ThesisPDF Available

Auxiliary verb constructions in Modern Spoken Kazakh

Authors:

Abstract

The Turkic language Kazakh has a remarkably large set of eighteen auxiliary verbs, which express various tense-aspect-modality (TAM) values in combination with a lexical verb. This thesis presents the first multivariate analysis of auxiliary verb constructions (AVCs) to precisely define the function and distribution of every auxiliary in Modern Spoken Kazakh. Based on corpus data and first-hand fieldwork, I demonstrate that contemporary Kazakh has 28 AVCs, each with a distinctive function. I argue that some morphologically or syntactically ambiguous constructions, hitherto analyzed as AVCs, should be treated as lexical verbs in separate clauses. Previ-ously undescribed AVCs are described, including an unexpected case of stem alternation. The distribu-tional analyses demonstrate that some AVCs are sensitive to syntactic parameters, such as the independ-ence of the clause they head. Others are expressions of a purely semantic feature and thus are insensitive to syntax, such as the modal abilitative. Thus, AVCs are grouped into six classes in order to contrast distinctive distributional behavior with characteristics shared across constructions. The analysis assumes that auxiliaries are periphrastic and thus are part of the lexical verb’s paradigm. Therefore, alongside AVCs, synthetic TAM expressions are investigated and the results include precise descriptions and a novel contrastive analysis of three past tense expressions. The description is complemented by an HPSG style analysis in order to present the system using a rigorous, feature-based approach. This is the first attempt to formalize a large auxiliary system with implemented solutions that lay the grounds for future work on the diachrony of auxiliaries. I propose novel semantic features that account for distinctions including boundedness, phase specification and focus. The main contribution is a systematic, synchronic, fine-grain examination of every Kazakh aux-iliary verb, which makes this complex system available for the general linguist, as well as specialists of TAM and periphrasis.
Auxiliary verb constructions in
Modern Spoken Kazakh
by
Dávid Győrfi
Submitted for the Degree of Doctor of Philosophy
University of Surrey
Faculty of Arts and Social Sciences
School of Literature and Languages
Supervisors:
Dr. Oliver Bond
Prof. Greville G. Corbett
©Dávid Győrfi 2022
2
Declaration of originality
This thesis and the work to which it refers are the results of my own efforts. Any ideas, data,
images or text resulting from the work of others (whether published or unpublished) are fully identified
as such within the work and attributed to their originator in the text, bibliography or in footnotes. This
thesis has not been submitted in whole or in part for any other academic degree or professional qualifi-
cation. I agree that the University has the right to submit my work to the plagiarism detection service
TurnitinUK for originality checks. Whether or not drafts have been so assessed, the University reserves
the right to require an electronic version of the final document (as submitted) for assessment as above.
Dávid Győrfi, University of Surrey, Guildford, 3rd of March 2022
3
Abstract
The Turkic language Kazakh has a remarkably large set of eighteen auxiliary verbs, which express
various tense-aspect-modality (TAM) values in combination with a lexical verb. This thesis presents the
first multivariate analysis of auxiliary verb constructions (AVCs) to precisely define the function and
distribution of every auxiliary in Modern Spoken Kazakh.
Based on corpus data and first-hand fieldwork, I demonstrate that contemporary Kazakh has 28
AVCs, each with a distinctive function. I argue that some morphologically or syntactically ambiguous
constructions, hitherto analyzed as AVCs, should be treated as lexical verbs in separate clauses. Previ-
ously undescribed AVCs are described, including an unexpected case of stem alternation. The distribu-
tional analyses demonstrate that some AVCs are sensitive to syntactic parameters, such as the independ-
ence of the clause they head. Others are expressions of a purely semantic feature and thus are insensitive
to syntax, such as the modal abilitative. Thus, AVCs are grouped into six classes in order to contrast
distinctive distributional behavior with characteristics shared across constructions. The analysis assumes
that auxiliaries are periphrastic and thus are part of the lexical verb’s paradigm. Therefore, alongside
AVCs, synthetic TAM expressions are investigated and the results include precise descriptions and a
novel contrastive analysis of three past tense expressions.
