Professional Documents
Culture Documents
By BERNARDCOMRIE
0 INTRODUCTION
MS CAUSE S
ES V ED0 EIO
John make
I
Mary
I
give
I
book
I
Fred
John make-give
I
Mary
I
book Fred
I
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 3
Comparing (3) with its noncausativeequivalent:
(4)
S
1 TURKISH
BLOCKAGE
2 CAUSATIVE
3 SYNTACTIC
DOUBLING
4 EXTENDED
DEMOTION
In the languages examined so far, it has for the most part been the
case that when ES is demoted down the Hierarchy, it is demoted to
the next available empty position on the Hierarchy, unless of course
the language in question allows doubling on certain syntactic posi-
tions higher up the Hierarchy. However, even in some of the lan-
guages we have mentioned so far there are apparent exceptions to
this paradigm case. Although French in general demotes ES in this
way, when tbere is an overt E D 0 it is often also possible for ES to
be demoted to Obl, to a prepositional phrase withpar:
(59) Je ferai manger les gdteaux par Jean
I shall make Jean eat the cakes
However, this may not in fact be an exception to the paradigm case
of stepwise demotion down the Hierarchy if, as argued for instance
by Seuren (1972: 4-7), the occurrence of the prepositional phrase
with par arises through passivization of the embedded sentence.
The derivation of (59) would then be:
(60) a Je ferai ,JJean manger les gdteaux]
b Je ferai B[lesgdteaux manger par Jean]
c Je ferai manger les gdteaux par Jean
Causativization, which transforms (b) into (c), would demote les
gdteaux from subject position (in (b)) to direct object position in (c),
but would have no effect on par Jean. Other languages with similar
phenomena include Urdu (Bailey, 1956: 64-6), where ES of a transi-
tive verb may appear as I 0 (postposition ko) or Obl (postposition
se ‘with, by’) and Dutch (Seuren, 1972: 50-2).
20 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
In Sanskrit, it is in fact impossible for ES to appear as 10;if it
does not appear as DO (which may be a doubled DO), then it must
appear in the instrumental:
(61) Tim habhih khidayet
her (DO) by-dogs (Instr.) devour-Caus-Optative
He should have dogs devour her
Again, the passive analysis, rather than extended demotion,
may be able to account for the phenomenon. I want now to turn
to some examples where the passive analysis is rather less attrac-
tive.
In Hungarian, the subject takes no inflection, the direct object
has the ending -t, and the indirect object -nakl-nek (the variants are
conditioned by vowel harmony) :
(62) A no" adta a Jili-nak az almri-t
the woman gave the boy I 0 the apple DO
The woman gave the boy the apple
Causative verbs have the suffix -(t)at/-(t)et.With the causative of the
intransitive, ES appears as direct object:
(63) A tanuldk vdr -at -jdk a tandr -t
the pupils wait Caus the kacher DO
The pupils make the teacher wait
With the causative of a transitive verb, it is not possible for there to
be two direct objects :
(64) *A professzor kitisztit-tat -ta a szabd-t a
the professor clean Caus the tailor DO the
ruhrijri-t
suit DO
The professor had the tailor clean his suit
nor is it possible for ES to appear as an indirect object:
(65) *A professzor kitisztit-tat-ta a szabd-nak a ruhdjd-t
Instead, ES takes the sufKx -Val/-vel:
(66) A professzor kitisztit-tat-ta a szabd-val a ruhdjd-t
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 21
The basic use of the suffix -Val is to indicate the instrument, e.g.
(67) A tanuld ir a ceruzci-Val
the pupil writes the pencil with
The pupil is writing with the pencil
Modern Hungarian does not have a productive passive construction
for finite verbs, in the way that most European languages do. How-
ever, it does have participles that may take a passive construction, as
in :
(68) Az occst -vel kiildott levelez6lapot megkaptam
the his-brother by sent postcard I-received
(69) Az occse dltal kiildott levelez6lapot megkaptam
I received the postcard sent by his brother
(The form with a'ltal 'through' is sometimes found instead of -Val
in causative constructions, although my informant considered it not
in accordance with current usage.) Thus although Hungarian does
not have a fully-fledged passive, we may be able, on the basis of the
participial construction, to make some case for the passive analysis
with regard to causatives.
