ESZTER ÓTOTT-KOVÁCS

The syntax of non-finite clauses in Kazakh

PhD dissertation summary

Supervisors:
DR. ÉVA KINCSES NAGY
PROF. DR. ANDRÁS RÓNA-TAS

University of Szeged
Graduate School in Linguistics
Altaic Studies Programme
Szeged, 2015.
The dissertation deals with the syntax of Kazakh non-finite clauses. Kazakh belongs to the
South Kipchak branch of Turkic languages; it is manly spoken in the Republic of Kazakhstan
and in the neighbouring countries such as Uzbekistan, China, Mongolia, Russia.
Kazakh syntax, and especially the syntax of non-finite clauses, is an unresearched
field: grammars usually neglect this area and there are very few insightful linguistic papers on
the topic. There are no detailed descriptive works available about Kazakh non-finites, let
alone theoretical ones. This dissertation intends to be the first in-depth contribution to this
topic offering not only descriptive observations, but also an explanatory (i.e. theoretical)
account on Kazakh non-finite clauses. It is noteworthy that I couch my theoretical analysis in
the Minimalist Program framework, but I also use Distributed Morphology.
Since there are no detailed descriptive works concerning Kazakh non-finite clauses, I
had to begin my research with collecting and organizing the relevant Kazakh language
material. I compiled a corpus from Kazakh literary works (fairy tales, short stories),
traditional paper-based or online newspaper articles. Moreover, I put two questionnaires
together, which I asked Kazakh native speakers to fill out during my stay in Kazakhstan
(2013. May-June, 2014. May, 2014. December). The results of the questionnaires provided
useful material not only for the descriptive but for the theoretical analysis, too.
The dissertation has five chapters, from which the first one is the Introduction and the
last one is the Conclusions. The remaining three chapters address crucial issues of all Kazakh
non-finite clause types, which I claim to be converb, nominalized and Inflectional non-finite
clauses. The results of these chapters are summarized in the following.

1. The syntactic position of non-finite clause heads

First of all, I determined the syntactic position of non-finite heads. For this, I relied on the
syntactic behaviour of non-finite heads with respect to constructions I called high light verbs,
and which are traditionally called auxiliaries. Therefore, first, I needed to take a closer look at
high light verbs. High light verbs are constructions that follow the verb stem but precede
Inflectional and Tense markers, and express meanings such as Benefactivity, Completiveness,
Manner or Continuousness.
To my knowledge, I was the first to show that the groups of high light verbs are
strictly ordered: Benefactives must be followed by Completives, which must precede Manner

2
high light verbs, and the Continuous comes after all of them. (Naturally, not all of these high
light verbs are present all the time.) This is summarized in table (1).

(1) The order of the groups of high light verbs

Benefactive Completive Manner Continuous
main -(I)p al- -(I)p qal- -y/A qoy- -(I)p žat-
verb -(I)p ber- -(I)p žiber- -y/A sal- -(I)p žür-
-(I)p tasta- -(I)p otïr-
-(I)p ket- -(I)p tur-

After establishing this, it was shown that not all non-finite heads can embed high light verb
phrases. Based on the results of Questionnaire 1, it was demonstrated that converb heads -y/A,
which expresses manner, and -(I)p when used to head manner clauses cannot follow high light
verbs. In contrast, other non-finite heads can embed high light verb phrases, therefore I
concluded that they are situated higher up in the structure, as shown in the following tables.

(2) The structure of Inflectional non-finite clauses

Verb Benefactive Completive Manner Continuous Inflection
-(I)p al- -(I)p qal- -y/A qoy- -(I)p žat- I n f l e c t io na l n o n-
-(I)p ber- -(I)p žiber- -y/A sal- -(I)p žür- f i n it e s :
-(I)p tasta- -(I)p otïr- -ĠAn, -y/AtIn, -(A)r
-(I)p ket- -(I)p tur- C o nve r bs :
-(I)p, -MAy, -ĠAlI,
-ĠAsIn, -MAyInšA

(3) The structure of nominalized clauses

Verb Benefactive Completive Manner Continuous Nominalizer
-(I)p al- -(I)p qal- -y/A qoy- -(I)p žat- -w
-(I)p ber- -(I)p žiber- -y/A sal- -(I)p žür- -(I)s
-(I)p tasta- -(I)p otïr-
-(I)p ket- -(I)p tur-

3
Thus it was shown that, not surprisingly, Kazakh non-finite clauses do not form a uniform
class syntactically: there is group of clauses (the manner expressing converbs -y/A and -(I)p)
which is more truncated than other non-finite clauses.

