You are on page 1of 8

Linguistic Society of America

Intervocalic l in Umbrian Author(s): James W. Poultney Source: Language, Vol. 25, No. 4 (Oct. - Dec., 1949), pp. 395-401 Published by: Linguistic Society of America Stable URL: http://www.jstor.org/stable/409867 . Accessed: 11/04/2011 15:57
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=lsa. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to Language.

http://www.jstor.org

INTERVOCALIC 1 IN UMBRIAN
JAMES W. POULTNEY

The Johns Hopkins University For approximately a century students of the Iguvine Tables have recognized that the Umbrian phoneme represented by r in the native alphabet and by rs in the Latin alphabet must in a few cases, when in intervocalic position, be etymologically equated not with Latin d but with 1.1The theory did not at first win universal acceptance in the sense that i may be derived from original 1; some accepted d as the only possible source of f and explained the Latin 1 as secondary or else denied the etymologies which assumed the change 1 > r.Since the originality of the 1 in some cases is proved by cognates outside Italic, provided we accept the etymological equations (and some of them can scarcely be denied), practically all scholars now accept the change 1 > r for at least a few words. Von Planta (loc.cit. fn. 1) carefully examined the forms alleged to exemplify the change, attempting to define the conditions under which it occurred, and concluded that it was restricted to situations where intervocalic 1 was followed by a close e or by i. Since the number of relevant examples is small, Buck (69) declines to make any exact formulation of conditions, and editors generally have been satisfied with merely listing the probable instances.3 It is therefore with some feeling of temerity that I attempt in this article to reexamine the conditions of the sound-change. If my conclusions are acceptable, they may have some value for certain problems of Umbrian etymology and interpretation. If they are too uncertain, the list of new applications of the change 1 > r since the appearance of von Planta's work may be of some service. The strongest single support for i < 1 is kaietu I b 33, kafitu III 21, carsitu VI a 17, VII a 43. It is generally translated by such imperatives as appellat5, calat3, vocata,and most scholars now regard it as cognate with Lat. calare (e-stem reflected in Lat. kalendae), Gk. KaX ev.4 That 1 and not d is original is proved by and OHG in deriving the Umbrian form from *kadeto Devoto halan, KaXE? holSn.
1 It seems best to give the principal bibliographical references at the outset. Von Planta, Grammatik der oskisch-umbrischen Dialekte 1.291-7 (Strassburg, 1892-97), has the most comprehensive discussion of the problem. Of literature prior to von Planta, some of which is difficult of access and antiquated in matters of phonology, I have made sparing use, and here cite only Biucheler, Umbrica (Bonn, 1883). The most frequently cited works subsequent to von Planta are: Robert S. Conway, The Italic dialects (2 vols., Cambridge, 1897); F. Muller, Altitalisches Worterbuch (Gbttingen, 1926); C. D. Buck, A grammar of Oscan and Umbrian (Boston, 1928); A. von Blumenthal, Die iguvinischen Tafeln (Stuttgart, 1931); G. Devoto, Tabulae Iguvinae (Rome, 1937). 2 Corssen, Aussprache, Vocalismus, und Betonung der lateinischen Sprache2 2.15-6 (Leipzig, 1868-70); C. Pauli, Altitalische Studien 5.82-6 (Hannover, 1883-7). Conway (1.359-61) explained many Latin words as Sabine borrowings with 1 in place of original d, and in his glossary of Umbrian words in Vol. II he maintains a skeptical attitude toward 1 (= f) < 1, avoiding etymologies based on it. Similarly Petr, BB 25.127-58 (1899). 3 Cf. Brugmann, IF 18.532 (1906); Devoto 158. Biicheler 50; von Planta 1.291; Buck 69; Devoto 158, 170. 395