Finnish has essentially the same causative construction as Hun-
garian. When an intransitive verb is made causative, ES becomes
direct object, e.g. (examples from Volodin et al., 1969):'O
(70) Koko seura nauroi
all company laughed
The whole company laughed
(71) Han nauratti koko seura -n
he laugh-Caus all company DO
He made the whole company laugh
When a transitive verb is causativized, ES stands in the so-called
adessive case, with the suffix -lla/-lla. This case has a wide range of
functions, but includes that of instrument-although Finnish does
not have passive constructions, not even passive participial construc-
tions, with the underlying subject expressed in the adessive :
(72) Muurarit rakensivat talo -n
bricklayers build house DO
The bricklayers build the house
22 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
(73) Mina rakennutin talo -n muurarei -1la
I build-Caus house DO bricklayers Adessive
I get the bricklayers to build the house
As a final example, I want to consider the causative in Gilyak
(Nivkh), an unclassified language spoken at the mouth of the River
Amur and on Sakhalin Island in the Soviet Far East. In Gilyak,
there is one case, the absolute case, with no inflection, that marks
subject, direct object, and indirect object (Panfilov, 1962: 126):11
(74) Ytyk phmeot'u riykyn khim-d'
father his-gun elder-brother gave
Father gave his gun to my elder brother
The causative is formed in Gilyak by means of a suffix -gu. In the
causative construction, it is possible for ES to be expressed by a
noun in the absolute case, but this is rather unusual, being used
basically only with verbs in one of the numerous nonfinite ('gerund')
forms (Panfilov, 1962: 126):
(75) N'iPajan c'o l~yg -t vi-gu -d'-ra
I fish catch Gerund go Caus
I make Pajan go to catch the fish
The more usual construction is for ES to go into the dative-accusa-
tive case, with the suffix -ax. In fact, this is the only use in Gilyak
of this case-the name used by Panfilov is rather misleading. This
case is used irrespective of whether the verb in the causative already
has no EDO, EIO, or not (Panfilov, 1962: 131; 1965: 60):
(76) UtkuoHa ererx Zo -ax vi-gu -d'ra
son into-river fish Dat-Acc go Caus
The son let the fish go into the river
(77) N'i Xeugun-ax erx gala-gu -d'
I Dat-Acc to-him hate Caus
I made Xevgun hate him (lit. hate to him)
(78) Ytyk utkuoUIu-ax oxt ra -gu -d'
father son Dat-Acc medicine drink Caus
The father made his son drink the medicine
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 23
Thus Gilyak has the ultimate in extended demotion: in nearly all
cases, ES is demoted straight to the position of other oblique con-
stituents.
To conclude this section, we must re-examine the relation between
the passive solution, as outlined above with respect to French sen-
tence (59) with par, and extended demotion. In a number of lan-
guages, we find that ES may be expressed by the same means as
expresses the agent in the passive construction (e.g. French, Sanskrit,
Turkish, possibly Hungarian), and we must ask whether the oblique
cases (Obl) in such sentences arise by passivization, or strictly by
demotion, possibly extended demotion, down the Hierarchy. If
the passive solution were to be justified in all examples, then the
relevance of the Case Hierarchy to causativization would be
reduced, though not eliminated, in that the position Obl would
play no role.
In fact, the passive solution seems feasible for French, Sanskrit,
and Hungarian, of the languages we have examined in some detail.
For Finnish it is much less attractive, for the simple reason that
Finnish has no passive construction with the subject expressed in the
adessive (instrumental) case : a passive construction of this type
would have to be set up specificallyin underlying structure to account
for the causative. The passive analysis fails when applied to Turkish
where, it will be recalled, the postpositional phrase with turajindun
‘by’ may only be used with causatives that have both DO and 10,
although it is used to form passives where only DO is present, cp.
(24) and (25), repeated below:
(79) Adam kadin tarafindan dov-iil-dii
The man was hit by the woman
(80) * Dijci mektub-u mudur tarafindan imzala-t-ti
The dentist made the director sign the letter
Thus Turkish provides a clear example of the relevance of strict
stepwise demotion down the Hierarchy, the position Obl being called
into play only when required by the Hierarchy and constraints on
syntactic doubling. Gilyak, on the other hand, provides a clear
example of extended demotion, since the so-called dative-accusative
expresses neither Subj, nor DO, nor I 0 in simplex sentences, and
must therefore be Obl.