2. Subjects of non-finite clauses

Moreover, I determined which non-finite clauses can have an independent subject (i.e.
independent from the subject of the matrix clause), and then I explained why certain non-
finite clauses can have an independent subject, while others cannot.
First of all, I collected which non-finite clauses can have an overt independent subject,
and which cannot. This is given in table (4).

(4) Independent subjects in non-finite clauses

Can the clause have an Non-finite heads Type (and meaning if
overt independent subject? relevant)
yes -(I)p Converb (‘after’, ‘when’;
(when it expresses manner: ‘-ing’ (manner); ‘and’; ‘as’;
no) ‘since’, etc.)
no -y/A Converb (‘-ing’ (manner))
yes -MAy Converb (Negative
allomorph of -(I)p and -y/A;
‘without’; ‘until’)
yes -ĠAlI Converb (‘since’; ‘in order
to’)
yes -ĠAsIn Converb (‘when’; ‘because’)
yes -MAyInšA Converb (‘unless’, ‘until’)
yes -w Nominalizer
yes -(I)s Nominalizer
yes -ĠAn Inflectional non-finite
yes -y/AtIn Inflectional non-finite
yes -(A)r Inflectional non-finite

4
We found a striking correlation between the availability of an overt independent subject and
the degree of truncation: notice that only those clauses cannot have an independent subject
(i.e. manner expressing -y/A and -(I)p) whose head cannot embed high light verb phrases, i.e.
these are the clauses which are more truncated. Moreover, I suggested, in line with the
Minimalist Program approach, that the syntactic position of clause-heads and their capability
to licence subject case are connected. That is, in Kazakh only those clauses can have
independent subjects whose (clausal) head is in the Inflection slot, or whose head is a
nominalizer. If the head of the clause is located below the Inflection position, the clause
cannot have an independent subject, since that subject could not be licensed subject case.
With this approach, we managed to explain w h y certain non-finite clauses do not have
independent subjects.

3. A curious case: -(I)p-clauses

Two important features of -(I)p-headed “converb” clauses are mentioned in the literature,
which, I think, characterize only these non-finite clauses. The first is the abundance of
meanings. -(I)p-clauses can express: manner, purpose (with motion verbs), cause, a linking
relation (when the -(I)p-marked element seems to be independent of the main clause), and
temporal relation (perfect or imperfect). The second is the scope-over phenomenon, that is,
the scope of the matrix clause’s functional categories may extend over the -(I)p-clause as
well, which is quite odd if we assume that -(I)p-clauses are subordinated.
I argued that the explanation for these phenomena is that there are (at least) four
different syntactic constructions that -(I)p can realize: low subordination (in manner
constructions), high subordination, low coordination and high coordination. The claim that
Kazakh -(I)p can mark coordinated clauses too is novel in the literature. I used several
diagnostics to tease apart coordinative and subordinate -(I)p-clauses. First, I showed that
scope-over can only happen in coordinated clauses, but not in subordinate clauses. Thus the
only way to account for the fact that scope-over does happen in -(I)p-clauses is to assume that
(at least some types of) -(I)p-clauses are coordinated. An illustrative example is offered in (5),
where the scope of negation and the Inflection marker -y/AtIn, which is only indicated in the
matrix clause, scopes over the -(I)p-clause (töbeles-ip ‘fight-(I)p’) as well.

5
(5) [...] Ең жақсы қасиет-і –
SUPL good quality-POSS.3
[[ешкім-мен [төбелес-іп], сөз-ге кел]-ме-йтін-і еді].
[[nobody-INSTR [fight-CV] word-DAT come]-NEG-NF-POSS.3] COP.PAST.3
‘His/Her best quality was that (s)he wasn’t such who would fight or argue with
anyone.’

Moreover, I showed that some -(I)p-clauses pattern with coordinated clauses with
respect to other diagnostics as well. For instance, symmetrical operations (such as
symmetrical questions) are grammatical in them, while asymmetric operations are not
grammatical. This pattern is typical to coordinated clauses, and not to subordinate clauses.
The grammatical example (6) illustrates the case when the question formation is applied
symmetrically in both the -(I)p and the other clause. On the other hand, if only one of the
clauses includes a question (i.e. if it is asymmetrical), the result is an ungrammatical sentence,
as in (7).

(6) Кеше мейрамхана-да [Асқар кім-мен төбелес-іп],

yesterday restaurant-LOC [Asqar who-INSTR fight-CV]
[Болат кім-мен сөз-ге кел-ген]?
[Bolat who-INSTR word-DAT come-PERF.3]
‘Yesterday at the restaurant who did Askar have a fight with, and who did Bolat argue
with?’