396

JAMES W. POULTNEY

is not really in conflictwith the generalview, since d is commonlyassumedas an intermediatestage betweenI and r.' utetu III 12, IV 30 as an imperativeof another e-stem verb is conveniently treated after kafetu, but unfortunatelyits etymology is less sure. Accordingto the usual view' it is connectedin origin and meaningwith Lat. ad-olito 'burn, make burn' (sacrificialoffering,or altar), and possibly with several Germanic words including OE elan 'burn', OIcel. ylr 'heat'. These equations, if valid, give support to the original1, but Conway (2.666) suggestedSabine origin for the I in Lat. adolre 'inflame', and Devoto7assumedan earlierL. *odere (: olere = odor : olor). For famefias II b 2, relationshipwith Lat. familia is generallyrecognized, and none of the recenteditorsaccepts Corssen'sassumptions of a form *famedia based on a suffixdifferentfrom that of familia. Since no very satisfactoryetymology for this group of words has been found,' Devoto (328, 353) may be right in asserting pre-Indo-European origin, but the validity of famefias as evidence for f < 1 is not affected. Among the most frequently cited examplesof f < 1 is arsir VI a 6, 7, the equationof whichwith Lat. alius is stronglysupportedby von Planta.'oDevoto (155-6) translates 'dedicationibus',with derivation evidently from the same root *ad- seen in arsie and arsmor.I favor the equationof arsir with Lat. alius on the ground that the essential precautionin the passage where it occurs is third personappearingon the scene or makinga noise against any unauthorized while the augurand the flamenare taking the auspices.For the Latin dissyllabic form alis, which in this case wouldbe the exact etymologicalequivalentof arsir, divergent explanationsare offeredby Sommer,IF 24.17-25 (1909: syncope or dissimilationin phrasealius alium) and by Skutsch,G1.2.154-6 (1910:analogy of pronouns). afepes, afipes, afpes I a 6 and elsewherein I a, I b, and II a, is commonly taken as abl. pl. of an o- or a-stem cognatewith Lat. adipibus,"which is used as a translationfor it, both being ultimately borrowedfrom Gk. dxetoa. The line of transmissionof the word is uncertain. Muller suggests that it reached Latin through the medium of an Italic dialect and that the Romans used d to representthe sound of f, but it seemsmorelikely that they receivedit while it still had d, the transitionalsound between 1 and i. Devoto (201) on semantic groundsmakes a good case for a differentinterpretation,derivingafepes from
Planta 1.297; Buck, loc.cit. 6 Von Planta 1.291-2, 295; Muller 6; Buck 69; Walde-Pokorny 1.88. 7 Page 376, where he translates 'odoribus imbuito'. 8 Loc.cit.; unfavorable comment in von Planta 1.292. 9 Walde-Pokorny (1.828) suggests extension of a stem *dhq-mo-'Niederlassung, Wohnstitte' or 'Haufe, Schar'. 101.291, 295, with references. Supported by Brugmann, IF 18.532 (1906), who however Ges. 63.173 (1911), taking it as dat. pl. = 'sanctis', with later rejected this view, Ber. sAichs. no convincing etymology. Blumenthal (15) and Goidhnich, Historia 8.264 (1934), translate 'alius' without discussion. 11Conway 2.597; Muller 19; and in the translations of Biicheler, von Planta, Buck, and Blumenthal.
6 Von

INTERVOCALIC 1 IN tMBRUAN

39f

a stem (af)ep-o-,where < ad-, and ep is from a root *ephfound in Gk. 1wo, aArm. ephem'cook'. Ahmune,an epithet of Jupiter in II b 7, has been brought into relationship with Lat. almus,alimonia,Alemona;12 but most scholars,on groundspartly phonetic and partly semantic, have rejected this etymology for others based on roots containing -d.l Pumpetias II b 2 is translated'quintiliae'by Biicheler(140-1), who mentions its resemblance to the name of Numa Pompilius.Blumenthal (39), apparently alone among modern editors, explicitly assumes I < 1 here, without however to Osc. pumperias. making quite clear his view of the relation of pumpeHias The majority of scholars14 regard the equation of the two words (generally consideredequivalent in form and meaning to Lat. quincuriae)as inevitable and view f as an errorfor r, inducedperhapsby f in the adjacentfamefias. Puptike, -es, appearsin III 27, 35, and in several passagesof IV, always as an epithet of Puemune, -es. The fluctuationbetweenk and g may be ignored Full equation here, and the spelling Pupi~es IV 26 is undoubtedlyan error.'6 of Puphikeas dative with Lat. Piiblic5 is possible, and most scholarsassume it or some similar etymology with r < 1.16 Von Planta (1.293, 333, 2.74-5) is unfavorabletoward the equation with Pablic6 and suggests several etymologies with i < d. A few other words deserve brief mention here for the mere reason that for each of them an interpretation has been proposedwhich assumesr < 1, though in no case has the interpretation in questionwon generalacceptance. II a 29 is connected with Gk. A7reXos by Muller 27, who proposes ampefia as a conjectural meaning'(junger) Wein' and regardsthe Umbrianword either as a loan from &AreXos or as a loan in commonwith from an aboriginal i5,reos language. Blumenthal (7, 70-1), giving it the sense 'omentum',reconstructs a primitive form *am-pel-i-a,with a root common to L. pellis, Gk. frirrXoor. Otheretymologicaldiscussion,not based on r < 1and mostly very inconclusive, may be found in Bticheler (133), von Planta (1.466), Conway (2.599), Buck (99), and Devoto (332). feiehtru III 16, 18, is so obscure that interpretationis almost impossible, while applicationof the word in supportof a sound-lawis of course out of the question.We seem to have here a -tro-stemsignifyingsome part of the wooden structurefor the private sacrificeconductedby the AtiedianBrothers.Biicheler (155-6) derivesfrom a root *fal- in Lat. faliscas, so that it would indicatesome sort of wooden frame, but the correspondence of e and a is not accountedfor. Scarcely better is Muller's conjectureusing a stem file- (466, under sup + material'.For etymologieswith cf. 176) with the meaning 'combustible fb--i-; 1
Biicheler 141; Muller 21.