The description is complemented by an HPSG style analysis in order to present the system using
a rigorous, feature-based approach. This is the first attempt to formalize a large auxiliary system with
implemented solutions that lay the grounds for future work on the diachrony of auxiliaries. I propose
novel semantic features that account for distinctions including boundedness, phase specification and
focus. The main contribution is a systematic, synchronic, fine-grain examination of every Kazakh aux-
iliary verb, which makes this complex system available for the general linguist, as well as specialists of
TAM and periphrasis.
4
Acknowledgments
This work could not have happened without so many people, events, and maybe a bit of luck.
I will follow some chronology and I apologize for anyone I did not include it’s all my bad.
First, I thank my father, without whom one of my first memories would not have been hearing
him greet his friend and colleague, in Arabic. A memory of true awe. I also thank him for teaching me
the first word I could write in Arabic, and maybe the only one he could [baːbaː] ‘father’.
With a bit of a leap, I thank Prof. István Vásáry, my supervisor at ELTE University, for all his
help with Kipchak languages and the many friendly conversations. Similarly, I am grateful to Prof.
Dunstan Brown for supervising my dissertation at York, my first work on Kazakh.
Most importantly, I thank my supervisors at Surrey, Dr. Oliver Bond, and Prof. Greville G. Cor-
bett for their devoted guidance and help throughout this work. They always found time for small and
big questions and issues I wanted to discuss. Some of the big issues took entire afternoons to sort out.
From Surrey Morphology Group I thank everybody for their kindness, help, and conversations.
A huge that you is due to Dr. Marina Chumakina, Dr. Masha Kyuseva, Dr. Steven Kaye, Dr. Mike
Franjieh, and Dr. Borja Herce for reading chapters of this thesis and commenting on the bits that could
use some additional attention. I especially thank Dr. Borja Herce and Dr. Jérémy Pasquereau for being
good friends, when it comes to both business and leisure. I thank Dr. Steven Kaye in particular for
helping me with hundreds of small and big issues, and for reading the most crucial pieces to make sure
they make some sense in English. The help, time, and effort of Prof. Olivier Bonami and Dr. hab.
Berthold Crysmann enormously increased the quality of the HPSG analysis, for which I am very grate-
ful. I would also like to thank my examiners for their insights and for making sure that the viva was a
great experience. The comments of Prof. Irina Nikolaeva and Dr. Matthew Baerman have made this
thesis a more cohesive, better piece of work.
The Kazakh data has been validated by hard-working consultants, most notably Aqmerey Sei-
dikhanova, Aytolqïn Abdigali, Toɣan Turɣanbayeva, and Ulan Rojik, whom I truly admire for their
patience and devotion Көп рақмет сендерге!
I am indebted to the University of Surrey (Faculty of Arts and Social Sciences) and TECHNE for
supporting my work by providing both the tuition fees and the stipend. Without their help, I would be
much slimmer. I am grateful to TECHNE for their additional funding for fieldwork.
Last but not least, I am grateful to my amazing family for all their support throughout these years.
Thank you, Madzio and Kazimír!
