sasson. 1991. Genre and Language in Modern Arabic Literature. Wiesbaden:
O. Harrassowitz.
. 1993. Colloqialized fus h in modern Arabic prose fction. Jerusalem Studies in
Arabic and Islam, 16:17694.
Stetkevych , J. 1970. Te Modern Literary Arabic Language: Lexical and stylistic devel-
opments. Chicago: University of Chicago Press.
Suleiman , Yaser. 2006. Arabiyya. In Kees Versteegh et al., eds. Encyclopedia of Arabic
Language and Linguistics, vol. 1, 173178.
Swadesh , Morris. 1952. Lexico-statistics dating of pre-historic ethnic contacts with
special reference to North American Indians and Eskimos. Proceedings of the
American Philosophical Society, XCVI:452463.
Taine-Cheich , Catherine. 2000. Refexions sur le statut des emprunts dans les langues
mixtes partir du cas mauritanien. In M. Mifsud, ed. Proceedings of the Tird
International Conference AIDA, 10712.
Tomason , Sara Grey. 2001. Language Contact: An Introduction. Washington D.C.:
Georgetown University Press.
, and Terrence Kaufman. 1988
1
/1991
2
. Language Contact, Creolization and Genetic
Linguistics. Berkeley: University of California Press.
Versteegh , Kees. 1984. Pidginization and Creolization: Te Case of Arabic. Amsterdam-
Philadelphia: J. Benjamins.
. 1997. Te Arabic Language. Edinburgh: Edinburgh University Press.
, et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics. Vol. 1. Leiden:
E.J. Brill.
674 judith rosenhouse
Wehr , Hans; J. Milton Cowan, ed. 1971. A Dictionary of Modern Written Arabic.
Wiesbaden: O. Harrassowitz.
Weinreich , Uriel. 1967. Languages in Contact: Problems and fndings. Te Hague:
Mouton.
Xarb, A. 2002. arakat at-tarb f l-urdunn. Amman: Ministry of Education.
Youssi , Abderrahim, Fauzia Benjelloun , Mohamed Dahbi , and Zakia Iraqui-Sinaceur,
eds. 2000. Aspects of the Dialects of Arabic Today: Proceedings of the 4th Conference
of the International Arabic Dialectology Association (AIDA), Marrakesh, April 14.
Rabat: Ampatril, 317328.
EVERYTHING YOU ALWAYS WANTED TO KNOW ABOUT
L, YIL TO SAY IN EGYPTIAN ARABIC *
Manfred Woidich
University of Amsterdam
1. Introduction
Te idea of contributing to the Festschrif for our esteemed friend and
colleague Kees Versteegh with an article on the verb l, yil (gl, yigl )
to say in Egyptian Arabic dialects and its various idiomatic uses, came
to me when I was reading a draf of his lemma Serial Verbs , which he
had written for the Encyclopedia of Arabic Language and Linguistics.
1
In
this lemma, he reconsiders my view of items like (01) as an originally
paratactic but now grammaticalized construction in order to introduce
a pseudo-complement,
2
and prefers to regard it as serialization, that is,
as a serial verb construction , albeit not without hesitation.
(01) huwwa raddi alk allak ?
huwwa radd al-k l-l-ak
he answered to-you said-to-you what
What did he answer you? LAB 118,8
Unfortunately, I do not share this view and I shall reafrm my position
in section 4 below. Since this pseudo-complementation is not the
only case for l, yil to appear in contexts and functions that involve
grammaticalization , it seemed appropriate to give an overview of several
other cases where l, yil clearly has lost its original lexical meaning
by a process known as bleaching and acquired new meanings and
developed into a function word . Afer making some brief remarks on
the syntactic behaviour of l , yil as a lexical verb in section 2, several
steps and ramifcations in the history of l , yil will be described in
* I should like to thank Rudolf de Jong for going through an earlier draft of this arti-
cle and giving me some valuable hints. Needless to say, any remaining errors are mine.
1
To appear as Versteegh 2007.
2
Woidich 2002, in particular pp. 183184.
676 manfred woidich
section 3 based on data collected over the last 40 years mostly from
written sources but also from some recordings of rural dialects . Te
starting point of these developments apparently is the use of l , yil by
the speaker to introduce reported speech
3
on discourse level
4
within a
pragmatic strategy, namely introducing a direct or an indirect quote, be
it the speakers own words or the speech of somebody else, in order to
show rejection and non-acceptance of an utterance, to give reasons for
acting in a certain way, to explain intent, to make a comparison between
two things or for other reasons. Te fnal sectionsection 5deals
with a lexical aspect and gives some examples of the use of l , yil with
vocatives and in delocutive derivations.
2. On propositional level: direct vs. indirect speech
Here I will briefy describe a syntactical point, that is, the introduction
of direct and indirect speech . Direct or quoted speech follows l, yil
as an asyndetic sentence without a complementizer :
(02) ana mu ayllik hti aa
Did I not say to you, bring an ashtray? RUH 112,-1;
(03) alitlaha tali uudi maya
She said to her, sit down with me! FWQ 107,-7
(04) ns kitr bau ylu f-nafsuhum w ana mli
Many asked themselves, what business is this of mine? BAHN 122,9
A sentence reporting indirectly what was said, however, may either
be introduced by the complementizer inn (05, 06) or be connected
asyndetically (07, 08):
(05) iddakatra lu inn abak tabna wayya
Te doctors said that your nerves are a little bit of HAM 95,7f
(06) l innu aynm andina llla diyyat wi bass
He said that he would sleep with us only that night LAB 16,8
(07) a kkallim wi l aybt baa
Daddy called and said that he would spend the night out of the house
LAY 113,3
3
Following Gldemann et al. (2002, viii), I use the term reported speech as a gene-
ric term for both direct and indirect speech .
4
In many languages, the verb say constitutes a source of various pathways in gram-
maticalization that can lead to distinct types of function words , see Heine et al. (1993),
and Heine et al. (2002, 261 ff ).
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say
t?
Tey said that foreigners came to ilBawi , which is the main town. Tey
say Americans . Tey came, allegedly, they took [with them] a golden
statuette in the shape of a cock. Tis is babble: they neither took anything
nor did anything. Would foreigners go around digging up things?
(Baariyya : Manda )
(39) lit imbri innak l ma-bitru ilmadrasa
She said yesterday that youincredibly enough!do not go to school
TAY 38,11
In all cases from (31) through (39), the rejection of real speechthat is,
utterances that were really madehappens by reporting it, introduced
by l and pronounced with the appropriate intonation. In frst instance,
it is this typical intonation of indignation and annoyance that makes it
clear to the hearer what the speaker means: disapproval and rejection.
Reporting it alone would not be sufcient for this purpose; intonation
must be an integral part of the construction. Te semantic content of
this intonation materializes in grammaticalized l. Tis means that
disapproval and rejection now belong to the semantic content of l ,
which in this way is recruited as a pragmatic marker for disapproval
and rejection.
13
Tis makes it possible that the development goes further
13
Indeed, for at least some speakers/writers l seems to be no longer associated with
the verb ql in this meaning. In MRR and BAHN, for example, we find it written with
Alif Mdda as whereas l as a verb preserves the original orthography and is written
with qf throughout.
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say
l, yi
to say