(7) *Кеше мейрамхана-да [Асқар төбелес-іп],

yesterday restaurant-LOC [Asqar fight-CV]
Болат кім-мен сөз-ге кел-ген?
Bolat who-INSTR word-DAT come-PERF.3
~‘Yesterday at the restaurant Askar had a fight, and who did Bolat argue with?’

These facts clearly indicate that some -(I)p-clauses are coordinated. But crucially, not all of
them: the dissertation showed that other types of -(I)p-headed clauses pattern like
subordinated clauses.

6
 Therefore, I proposed that there are (at least) four different underlying syntactic
structures that are headed by -(I)p. Moreover, I claimed that this wide range of usage can only
be explained if we assume that -(I)p is an underspecified vocabulary item, which can realize
more than one syntactic configuration.

4. Nominalized and non-nominalized non-finite clauses

In Chapter 4, I turned to non-converbial non-finite clauses. These clauses can appear as

argument clauses, as complement clauses of postpositions/semantic cases, and (some of them)
as relative clauses. First, I proposed that these clauses have two subtypes: nominalized and
not nominalized, which I named “Inflectional non-finite”, clauses; the table in (8) shows this
classification. (It is noteworthy that my analysis of Kazakh was greatly influenced by
Kornfilt’s (2001a, 2003, 2006, 2007) works on similar Turkish non-finite clauses.)

(8) Heads of nominalized and non-nominalized non-finite clauses (first version)

Nominalized clauses -w, -(I)s, (-MAq)
(Non-nominalized) Inflectional non-finites -ĠAn, -y/AtIn, -(A)r

The syntactic behaviour of these clauses supports this classification. For instance, if an -w-
headed nominalized clause modifies a noun head, as shown in (9), the compound marker -(s)I
appears on the modified noun head, which is exactly the same pattern that can be observed in
case of nominal compounds (cf. (10)). However, when Inflectional non-finites modify a noun
phrase, the compound marker is absent (cf. (11)), indicating that these non-finite constructions
are not nominalized. In Chapter 4, other criteria were offered, too, supporting the
classification in (8).

(9) [«Коста Конкордиа» кеме-сін көтер-у] жұмыс-тар-ы

[Kosta Konkordįa ship-CM.ACC raise-NNF] work-PL-CM
аяқтал-ды.
finish(intr)-PAST.3
‘The works of lifting the ship Costa Concordia have come to an end.’

(10) үй жұмыс-ы

7
 house work-CM
‘housework’

(11) Маған [демалыс кун-дер-і істе-йтін] жұмыс керек еді.

I.DAT [rest day-PL-CM do-NF] work necessary COP.PAST.3
‘I would need a job that can be done on the weekends.’

Moreover, I needed to account for the fact that some -ĠAn (-y/AtIn and -(A)r)-clauses,
which I claimed to be non-nominalized, can appear in typical nominal positions (such as in
argument position or as complements of certain semantic cases/postpositions) with nominal
agreement (i.e. the possessive) marked on their predicates. This seemingly contradicts the
classification in (8). However, it was demonstrated in Chapter 4 that the originally non-
nominalized -ĠAn (-y/AtIn and -(A)r)-clauses can get nominalized, which allows them to
appear in the above mentioned “typical” nominal positions. In Kazakh the suffix -LIq can
appear following (certain types of) -ĠAn (-y/AtIn and -(A)r)-clauses; an illustrative example is
given in (12).

(12) Өмірбек [өткен апта Алматы-да бол-ған-дығ-ын] айт-ты.

Ömirbek [last week Almatï-LOC COP-NF-LIq-POSS.3]ACC say-PAST.3
‘Ömirbek said that he was in Almaty last week.’

I showed that -LIq is nominalizer that turns the originally non-nominalized Infinitival non-
finites into nominalized clauses. More than 60 sentences of Questionnaire 2 were concerned
with the possibility (and details) of -LIq-attachment. The results are summarized in the
following table.