5
Table of contents
List of figures .......................................................................................................................... 10
List of tables ............................................................................................................................ 10
Abbreviations .......................................................................................................................... 12
Transcription conventions ....................................................................................................... 13
Chapter 1: Introduction ........................................................................................................... 14
1.1 Motivation and the aim of this study ............................................................................. 14
1.2 The Turkic languages and Kazakh ................................................................................ 15
1.3 Data collection and sources ........................................................................................... 17
1.3.1 Fieldwork methodology......................................................................................... 17
1.3.2 Corpus data ............................................................................................................ 18
1.4 Introduction to frameworks of TAM ............................................................................. 18
1.4.1 Lexical aspect ........................................................................................................ 19
1.4.2 Grammatical aspect ............................................................................................... 22
1.4.3 The neo-Reichenbachian framework and its limits ............................................... 24
1.4.4 Lars Johanson’s framework on tense and aspect ................................................... 29
1.4.5 Modality ................................................................................................................ 31
1.5 Kazakh TAM expressions: Morphology and Syntax .................................................... 32
1.5.1 Phonological properties of suffixes ....................................................................... 32
1.5.2 TAM suffixes ........................................................................................................ 34
1.5.3 Auxiliary Verb Constructions from a cross-linguistic perspective ........................ 36
1.5.4 Auxiliary Verb Constructions in Kazakh .............................................................. 39
1.5.5 Converb A ............................................................................................................. 44
1.5.6 Converb B .............................................................................................................. 46
1.5.7 Negated converb and negation in AVCs ............................................................... 49
1.5.8 The ambiguity between AVCs and dependent clauses .......................................... 52
1.5.9 Stacked AVCs ....................................................................................................... 54
1.5.10 AVCs as periphrasis .............................................................................................. 56
6
1.6 The structure of this thesis ............................................................................................. 60
Chapter 2: Synthetic TAM expressions .................................................................................. 63
2.1 Past tense -DI ................................................................................................................. 63
2.2 Nonpast tense -y/A ......................................................................................................... 66
2.3 Progressive -UwdA ........................................................................................................ 68
2.4 Anterior and prospective participles -GAn and -(y/A)tIn ............................................... 69
2.5 -GAn as an independent expression ............................................................................... 79
2.6 -GAn in combination with the existential verb e- .......................................................... 82
2.7 -(y/A)tIn as an independent expression .......................................................................... 88
2.8 Aorist -(A)r .................................................................................................................... 90
2.9 Mirative -(I)p ................................................................................................................. 93
2.10 Intentional modality -MAQ or -MAQI ..................................................................... 96
2.11 Conditional -sA .......................................................................................................... 98
2.12 Imperative .................................................................................................................. 99
2.13 Summary ................................................................................................................. 100
Chapter 3: Imperfective AVCs .............................................................................................. 101
3.1 Introduction ................................................................................................................. 101
3.2 Idiosyncratic properties of the four imperfective auxiliary lexemes ........................... 102
3.3 Copula function of the four imperfective auxiliaries ................................................... 105
3.4 at ‘lie’ ........................................................................................................................ 107
3.4.1 The progressive and iterative uses of at ............................................................. 107
3.4.2 The frequentative at ........................................................................................... 111
3.4.3 Pluractionality of the four motion verbs in converb B + at ................................ 114
3.4.4 Resultative and pluractional readings (lexical verb: ‘come’) .............................. 118
3.4.5 Resultative and pluractional readings (lexical verbs: ‘bring’ and ‘take away’) .. 123
3.4.6 High hypotheticality syntax and semantics .................................................... 127
3.4.7 High hypotheticality a generalization ............................................................. 132
3.4.8 atïr is it a new stem? ...................................................................................... 136