24 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
When ES is demoted to Obl, the Hierarchy does not make any
prediction as to which particular other oblique case it should stand
in, although in all the languages we have looked at, except Gilyak,
ES stands in the same case as the agent in the passive construction,
or in the instrumental (case used to express the instrument, e.g. the
adessive in Finnish). The passive analysis might seem to provide some
explanation for this, yet the same phenomenon occurs in Turkish,
where the passive analysis must be rejected on other grounds. The
generalization seems rather to be that wherever a subject noun phrase
is demoted to Obl, irrespective of the rule or convention responsible
for the demotion, it always moves to the same case within Obl. Thus
passivization and causative demotion would be expected to have the
same effect, and in each individual language the linguist will have
to look for other criteria to enable him to choose between the two
solutions.
5 CAUSATIVES PHENOMENON
AS AN INDEPENDENT
One of the effects of the various features that make up the ‘paradigm
case’ is that causative constructions and noncausative constructions
end up having, in derived structure, exactly the same syntactic
structure. Causatives of intransitives are just like noncausative tran-
sitives, and causatives of transitives are just like noncausative verbs
with a direct and an indirect object. There are two objections one
might make to this line of analysis, both of which would tend to blur
the distinction between noncausatives and causatives (or rather
more generally, superficial simplex sentences for which a complex
underlying structure with embedding is presupposed), if not eliminate
it a1together .
The first objection would be that there is no need to set up
for causative constructions a structure different from that for
ordinary nonderived verbs. If in surface structure causatives look
like noncausatives, why bother to set up a divergent underlying
structure with extra embeddings 7 The constraints of the ‘para-
digm case’ serve to bring this abstract underlying structure into
line with other surface structure types, but are only required be-
cause we chose to set up the abstract underlying structures in the
first place.
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 25
The second objection is almost the exact opposite of the first. It
grants that the underlying structure of causatives involves embedding,
but claims that this is equally true of other sentences that have
objects of various kinds. Seuren (1972) argues that indirect objects
(datives) in all languages arise through the displacement of ES (to
use our terminology) via the contraction of a matrix plus embedded
sentence into a simplex sentence; in the present paper, this is the
analysis proposed for surface I 0 deriving from ES. In his analysis
of*the verb kill as Cause-to-die Lakoff (1970: 98-100, and more
generally 9 1-107) argues that the superficially simple transitive con-
struction X kills Y derives from the underlying X causes [ Y to die]
(itself derived from X causes [ Y to become dead]);admittedly, Lakoff
is treating kill as a causative verb (which it may be semantically,
though not syntactically), but the analysis could conceivably be
extended to all transitive verbs.
However, both of these alternative suggestions fail to take into
account some of the variations on the paradigm causative construc-
tion. For instance, the second proposal outlined above runs into
difficulty with languages like Sanskrit that allow two direct objects
per simplex sentence: it predicts that the embedded subject should
become an indirect object. The first alternative suggestion can en-
compass these data, but would run foul of a language that allowed
double direct objects only in causative constructions. And neither
proposal makes any headway with Gilyak, where a distinct mor-
phological case exists only in causative constructions. It has been my
intention, in the body of this paper, to present an analysis of the
causative that is both sufficiently constrained to make interesting
predictions, and sufficiently flexible to accommodate the variations
on the causative that are found in different languages.
Even in languages that confirm more closely to the para-
digm case, there are sometimes other syntactic peculiarities that
set off causatives from other constructions. For instance, in
Japanese, the reflexive pronoun zibun can normally only be co-
referential with the subject of a simplex sentence (Kuno, 1973:
292-293), e.g.