8
(13) Possibility of -LIq-attachment
Function of the -ĠAn/-y/AtIn/-(A)r-headed Can -LIq follow the clause?
clause
Relative clause no
Argument clause yes
Complement of a semantic case/postposition no
(α)1
Complement of a semantic case/postposition yes
(β)2
Complement of a semantic case/postposition ?
(γ)3

These results indicate that -LIq can only show up following clauses in typical nominal
positions (cf. the ungrammaticality of -LIq-attachment after relative clauses, but the
acceptability after argument clauses; the latter being a typical nominal position).
A further argument formulated in Chapter 4 was that nominal agreement marking can
only be marked on predicates of nominalized clauses, but not on non-nominalized clauses.
First, I provided a detailed dataset of agreement marking patterns in non-finite clauses based
on the Kazakh corpus I compiled and Questionnaire 2, which will hopefully be useful for
Kazakh descriptive linguistics, too. Moreover, interesting correspondences can be discovered
between nominalization and the grammaticality of nominal agreement marking (cf. the table
in (14)).

1
Complement clauses of the following postpositions belong to this group: -(ABL) soŋ ‘after’, sayïn ‘every’.
2
Complement clauses of the following postpositions/semantic cases belong to this group: twralï ‘about’, üšin
‘for; in order to’, -Men (INSTR semantic case)
3
Complement clauses of the remaining postpositions/semantic cases belong to this group, such as locative,
dative, ablative, -šA, -DAy, -ABL keyin ‘after’, etc.
9
(14) Agreement marking patterns in non-finite clauses
Non-finite clause type Can the nominal agreement marking be
present on the non-finite predicate?
Converb clauses no
(nominalized) -w-clauses yes
(nominalized) -(I)s-clauses yes
Relative clauses (headed by -ĠAn/-y/AtIn/ no
-(A)r)
Argument clauses (headed by -ĠAn/-y/AtIn/ yes
-(A)r)
Complement clauses of a semantic no
case/postposition (α)
Complement clauses of a semantic yes
case/postposition (β)
Complement clauses of a semantic (preferred:) no
case/postposition (γ)

The nominal agreement can be marked on predicates of the clauses that I analyzed as
nominalized non-finites (cf. -w, -(I)s, and Argument -ĠAn/ -y/AtIn/ -(A)r-clauses); but it
cannot be indicated in converb clauses or on the predicate of relative clauses, which are non-
nominalized clauses. -ĠAn, -y/AtIn and -(A)r-headed complement clauses of semantic cases/
postpositions have three subtypes, indicated by the notations α, β and γ. If we compare the
relevant parts of tables (13) and (14), we will see that in those complement clauses which -LIq
can be attached to, and which are consequently nominalized, the nominal agreement marking
can be indicated (cf. β). On the other hand, those complement clauses where -LIq cannot
follow the agreement cannot be marked either (cf. α). (Note that in group γ the agreement
marking, like the -LIq-attachment, is not preferred.) Therefore, there is a clear correlation
between nominalization and agreement marking in Kazakh non-finite clauses.
To sum up, I argued in Chapter 4 that there are two types of non-converbial non-finite
clauses in Kazakh: nominalized and non-nominalized non-finites, as shown in (15), which is a
revised version of (8). -ĠAn, -y/AtIn and -(A)r-headed clauses are non-nominalized, but they
can get nominalized through the attachment of -LIq. Moreover, I demonstrated that the
agreement marking can only be indicated on nominalized non-finites.

10
(15) Heads of nominalized and non-nominalized non-finite clauses (final version)
Nominalized clauses -w, -(I)s, (-MAq); -ĠAn(dIq), -y/AtIn(dIq),
-(A)r(lIq)
(Non-nominalized) Inflectional non-finites -ĠAn, -y/AtIn, -(A)r