"sVon Planta 1.294-5; Conway 2.597; Buck 183, 329; Blumenthal 71-2; Devoto 194. 14Von Planta 2.200; Muller 331; Buck 185; Devoto 353. 16Cf. R. G. Kent, Textual criticism of inscriptions 38 (Philadelphia, 1926). 16Biicheler 159;Muller 350-1; Buck 69. Blumenthal (11-3) translates 'Publico, -i'; Devoto 158, 384, where d in Popdico of the translation evidently represents an intermediate stage in the development of f from Etr. 1, according to the etymology which he adopts.

398

JAMESW. POULTNEY

< d cf. von Planta (1.462); the view which connects ferehtru with Skt. bhinatti, Lat. find5, giving a sense 'tignum', is adopted also by Conway (2.621) and Devoto (380). ?ihgefa III 15, apparently a sort of lattice railing, is generally translated The primitive form is *kinkelij 'cancellos' and compared with Gk. KLyKX,5. or *kinkeda (Buck 348); von Planta (1.361, 2:25, 42) wavers be(Muller 84) tween the two. The difficult term mersto VI a 3, 4, etc., which refers to the significance of the augural birds, or possibly to their location, is commonly translated 'iusto' and brought into the family of mers 'ius', Lat. modes-tus, Osc. med-dis, by assuming a form *medes-to-.17 Goidinich'8 connected mersto with Lat. meltom (meltom meliorem dicebant Paul. Fest. 122). surum, sorsom I a 27, 30, and in various other passages of I a, II a, V b, and VI b, is translated 'suillum' in most editions, the connection of r with Lat. 1 being explicitly stated by Biicheler (39); but von Planta (1.294, 2.43) and Buck (40, 190) prefer to assume a -do-stem as actual source of the Umbrian forms. Devoto (158, 244-5, 414, 438) equates only sorser of V b 12, 17, with Lat. suillo(as an adaptation), referring all the other examples to a stem extended from *seu-d- with the sense 'liquid'. tehtefim IV 20 is evidently a cover to be placed over the persuntru and is often translated 'tegumentum'. Derivation from the root of Lat. teg6 is extremely probable, but whether r represents 1, as in Lat. tictilem, or d, is doubtful.'9 Of all the examples in which the change 1 > f has been assumed, only a hand" ful can be used either as proof of the change itself or as argument for the conditions under which it occurs. The danger of circular argument here is very great, but if the sound-change can be strongly supported by even a few examples, there is no reason why it should not receive a moderately wide application where the phonetic environment is suitable, if semantic and other factors are equally favorable. I follow von Planta (1.295) in regarding kafitu karsitu, uifetu, fameiias as very difficult of analysis on any other basis than r' < 1; in arsir, despite my own preference for equation with Lat. alius expressed above, I have less faith, because of the current popularity of the interpretation whereby arsir = 'dedicationibus'. There appears no reason to doubt the view of von Planta (1.297) and Buck (69) that d was an intermediate stage in the change 1 > f. Since the change d > r in intervocalic position is regular and undisputed,20 the question here is to determine the possibility and conditions of a change of 1 to d, which would then be followed by the change of d to f. We cannot be surprised at the absence of forms showing the intermediate stage, since even in the oldest tables the change d > f has already taken effect, and the minor Umbrian inscriptions are so few and so short as to be of little help, but the frequency of gentile names
17Cf. Biicheler 43; von Planta 1.407; Conway 2.633; Muller 270; Buck 12, 58. 18 Historia 8.260 (1934). Unfavorable comment by J. B. Hofmann, Bursian's Jahresber. 270.96, and Ribezzo, RIGI 18.69 (1934). According to von Planta (1.407 fn. 1), this view had already been proposed by Panzerbieter, Quaestiones Umbricae 10 (Meiningen, 1851). "iVon Planta 2.25, 42. 20Von Planta 1.398-405;Buck 82-3.