(81) John ga Mary o zibun no uti de mita
Subj DO self 's house in saw
John saw Mary in his (*her) house
26 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
Where the verb is causative, however, the reflexive may refer to
either MS or ES (Kuno, 1973: 294-8):
(82) John ga Mary ni zibun no uti de hon o yom
Subj I 0 self 's house in books DO read
-use -ta
Caus Past
John made Mary read books in his/her house
On the one hand, this provides an argument for an underlying
structure where ES is subject of an embedded sentence (whence
zibun may be coreferential with Mary, both in the embedded sen-
tence); on the other hand, it sets off this causative construction from
sentences with nonderived verbs that might seem, semantically, to be
causatives, such as korosu 'to kill':
(83) John ga Mary o zibun no uti de korosita
Subj DO self 's house in killed
John killed Mary in his (*her) own house
(This argument for not deriving kill from Cause-to-die is given by
Kuno (1973: 298-9); other arguments are given by Shibatani
(1973b), who has extended this argument, and others, to Korean
(Shibatani, 1973a).) In Japanese, the class of syntactic causatives is
well-defined, and nonsyntactic semantic causatives clearly do not
belong to it.
In Tagalog, if we restrict ourselves to the prepositions that indicate
the syntactic functions of noun phrases in the sentence, then we
find that MS appears as derived subject, ED0 and EIO as derived
direct and indirect object respectively; while ES appears as direct
object if there is no EDO, otherwise as indirect object. So causatives
of transitives look just like transitives, and causatives of transitives
look like noncausatives with direct and indirect object. But Tagalog
has another way of obligatorily marking syntactic positions: one
of the noun phrases of the sentence must be focused, and the
morphology of the verb encodes which syntactic position has been
focused on. In noncausative sentences, there is one set of verbal
affixes for focusing on Subj, one for DO, one for I 0 (plus others
not relevant here for other syntactic positions). In causative con-
structions, however, there is a set of verbal affixes reserved for
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 27
focusing on ES, whether this is realized as DO or 10, and distinct
from the verbal affixes used for focusing on ED0 and EIO. If we
take verbal focus markers into account as well as the markers of
syntactic position on the noun phrases, then there is a clear dicho-
tomy between causative and noncausative sentence types. (Causa-
tives in Tagalog are discussed by Schachter and Otanes (1972: 321-
330), who use the term ‘indirect-action verbs’. The system is reason-
ably complex, especially in its interrelations with other aspects of
Tagalog syntax (marking of syntactic positions by prepositions,
coding of focus in the verbal morphology), and I hope to be able to
present a fuller report at a later date.)
Thus we see that several languages do make a rigid distinction
between causative and noncausative sentence-types, often by leaving
traces in other syntactic phenomena of a more complex underlying
structure with an embedded sentence. Even in languages that do
have such a distinction, we find a certain amount of give-and-take
between the two classes of verbs. Presumably most linguists, given
Abaza ba ‘see’, rba ‘show’, and the information that Abaza causa-
tives are formed by prefixing r-, would conclude that rba is a causa-
tive verb. But Allen (1956: 157) notes that rba follows the concord
type of a noncausative verb, which is distinct from that of a causa-
tive. Presumably in the development of the language rba has moved
from one class to the other. There are clear criteria for differentiating
between the classes, but individual verbs may belong to different
classes at different points of time; there is thus no conflict between
the ‘rigid dichotomy’ and the ‘give-and-takebetween the two classes’.
In languages that lack such clear criteria, on the other hand, indi-
vidual linguists may well disagree on whether a given verb is causa-
tive or not, and I see no reason why individual native speakers
should necessarily be expected to internalize exactly the same gram-
mar on the basis of such data.
6 CONCLUSION
King’s College
Cambridge
NOTES
* The work reported on in this article is part of a much larger project to investi-
gate the applicability of the Case Hierarchy to natural languages, which I am
carrying out together with E. L. Keenan (cp. Keenan and Comrie, 1972). We hope
to report in greater detail, at a later date, on other aspects of this work, which has
necessarily been condensed in articles for publication. Most of the work on
causatives has been carried out by myself, though in view of our close cooperation
on all aspects of the Case Hierarchy I should like to express my obligation and
gratitude to E. L. Keenan. Preliminary versions of this paper have been presented
to seminars at the universities of Bremen, Cambridge, and Hanover, and at the
October 1973 meeting of the Philological Society; I am grateful to all those who
participated in the ensuing discussions. I should also Iike fo thank my numerous
informants, without whose generosity and patience this paper could never have
been written.
The present paper contains all the theoretical apparatus I have built up as a
result of this work on causatives, though in the fuller version I hope ultimately
to expand considerably the exemplificatory material.