11
Selected references

Abish, Aynur (2014): Modality in Kazakh as spoken in China. Doctoral dissertation. Uppsala
Universitet. Uppsala.
Adger, David (2003): Core syntax. A minimalist approach. Oxford.
Akbaba, Dilek Ergönenç (2011): Kazak ve Nogay Türkçesi Yazı Dillerinde Tasvir Fiilleri.
Ankara.
Alexiadou, Artemis & Iordăchioaia, Gianina & Schäfer, Florian (2011): Scaling the variation
in Romance and Germanic nominalizations. In. Sleeman, P. & Perridon, H. (eds.) The
Noun Phrase in Romance and Germanic. Structure, variation, and change.
Amsterdam/Philadelphia. 25-40.
Balakaev, Mäulen Balaqayulï (1959): Sovremennyj kazahskij âzyk. Sintaksis slovosočetaniâ i
prostogo predloženiâ. Alma-Ata.
Balaqaev, Mäulen & Isqaqov, Ahmet (1954): Qazirgi qazaq tili. Leksika, fonetika,
grammatika. Almatï.
Biacsi, Mónika & Mukuseva, Raushangul (selected, translated) (2011): A kán és a vezír.
Kazak népmesék. Szeged.
Butt, Miriam & Ramchand, Gillian (2005): Complex Aspectual Structure in Hindi/Urdu. In.
Erteschik-Shir, N. & Rapoport, T. (eds.) The Syntax of Aspect. Oxford. 117-153.
Foley, William A. (2010): Clause linkage and Nexus in Papuan languages. In. Bril, I. (ed.)
Clause Linking and Clause Hierarchy. Syntax and pragmatics. Amsterdam. 27-50.
Haspelmath, Martin (2007): Coordination. In. Shopen, T. (ed.) Language typology and
linguistic description. Second edition. Volume II: Complex Constructions. Cambridge. 1-
51.
Huddleston, Rodney & Pullum, Geoffrey K. (2006): Coordination and Subordination. In.
Aarts, B. & McMahon, A. (eds.) The Handbook of English Linguistics. Oxford. 198-219.
Johanson, Lars & Csató, Éva Á. (eds.) (1998): The Turkic Languages. London–New York.
Koç et al. 2003 = Koç, Kenan & Bayniyazov, Ayabek & Başkapan, Vehbi (2003): Kazak
Türkçesi Türkiye Türkçesi Sözlüğü. Ankara.
Koç, Kenan & Doğan, Oğuz (2004): Kazak Türkçesi Grameri. İstanbul.
Kornfilt, Jaklin (2001a): On the syntax and morphology of clausal complements and adjuncts
in the Turkic languages. In. Bisang, W. (ed.) Aspects of Typology and Universals. Berlin
63-82.

12
Kornfilt, Jaklin (2001b): Functional projections and their subjects in Turkish clauses. In.
Erguvanlı Taylan, E. (ed.) The Verb in Turkish. Amsterdam–Philadelphia. 183-212.
Kornfilt, Jaklin (2003): Subject Case in Turkish nominalized clauses. In. Junghanns, U. &
Szucsich, L. (eds.) Syntactic Structures and Morphological Information. Berlin–New
York. 129-216.
Kornfilt, Jaklin (2004): Remarks on Complex Predicates in a Comparative Altaic Perspective.
In. Csirmaz, A. & Lee, Y. & Walter, M.A. (eds.) Proceedings of WAFL 1: Workshop in
Altaic Formal Linguistics. MIT Working Papers in Linguistics. Volume 46. 1-14.
Kornfilt, Jaklin (2005): Agreement and its placement in Turkic Nonsubject Relative clauses.
In. Cinque, G. & Kayne, R. S. (eds.) The Oxford Handbook of Comparative Syntax.
Oxford. 513-541.
Kornfilt, Jaklin (2006): Agreement: The (unique and local) syntactic and morphological
licenser of the subject Case. In. Costa, J. & Silva, M.C.F. (eds.) Studies on Agreement.
Amsterdam–Philadelphia. 141-171.
Kornfilt, Jaklin (2007): Verbal and nominalized finite clauses in Turkish. In. Nikolaeva, Irina
(ed.) Finiteness. Theoretical and Empirical Foundations. Oxford – New York. 305-332.
Kornfilt, Jaklin (2008): DOM and Two Types of DSM in Turkish. In. de Hoop, H. & de
Swart, P. (eds.) Differential Subject Marking. Dordrecht. 79-111.
Kornfilt, Jaklin (2009): Subject-Agreement correlations and their syntactic effects in some
Turkic relative clauses. Turkic Languages Vol. 13. 70-96.
Meral, Hasan Mesut (2012): Kazakh complex verb structures: A Distributed Morphology
analysis. Turkic Languages Vol. 16. 239-256.
Oralbaeva, Nuržamal (1979): Qazaq tilindegi etistiktiŋ analitikalïq formanttarïnïŋ qurïlïsï
men maġïnasï. Almatï.
Straughn, Christopher A. (2011): Evidentiality in Uzbek and Kazakh. Doctoral dissertation.
University of Chicago, Chicago.
Tanç, Mustafa (2002): Kazak Türkçesinde Zarf-Fiiller -Muhtar Evezov’un Tandamalı
Engimeler Adlı Eseri Üzerinde Bir Çalışma- (Şekil-Anlam-İşlev ve Türkiye Türkçesindeki
Karşılıkları). Doctoral Dissertation. Çukorova Üniversitesi, Sosyal Bilimler Enstitüsü,
Türk Dili ve Edebiyatı Ana Bilim Dalı, Adana.
Žanpeyisov, Erbol (et al.) (2002): Qazaq Grammatikasï. Fonetika, Sözžasam, Morfologiya,
Sintaksis. Astana.

13