3.5 Otïr ‘sit’ ....................................................................................................................... 141
7
3.5.1 Independent clauses ............................................................................................. 141
3.5.2 Dependent clauses ............................................................................................... 145
3.6 Tur ‘stand’ ................................................................................................................... 149
3.6.1 Converb B + tur ‘stand’ ....................................................................................... 150
3.6.2 Converb A + tur ‘stand’ ...................................................................................... 154
3.6.3 Dependent clauses ............................................................................................... 155
3.7 ür ‘walk’ .................................................................................................................... 156
3.7.1 Converb B + ür ‘walk’ ....................................................................................... 156
3.7.2 Converb A + ür ‘walk’ ....................................................................................... 162
3.8 Summary ..................................................................................................................... 163
Chapter 4: Perfective AVCs .................................................................................................. 167
4.1 Introduction ................................................................................................................. 167
4.2 Ket ‘leave’ ................................................................................................................... 167
4.2.1 Converb A + ket ‘leave’ ...................................................................................... 168
4.2.2 Converb B + ket ‘leave’ ....................................................................................... 169
4.3 Converb B + iber ‘send’ ............................................................................................ 175
4.4 Converb B + bol ‘copula’ ............................................................................................ 179
4.5 Converb B + ïq ‘exit’ ................................................................................................. 188
4.6 Qoy ‘place, put’ ........................................................................................................... 192
4.6.1 Converb A + qoy ‘place, put’ .............................................................................. 192
4.6.2 Converb B + qoy ‘place, put’ .............................................................................. 193
4.7 Converb A + azda ............................................................................................... 201
4.8 Comparison of perfective AVCs: a distributional analysis ......................................... 203
4.8.1 iber ‘send’ vs ïq ‘exit’ ...................................................................................... 204
4.8.2 iber ‘send’ vs ket ‘leave’ .................................................................................... 204
4.8.3 iber ‘send’ vs bol ‘copula’ ................................................................................. 205
4.8.4 iber ‘send’ vs qoy ‘put, place’ ............................................................................ 207
4.8.5 ket ‘leave’ vs qal ‘stay’ ........................................................................................ 208
4.8.6 ïq ‘exit’ vs ket ‘leave’ ........................................................................................ 210
8
4.8.7 bol ‘copula’ vs ïq ‘exit’ ...................................................................................... 211
4.8.8 ïq ‘exit’ vs qoy ‘place, put’ ................................................................................ 211
4.8.9 qoy ‘place, put’ vs qal ‘stay’ ............................................................................... 212
4.8.10 qoy ‘place, put’ vs ket ‘leave’.............................................................................. 214
4.8.11 ket ‘leave’ vs bol ‘copula’ ................................................................................... 216
4.8.12 bol ‘copula’ vs qoy ‘place, put’ ........................................................................... 217
4.8.13 qal ‘stay’ vs ïq ‘exit’ .......................................................................................... 218
4.8.14 qal ‘stay’ vs bol ‘copula’ ..................................................................................... 219
4.9 Summary ..................................................................................................................... 220
Chapter 5: Interval and Manner AVCs ................................................................................. 221
5.1 Introduction - Interval AVCs....................................................................................... 221
5.2 Converb A + tüs ‘fall’ ................................................................................................. 222
5.3 Kel ‘come’ ................................................................................................................... 226
5.3.1 Converb A + kel ‘come’ ...................................................................................... 226
5.3.2 Converb B + kel ‘come’ ...................................................................................... 229
5.4 Qal ‘stay’ ..................................................................................................................... 234
5.4.1 Converb A + qal ‘stay’ ........................................................................................ 235
5.4.2 Converb B + qal ‘stay’ ........................................................................................ 238
5.5 Introduction - Manner AVCs....................................................................................... 242
5.6 Sal ‘put’ ....................................................................................................................... 242
5.6.1 Converb A + sal ‘put’ .......................................................................................... 243
5.6.2 Converb B + sal ‘put’ .......................................................................................... 249
5.7 Converb B + Tasta ‘throw’ ......................................................................................... 250