INTERVOCALIC 1 IN UMBRIAN

399

feliuf I a 14 = filiu VI b 3, presoliafe VI a 12, Salier VI a 14. For this trouble-

in -dius in Latin inscriptionsfrom Umbriamay easily point to a change1 > d ratherthan to choice of a d- in place of an 1-suffix.21 Of the forms giving strongest evidence of i < 1 two, kafetu and uretu, are verbs with -&-as in Lat. habit5 etc. At the imperativesof second-conjugation same time maletomII a 18, pf. pass. ppl. of a verb cognatewith Lat. molS,and Miletinar VI a 13 probably have -e-.22Von Planta correctly observes that 1 changes before the close e and not before the open e,23but I believe that in treatingthe developmentof 1i and li (lii) he has gone astray. He assumesthat li remained,possibly as palatalized1,while l1 becamefi, and explainsthe maintenance of I in tefrali,uerfale, sorsalir,and perhapsdisleralinsust, as due to the dissimilatingeffect of r or i. But even if a precedingr (thougha differentphoneme) was able to prevent the developmentof r, it would scarcelyhave interferedwith I > d, whichvon Planta himselfassumesas the earlierchangein the developmentof 1. I believe, in otherwords,that I is regularlyunchangedbefore the open short i and that tefrali,uerfale,sorsalirare normal.24 Several difficulIn Pupfike,as was shownabove,it is uncertainwhether ties requireexplanation. i is from 1 or from d. Goetze, IF 41.96 (1923), in seekingan explanationfor the syncope (which must be assumed in order to provide an earlier intervocalic situation for the developmentof r) suggests that i may have been long. If his view is right, I, being close like j, may have occasioneda change 1 > r. The appearancein some words of i before an apparentlyshort i is admittedly emin view of the theory proposedfor tefralietc., but these three words, barrassing to Lat. animdlis,verb&lis, undoubtedlymembersof a stem-classcorresponding should carry great weight in comparisonwith forms containingr of doubtful origin.In Ahmune,though syncopemust have occurred,the vocalism as well as the sourceof the f is uncertain.For the foreignariepes, if the f cannotbe explained by Iguvine phonology,I (of iiUeta) may have becomed in some other dialect (whenceLat. adeps,adipis) in time to participatein the Iguvine changed > i. In ampefia (in the earlierformof arsirif it equatedwith Lat. alius), in famefias, and in pumpefias (if f is not an errorfor r), the i is prevocalic,but so is it in

some fluctuation between ii + vowel and ii + vowel there appearsto be a choice between two possible explanations: (1) the developmentof ii + vowel is connected with a variation between Li and lI (l1) as in von Planta 1.296; (2) ii + vowel regularlybecame fi, any differencebetween1i and 11being immaterial, and feliuf, presoliafe,Salier requirespecial explanationon the basis
21 Some examples are: Atidius CIL 11.6179(from Castelleone; = Lat. Atilius?); Ausidius CIL 11.4786 (from Spoleto; = Lat. Aurelius?); Turpidius CIL 11.4929 (from Spoleto; = Lat. Turpilius?); OrfidiusCIL 11.5967(from near Acqualagna; = Lat. Orbilius?). Bticheler, RhM 11.295-7 (1857), gives a list of such doublets, without however citing their provenance. A. Schulten, Klio 2.182 (1902), has a map showing the distribution of the -dius-type, with high frequency for the Umbrian and Sabellian territories. 22 Von Planta 1.296, 2.399-400; Buck 181. Cf. R. G. Kent, Lg. 14.215 (1938), in criticism of Devoto's view that maletu has -8-. 2 On the quality of the e- and i-vowels cf. von Planta 1.87, 94, 96, 107; Buck 34-5. 24 On disleralinsust I refrain from comment here. In faCefele and purtifele the vowel after the f may be secondary (von Planta 1.100, 2.29; Buck 190).