The following abbreviations are used throughout: Subj = Subject, DO=Direct
object, IO=Indirect object, Obl=Other oblique cases, M=Matrix, E=Embed-
ded, MS=Matrix subject, ES=Embedded subject, S=Sentence, V=Verb,
NP=Noun phrase.
1. In the present version I have not included a specificanalysis of French within
the framework I am proposing, although French examples are introduced to
illustrate various points of the discussion in terms of a more generally familiar
language. At many points of the discussion the reader familiar with French will
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 29
inevitably recall parallels between French and the language under discussion. For
recent treatments of the French causative, cp. Kayne (1969), Ruwet (1972:
126-80,254-67), Seuren (1972), and the literature there cited.
2. Perhaps by predicate-raising (McCawley, 1968: 73; Seuren, 1972), though
the particular mechanism used is not at issue here.
3. Where, as here, I establish correlations between (surface) syntactic positions
and morphological marking, it should be borne in mind that I am neither defining
the syntactic positions in terms of morphology, nor identifying surface syntactic
and morphological categories. In practice, morphological markers in many
languages do provide the most readily accessible exponent of syntactic positions,
and are therefore particularly useful in a brief exposition. For some languages
(e.g. Songhai in section 2, Sanskrit in section 3.2) other criteria are used for estab-
lishing syntactic positions, such as word order and susceptibility to passivization.
4. A sentence with two such indirect objects is cited explicitly by Sebiiktekin
(1971 : 82). Below (section 3.1) I discuss languages where doubling on the position
I 0 is possible. If some forms of Turkish do fall into this category, all this means is
that these forms of Turkish do not provide an instance of the ‘paradigm’ case
outlined in section 1.1.
5. At present, I have no real explanation for this generalization. Yet it is the
only one of (i)-(iv) to which I have not been able to find a clear counterexample.
An apparent counterexample is Georgian, which behaves like the paradigm case
until we come to causatives of noncausative verbs that have both direct and
indirect objects, e.g. of:
(i) Mdivan -ma da-cer-a mascavlebel-s ceril -i
secretary Subj wrote teacher I0 letter DO
The secretary wrote a letter to the teacher
Georgian has no possibility of moving ES lower down the Hierarchy than 10.
It is possible to construct the causative to (i) with two indirect objects (cp. (44)
below), but the preferred version is with EIO moved down the Hierarchy to Obl
(in fact, with the postposition -tvis ‘for’), ES moving into the vacated I 0 position
(Gecadzeetal., 1969: 144):
(ii) Mama-m da-a-qer-in-amascavleblis-tvis mdivan -s peril -i
father Subj write-Caus teacher for secretary I0 letter DO
Father made the secretary write a letter to the teacher
However, this is not simply a peculiarity of the Georgian causative. In the perfect
tenses, the subject of a transitive verb is transformed into an indirect object; the
original indirect object may either remain as such (since Georgian marginally
allows two 10s in a simplex sentence) or to be demoted down the Hierarchy to
Obl, with -tois (Vogt, 1971 :163-164):
(iii) Mdivan -s da-u-per-i-amasqavleblis-tvis peril- -i
secretary I 0 has-written teacher for letter DO
The secretary has (apparently) written a letter to the teacher
In Georgian, then, movement of I 0 to Obl to vacate the I0 position is required
quite independently of causative constructions, so the failure of condition (ii)
to hold need never be stated explicitly with specific reference to causative con-
structions. I hope to report in greater detail on the Georgian causative at a later
date.
6. This is interesting, in that Punjabi does have a constraint against doubling
on indirect objects in certain constructions. Direct-object-demotion transforms
30 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
certain direct objects into indirect objects. The rule is normally obligatory for
pronoun direct objects, but does not take place where there is also another in-
direct object:
(i) d n f i ( * 0 ) mez te rukkho!
it I 0 DO table on put
Put it on the table
(ii) Zuru may nri d (*nri) dpo!
just me I 0 it I0 give
Just give it to me
Thus different transformations may be sensitive to different degrees to restrictions
(whether absolute of tendencies) on syntactic doubling.
7. In Sanskrit, as in the older Indo-European languages generally, surface
structure syntactic positions are expressed, basically, by the inflectional endings
of nouns: Subj stands in the nominative, DO in the accusative, I0 in the dative;
the agent in the passive construction is in the instrumental (Obl.), like the instru-
ment in an active sentence. It will be recalled (see footnote (3)) that surface syn-
tactic positions and morphological categories are closely related, but not identical.