5.8 Summary ..................................................................................................................... 255
Chapter 6: Modal and voice AVCs ....................................................................................... 257
6.1 Introduction ................................................................................................................. 257
6.2 Kör ‘see’ ...................................................................................................................... 258
6.2.1 Converb A + kör ‘see’ ......................................................................................... 258
6.2.2 Converb B + kör ‘see’ ......................................................................................... 260
9
6.3 Ber ‘give’ ..................................................................................................................... 261
6.3.1 Converb A + ber ‘give’........................................................................................ 261
6.3.2 Converb B + ber ‘give’ ........................................................................................ 267
6.4 Al ‘get, take’ ................................................................................................................ 272
6.4.1 Converb A + al ‘get, take’ ................................................................................... 272
6.4.2 Converb B + al ‘get, take’ ................................................................................... 273
6.5 Summary ..................................................................................................................... 276
Chapter 7: HPSG analysis ..................................................................................................... 278
7.1 Motivation ................................................................................................................... 278
7.2 Introduction to HPSG .................................................................................................. 279
7.2.1 Analyses of auxiliary verbs ................................................................................. 281
7.2.2 Auxiliary verbs according to Ivan Sag ................................................................ 281
7.2.3 Analyses of Persian complex predicates ............................................................. 283
7.3 HPSG modeling of Kazakh AVCs .............................................................................. 285
7.3.1 The auxiliary verb lexeme ................................................................................... 285
7.3.2 Features in describing Kazakh AVCs .................................................................. 290
7.3.3 Modal and voice AVCs ....................................................................................... 291
7.3.4 Perfective AVCs .................................................................................................. 294
7.3.5 Manner AVCs ...................................................................................................... 297
7.3.6 Interval AVCs ...................................................................................................... 299
7.3.7 Imperfective AVCs .............................................................................................. 302
7.4 Summary ..................................................................................................................... 308
Chapter 8: Conclusions ......................................................................................................... 311
8.1 TAM expressions in Kazakh ....................................................................................... 311
8.2 Results ......................................................................................................................... 313
8.3 Areas for future research ............................................................................................. 318
8.4 Impact and concluding remarks................................................................................... 319
References ................................................................................................................................. 320
10
List of figures
Figure 1.1: The Turkic family tree (simplified) .......................................................................... 16
Figure 1.2: Typology of viewpoint (grammatical) aspect according to Comrie (1976) ............. 25
Figure 3.1 Representation of auxiliary verbs entailing posture from a textbook ...................... 102
Figure 3.2 Distribution of posture verbs over position of objects ............................................. 107
Figure 3.3 Visualization of types of eventualities expressed by at ......................................... 114
Figure 4.1 Event structure entailed by CVB.B + ket ................................................................... 169
Figure 4.2 Social media post with image and sentence shown in (337) ................................... 170
Figure 4.3 Event structure entailed by CVB.B + iber ................................................................ 175
Figure 4.4 Event structure entailed by CVB.B + bol................................................................... 180
Figure 4.5 Illustration for testing plannedness .......................................................................... 184
Figure 4.6 Spectrogram of the AVC boyap boldïm ‘finished painting’ .................................... 187
Figure 4.7 Spectrogram of the dependent construction oqïp boldïm ‘finished reading’ ........... 187
Figure 4.8 Event structure entailed by CVB.B + ïq ................................................................... 188
Figure 5.1 Visualizations of the event structure entailed by CVB.A + tüs ‘fall’ ........................ 223
Figure 5.2 Plotted event structure and topic time in (501) ........................................................ 233
Figure 7.1 Partial hierarchy of the type lexeme ......................................................................... 287
Figure 7.2 Polysemy between the lexical and auxiliary uses of a member of aux .................... 288
Figure 7.3 Feature specification of type aux-modal&voice ...................................................... 292
Figure 7.4 Imperfective AVCs with at ‘lie’............................................................................. 307
Figure 7.5 Type hierarchy of aux .............................................................................................. 310
List of tables
Table 1.1 Schematic rules of vowel harmony ............................................................................. 33