400

JAMES W. POULTNEY

of analogy, dialect borrowing,or the like. If we accept the first explanation,it seems better to reversethe relationset up by von Planta, so that Ii wouldmaintain its I as in tefrali, etc., and li would becomeii. Von Planta recognizesthe palatalizingeffect of i here, but if I did not changebeforeboth I and i, the palatalizing second form is more likely to have producedthe change. Yet actually I believe that palatalizationand change of 1 to r were producedboth by j and to Sturtevant (Pron.2?118) Lat. was closerbeforea by prevocalic . According than before a vowel consonant, and although the phenomenonin Latin may have been restrictedto Rome,25 there is evidence that it held good in Oscan and in Buck 34-5, ?44a) Umbrian,wheree, at least in the native alphabet, (cf. is frequently written for short i before a consonantbut very rarely before a vowel.26 Even if it were probablethat treatment of 1 was conditionedby a variation li : li, it would be difficultto set up such a variationamong the words now in is a difficultproblem,27 question.The distributionof the IE suffixes-/o- and -ii7obut there seems to be no good reason for assigningany of the words ampefia, arsir, famefias, pumpefias,feliuf, presoliafeto an -iio-class.Four of them fall within the scope of Buck's remark (66) that single writing of postconsonantal i is evidenceof its consonantal [and prevocalic] value; while arsir and presoliafe, In that these being in the Latin alphabet,are ambiguous. short, it is improbable 1 that was treated wordswere based on differentsuffixesand equallyimprobable
differently before i and prevocalic 1. feliuf, presoliafe, and Salier thus form a

residue in which the 1 seems to call for special explanation.feliuf may easily verb = Lat.fdl~re; have retainedits 1throughthe influenceof a first-conjugation 1 in such a verb, havingthe voweld, the Umbrianchange > i wouldnot operate, and the semantic connectionwith feliuf 'sucking'would be much closer (and hence much more likely to exert analogical influence) than with Lat. filius. presoliafeis more difficult,but if it containsthe same root found in Lat. salum (thus Biicheler48, Devoto 167), analogicalinfluencemight be exerted by an Umbrian*solowith I not subjectto change.Saliermay be a gentilenamewith an -iso-suffix,but is in general so uncertainthat it had better be left alone. The conclusionswhich seem to me most probableon the basis of our rather meagermaterialare as follows.Postvocalic1 in the dialect of Iguvium remained unchangedbeforeall back vowels and beforee (as in maletom) and i if 7 was not prevocalic (as in tefrali) but becamer (rs) before j (as in kafetu) and i (as I (1~)the same changeis a priori probablyin famefias). Beforei and prevocalic in question,but cannot be adeof the vowels probablein view of the closeness on The argument the positive side has centered quately supportedby examples.
evidence that a 26 veham,cited by Varro, RR 1.2.14, as a rustic form of viam, is taken as closer pronunciation of i before a vowel was not universal. 26 Cf. von Planta 1.100 for examples. sakreu V a 6, perakneu V a 7, appear to be the only instances of e for i before a vowel. 27 On these variants cf. von Planta 1.165-6; Conway 2.470-2; Brugmann, Gdr. 2.1.182-98. For Italic an exceptionally clear explanation appears in Buck, Oscan-Umbrian verb system 158-9 (Chicago, 1895). The suffix -iio-, indicated by ii in the native alphabet, appears especially in triia, in fourth-conjugation verb forms of the type herliei, Ose. fakiiad, and in Oscan gentile names of the type Kluvatiis, etc.

1 IN UMBRIAN INTERVOCALIC

401

largely around karfetu,uietu, and famefias, and on the negative side around maletom, tefrali,uerfale,and sorsalir.Other things being equal the conclusions here reached are favorable toward etymologies based on r < 1 for ampefia, and for feiehtru if h is a markof vowel length (cf. von Planta 1.58),but unfavorable for surum,sorsom, if taken as an o-stem (cf. von Planta 1.293), for Pupiike if i is short, and for Afmune and merstoif we assume9 or I beforesyncopetook place. In afepes the vocalism of the medial syllable is not very clear. gihgefa can have r < 1 only if based on a suffix-ia, with i absorbedafter palatalizing the precedingconsonant. Horse VI b 43 = Hutie I b 2, used by von Planta (1.535) to exemplifysuch treatmentof i, may have its own rs r fromd or from1.
arsir, if = Lat. alius, alis, probably has its i by samprasdrana; hence it would

be a close i capableof causing1 > f (cf. Buck 35, 60, 120). The i in tehterim may be of the same origin;a stem in -l1o-or -djo-seemsmoreprobablethan one in -di-, a very rare type, and the 1 in a -li-stem would not yield r accordingto the conclusionsreachedon the basis of tefrali,etc.

You might also like