In Latin sentence (i) Athenus (accusative) is not a second direct object:
(i) Legutos Athenus rnitto
I send envoys to Athens
just as in (ii) it is not the first direct object:
(ii) Athenus iui
I went to Athens
cp. the impossibility of making Athens subject in the passive:
(iii) *Atheme itue sunt
Often, the possibility of transforming an accusative into a subject under the
passive transformation does provide a test, independent of morphological criteria,
of direct object status. Where a verb takes two direct objects, only one can be
made subject at a time, and the value of the passive test is reduced by the fact that
some languages place quite strict limitations on which direct object this must be
(compare, for Latin, Gildersleeve and Lodge (1895: 216); for Sanskrit, Mac-
donell (1927: 185), and examples (53)-(55) below).
8. There is one quasi-argument, but even this is circular, and seems based on a
dubious interpretation of the definitions of syntactic positions (syntactic relations)
given by Chomsky (1965: 68-73). Park claims that under Chomsky’s definitions,
only one NP per simplex sentence will be defined as subject. Even if this were true,
the argument would be no stronger than Chomsky’s definition of subject. Indeed,
Chomsky speaks usually of ‘the Subject-of the sentence’ (my italics-BC) (and
equally, of ‘the Direct-Object-of the Verb Phrase’ (my italics-BC), but I am
dubious about attaching too much weight to the presuppositions of the definite
article here, especially as Chomsky concedes (1965: 73) that his definitions are
probably too restricted anyway. A natural extension would be to define ‘subject
of a sentence’ as ‘(any) NP immediately dominated by S’, if this type of definition
is retained. The rule assigning the correct morphological marker to subjects will
operate irrespective of how many subjects there are. Both Japanese and Korean
lack subject-verb agreement, i.e. the most obvious example of a transformation
requiring identification of the subject of a simplex sentence.
9. More neutral means of expressing sentences of this type would be to
BERNARD COMRIE - CAUSATIVES AND UNIVERSAL GRAMMAR 31
topicalize on m e of the ga noun phrases. But (57) and (58) are possible in Japa-
nese, a more accurate translation would be something like ‘it is I, and only I, who
like the film’, ‘it is in Japan, and only in Japan, that men have a short life-span’
(Kuno, 1973: 67).
10. It is sometimes argued that Finnish has no category ‘direct object’, since it
has no single morphological category corresponding to direct object, but uses
the nominative, genitive, or partitive. But this scepticism rests on an unnecessary
genuflection to morphological criteria; in fact, it is no more difficult to identify
a class of direct objects in Finnish than in better-known European languages, on
the basis of the full range of morphological possibilities in a given position.
11. Gilyak has cases that Panfilov (1962: 131-2,13942) calls dative-accusative
and dative-directional. But these are not used to express the indirect object. The
dativeaccusative is discussed below ; the dative-directional indicates motion
towards.
REFERENCES
ALLEN,W. S. 1956. ‘Structure and system in the Abaza verbal complex’, in TPS
1956: 127-76.
APTE,V. S . 1898. The student’s guide to Sanskrit composition. 4th edition. Poona.
(Published by Mrs R. A. Sagoon, Bombay.)
BAILEY, T. G. 1956. Teach Yourself Urdu. Edited by J. R. Firth and A. H. Harley.
London: English Universities Press.
CHOMSKY, N. 1965. Aspects of the theory of syntax. Cambridge, Mass.: MIT
Press.
GECADZE, I. 0..NEDJALKOV, V. P., and XOLODOVIE, A. A. 1969. ‘Morfologikskij
kauzativ v gruzinskom jazyke’ (‘The morphological causative in Georgian’),
in A. A. XOLODOVIE (ed.).
GILDERSLEEVE, B. L. and LODGE,G. 1895. Latin grammar, 3rd edition. London:
Macmillan, and New York: St. Martin’s Press.
GOODWIN, W. W. 1894. A Greek grammar. New edition. London: Macmillan,
and New York: St. Martin’s Press.
KAUTZSCH, E. 1910. Gesenius’ Hebrew grammar. 2nd English edition, translated
from the 28th German edition by A. E. Cowley. Oxford: Clarendon Press.