Table 1.2 Consonant and vowel harmony of the genitive case suffix ......................................... 33
Table 1.3 The lenition of the AVC CVB.B + at and a comparison with other suffixes .............. 43
Table 1.4 Featural intersectivity in Latin and the lack thereof in Kazakh .................................. 57
Table 2.1 The distribution of the allomorphs of -GAn and -(y)AtIn ............................................ 70
Table 2.2 Examples of -GAn and -(y/A)tIn in dependent clauses. .............................................. 73
Table 2.3 Partial paradigm of the affirmative and negated AVC CVB.B + at in present tense .. 78
Table 2.4 Partial paradigms of -GAn and -GAn e- with the lexical verb ‘help’ .......................... 84
Table 2.5 Elicited partial paradigm of kömektes ‘help’ in the -GAn past. .................................. 84
Table 2.6 Partial anterior paradigm in independent and dependent clauses ............................... 86
Table 2.7 Partial nonpast paradigm and converb A in Tatar (Poppe 1963) ................................ 87
Table 2.8 Partial nonpast paradigm and converb A in Kazakh ................................................... 87
11
Table 2.9 Partial paradigm of the mirative and converb B ......................................................... 93
Table 2.10 Agreement in the protasis of a conditional sentence ................................................. 98
Table 3.1 Otïr, tur and ür in indicative and imperative ........................................................... 103
Table 3.2 Otïr, tur and ür in old aorist and in modern present tense ....................................... 104
Table 3.3 at in old aorist and in modern present tense ........................................................... 104
Table 3.4 converb selection of the auxiliary at (preliminary 1) .............................................. 115
Table 3.5 Boundedness and the distribution of the resultative and the pluractional ................. 126
Table 3.6 Converb selection of the auxiliary at (preliminary 2) ............................................. 127
Table 3.7 Converb selection in realis and irrealis apodoses with at ........................................ 130
Table 3.8 Testing hypotheticality against tense and AVC ........................................................ 133
Table 3.9 Converb selection of the auxiliary at (final) ............................................................ 136
Table 3.10 Old aorist, modern present and modern aorist in the four imperfective auxiliaries 137
Table 3.11 at vs atïr stem in various verb forms ................................................................... 138
Table 4.1 Ratio of types of the lexical verb in CVB.B + iber ‘send’ in the corpus ................... 178
Table 4.2 Summary of systematic contrasts between the auxiliaries iber and ïq ................... 204
Table 4.3 Summary of systematic contrasts between the auxiliaries iber and ket ................... 205
Table 4.4 Summary of systematic contrasts between the auxiliaries iber and bol .................. 206
Table 4.5 Summary of systematic contrasts between the auxiliaries iber and qoy.................. 208
Table 4.6 Summary of systematic contrasts between the auxiliaries ket and qal ..................... 210
Table 4.7 Summary of systematic contrasts between the auxiliaries ïq and ket ...................... 210
Table 4.8 Summary of systematic contrasts between the auxiliaries bol and ïq ..................... 211
Table 4.9 Summary of systematic contrasts between the auxiliaries ïq and qoy ..................... 212
Table 4.10 Ten most frequent (from top to bottom) lexical verbs in CVB.B + qoy/qal. ............ 213
Table 4.11 Summary of systematic contrasts between the auxiliaries qoy and qal .................. 214
Table 4.12 Ten most frequent (from top to bottom) lexical verbs in CVB.B + qoy/ket. ............. 215
Table 4.13 Summary of systematic contrasts between the auxiliaries qoy and ket ................... 216
Table 4.14 Summary of systematic contrasts between the auxiliaries ket and bol.................... 217
Table 4.15 Summary of systematic contrasts between the auxiliaries bol and qoy .................. 218
Table 4.16 Summary of systematic contrasts between the auxiliaries qal and ïq ................... 219
Table 4.17 Summary of systematic contrasts between the auxiliaries qal and bol ................... 219
Table 4.18 Contrastive features in the perfective AVCs ........................................................... 220
Table 5.1 Featural exhaustivity in the interval AVCs ............................................................... 222
Table 5.2 Ratio of syntactic dependence in CVB.A + kel ........................................................... 229
Table 5.3 The aspect-semantic comparison of the manner AVCs ............................................ 256
Table 6.1 Restriction of the lexical verb in CVB.B + ber ........................................................... 270
Table 6.2 Token rates of transitivity and stativity in the AVC CVB.B + al ‘take’. .................... 275
Table 7.1 Summary of the properties governing the distribution of the modal AVCs ............. 291
12
Table 7.2 Featural exhaustivity in the interval AVCs (repeated) .............................................. 299
Table 7.3 Summary of the functions of at according to converb and lexical verb selection ... 304
Table 8.1 Summary of feature specification of every single-auxiliary AVC ............................ 315
Word count: 98602
Abbreviations
The interlinear glosses follow the conventions of the Leipzig Glossing Rules, with the addition of
the following abbreviations:
ABIL
abilitative
ACCO
accomplishment
ACUTE
acute achievement
ADVZ
adverbalizer
ANT
anterior
AOR
aorist
ATIN
polyfunctional particle
2.4, §2.7)
ATTEM
attemptive
ATTR
attributive
CEXP
counterexpectational
COMPAR
comparative
COMPL
completive
CVB
converb
DELIM
delimitative
EO
existential operator
GAN
polyfunctional particle
2.4-§2.6)
H.ENER
high energy
HAB
habitual
IAM
iamitive (‘already’ flavor of
perfective)
IMM.SEQ
immediate sequential
INTENS
intensification
INTENT
intentional
L.ENER
low energy
LV
light verb
MIR
mirative
NINT
non-intentional
OPT
optative
ORD
ordinal numeral
POL
polite
PROPIN
propinquity
PROSP
prospective aspect
PURP
purposive
RECIP
reciprocal
REQ
request
S.RES
significant result
SH.INT
short interval
SSI
subordinate simultaneous inter-
val
SUB.AFF
subject affective
SUBORD
subordinator
TOP
topicalization
VBLZ
verbalizer
13
Transcription conventions
All Turkic languages exemplified in this work are transcribed using the same transcription system.
Even though most Turkic languages have official scripts, transcription is useful when comparing lan-
guages, as the official scripts can be radically different. In the transcription I follow the common Turko-
logical tradition (e.g. Johanson and Csató 1998), specified for Kazakh below (bold and italicized).
Cyrillic
(1940-
2018)
Latin
(since
2018)
Tran-
scrip-
tion
IPA
Latin
(since
2018)
Tran-
scrip-
tion
IPA
а
a
a
ɑ
ŋ
ŋ
ә
á
ä
æ
o
o
o
б
b
b
b
ó
ö
œ
д
d
d
d
p
p
p
е
e
e
je
q
q
q
ф
f
f
f
r
r
ɾ
г
g
g
g
s
s
s
ғ
ɣ
ʁ
t
t
t
х
h
h
χ
u
u
ʊ
і
i
i
ɪ
ú
ü
ʉ
и, й
ı
y
j
v
v
v
ж
j
ʑ
y
ï
ә
к
k
k
k
uw/w
ʊw/w
л
l
l
l
z
z
z
м
m
m
m
sh
ɕ
н
n
n
n
ch
t
ʃ
14
Chapter 1: Introduction
1.1 Motivation and the aim of this study
Turkic languages have a remarkably complex verbal paradigm. Tense-aspect-modality (TAM)
expressions include a plethora of synthetic (suffixal) markers as well as many periphrastic constructions
and most notably, a large number of auxiliary verb constructions (AVCs) Kazakh features 28 AVCs.
The Kazakh AVC is defined as the combination of one lexical verb standing in a converbial form, and
at least one auxiliary verb that contributes no lexical, only grammatical meaning while also bearing all
the inflection that a single verb would in the same environment, e.g. agreement and TAM markers, or
markers of a dependent clause. The AVCs and their combinations with other TAM markers have been
the central topic of many studies of Turkic (e.g. Johanson and Csató 1998; Anderson 2004; Akbaba
2011; Johanson and Csató 2018), and from a cross-linguistic perspective (Pullum and Wilson 1977;
Puglielli 1987; Kuteva 1999; Kuteva 2001; Anderson 2006; Müller 2010; Levine 2012; Bonami 2015;
Sag et al. 2020). Nonetheless, they are still not very well understood, especially with regards to Kazakh.
These analyses either encompass multiple Turkic languages and thus largely ignore fine-grain language-
specific peculiarities; focus on one language but use data from a large time span, leading to fallacious
statements regarding Modern Spoken Kazakh; do not include every AVC in the language, or are unable
to show distributional differences between AVC and conclude that some are synonymous. This thesis
aims to provide a theory based on empirical data regarding AVCs in Modern Spoken Kazakh.
This is the first study that systematically describes every Kazakh AVC. It reviews the extant lit-
erature and relies on extensive fieldwork to determine the use, function, and distribution of each AVC.
While some examples will support the idea that in certain cases, more than one AVC is possible, it will
be shown that even these, seemingly synonymous AVCs have properties that distinguish them, and
therefore, any occurrence of synonymy is due to some featural underspecification in the context. By
extension, it will be shown that those cases that are cited in the literature as examples of synonymy, or
cases where no contrast has been identified, are in fact different constructions and semantic contrasts
can be demonstrated.