KAYNE,R. S. 1969. The transformational cycle in French syntax. Ph.D. thesis,
MIT, Cambridge, Mass. Revised version to be published as: French syntax.
The transformationalcycle. Cambridge, Mass : MIT Press.
KEENAN, E. L. and COMRIE, B. 1972. Noun phrase accessibility and universal
grammar. Paper read at the Annual Meeting of the LSA.
KUNO,S. 1973. The structure of the Japanese language. Cambridge, Mass.: MIT
Press.
LAKOFF, G. 1970. ZrreguIarityin syntax. New York: Holt, Rinehart and Winston.
LEWIS,G. L. 1967. Turkish Grammar. Oxford: Clarendon Press.
LORENTZ,0. 1973. ‘Predicate raising and word structure in Lappish and in
general’, in Workingpapers in linguistics, University of Oslo, 4.
MCCAWLEY, J. D. 1968. ‘Lexical insertion in a transformational grammar with-
out deep structure’, in B. J. DARDEN, C.-J. BAILEY, A. DAVISON (eds.), Papers
from the Fourth Regional Meeting Chicago Linguistic Society. Chicago, Ill. :
Department of Linguistics, University of Chicago.
MACDONELL, A. A. 1927. A Sanskrit grammar for students. 3rd edition. Oxford:
Oxford University Press.
32 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1974
NEDJALKOV, V. P. and SIL’NICKIJ, G. G. 1969. ‘Tipologija niorfologiEeskogo i
leksieeskogo kauzativov’ (‘Typology of the morphological and lexical causa-
tives’), in A. A. XOLODOVIE (ed.).
PANFILOV, V. Z. 1962. Grammatika nivxskogo jazyka. I. (Grammar of the Nivkh
language I ) . Moscow: Izd-vo AN SSSR.
PANFILOV, V. 2. 1965. Grammatika nivxskogo jazyka. II. Moscow: Nauka.
PARK,B . 4 . 1973. ‘Onthe multiple subject construction in Korean’, in Linguistics
100: 63-76.
RENOU,L. 1961. Grammaire sanscrite. 2nd edition. Paris: Librairie d‘Am6rique
et #Orient Adrien Maisonneuve.
RUWET,N. 1962. Thkorie syntaxique et syntaxe du franpis. Paris: editions du
Seuil.
SCHACHTER, P. and OTANES,F. T. 1972. Tagalog reference grammar. Berkeley
and Los Angeles: University of California Press.
SEBUKTEKIN, H. I. 1971. Turkish-English contrastive analysis. Morphology and
corresponding English structures. The Hague: Mouton.
SEUREN,P. A. M. 1972. ‘Predicate raising and dative in French and sundry
languages.’ Unpublished, Magdalen College, Oxford.
SHIBATANI, M. 1973a. ‘Lexical versus periphrastic causatives in Korean’, in JL 9:
281-97.
SHIBATANI, M. 1973b. ‘Semantics of Japanese causativization’ in FoL 9: 327-73.
SHOPEN, T. and KONARE,M. 1970. ‘Sonrai causatives and passives: transforma-
tional versus lexical derivations for propositional heads’, in Studies in African
Linguistics 1 : 21 1-54.
TITOV, E. G. 1971. Sovremennyj amxarskij jazyk. (The modern Amharic language).
Moscow: Nauka.
VOGT,H. 1971. Grammaire de la langue gkorgienne. Oslo: Universitetsforlaget.
VOLODIN,A. P., XOLODOVIE, A. A., and XRAKOVSKU, V. S. 1969. ‘Kauzativy,
antikauzativy i kauzativnye konstrukcii v finskom jazyke’ (‘Causatives, anti-
causatives, and causative constructions in Finnish’), in A. A. XOLODOVIC (ed.).
WRIGHT,W. 1898. A grammar of the Arabic language. Volume 2. Translated from
the German of Caspari and edited with numerous additions. Third edition.
Cambridge: Cambridge University Press.
XOLODOVIE, A. A. (ed.). 1969. Tipologca kauzativnyx konstrukcij. MorfologiCeskij
kauzativ (Typology of causative constructions. The morphological causative).
Leningrad: Nauka.