A distinctive characteristic of my analysis is that the data on which it is based is all drawn from
Modern Spoken Kazakh. As discussed in §1.3.1, the arguments rely on empirical data acquired in field-
work from native speakers who are all from the southern region of the country and are of a single gen-
eration. Some works that focus on Kazakh AVCs, most notably Akbaba (2011), cite examples from
multiple sources spanning a relatively large timespan. His earliest source is from 1894, and the latest is
from 2002 (Akbaba 2011, 28795). Treating such data ignores the possibility that the language may
have changed within the space of a century; as will be shown, this approach leads to inaccuracy. It will
15
be demonstrated that some of the AVCs discussed in the literature are not in use anymore, or their
function and distribution have merged with another AVC. Equally, I argue that some functions described
in the literature regarding a given AVC, are not attested, which may stem from the fact that the data they
utilize is simply obsolete in the contemporary language.
1.2 The Turkic languages and Kazakh
The Turkic language family consists of some forty languages spoken mainly in Eastern Europe,
the Middle East, Central Asia, and Siberia (Glottolog, 2021 [Hammarström, Forkel, and Haspelmath
2018]). Countries, where the majority of the population speaks a Turkic language, are Turkey, Azerbai-
jan, Turkmenistan, Uzbekistan, Kazakhstan, and Kyrgyzstan. Apart from these nation states, Turkic
languages are recognized in partially sovereign areas, such as the Uyghur language in the Xinjiang Uy-
ghur Autonomous Region in China, or the many republics of the Russian Federation, such as the Chu-
vash Republic or the Tuva Republic.
Map 1: The geographic distribution of Turkic languages
1
The Turkic languages have been classified by various authors (e.g. Samoljovič 1922; Baskakov
1952; Benzing 1959; Johanson 2016, 2020). The following figure does not include every modern and
historical variety and is based on the family tree in Glottolog (Hammarström, Forkel, and Haspelmath
2018).
1
The map was taken from https://en.wikipedia.org/wiki/Turkic_languages#/media/File:Turkic_Lan-
guages_distribution_map.png on November 17, 2021. The colors represent the branches within Turkic. Kipchak
Turkic languages, including Kazakh, are shown in shades of green; red is Oghuz, blue is Turkestani (or Turki/Kar-
luk), purple is Siberian.
16
The oldest written sources are the Orkhon inscriptions from the 8th century AD, found in today’s
Mongolia. Since that period and especially throughout the expansion of Islam in Asia, Turkic languages
have been well-documented, including works describing Turkic languages, such as The Compendium
of the Turkic Dialects (Divânü Lügati't-Türk) written by Maḥmūd al-Kāšġarī in the 11th century.
In general, Turkic languages are head-final with a rather strict SOV word order. Their morphology
is characterized by suffixal agglutination. Suffixes obey varying rules of consonant and vowel harmony,
apart from some Turkic languages under the considerable influence of other languages (most notably
Uzbek under Persian influence). The verbal paradigms contain several dependent clause markers which
take functions that are expressed in some other languages by anaphoric or conjunctive constructions.
All Turkic languages, including Kazakh, have a complex tense-aspect-modality system featuring nu-
merous suffixal expressions and a myriad of periphrastic constructions, such as auxiliary verb construc-
tions, which is the core topic of this thesis. Good introductions to Turkic languages and Turkology, in
general, can be found in Menges (1994), Johanson and Csató (1998), Ercilasun (2007), and Johanson
(2021).
The region of today’s Kazakhstan was colonized by Turkic tribes in the 11th century AD. This
was followed by the foundation of the Kazakh Khanate in 1465. Following first the Russian, and the
subsequent Soviet rule, Kazakhstan became an independent state in 1990. Kazakh [ISO: kaz, Glottolog:
kaza1248] is spoken by approximately 13.2 million people (2009) mainly in Kazakhstan and neighbor-
ing China, Uzbekistan, Kyrgyzstan, and Russia. Within Turkic, has been classified as a South Kipchak
language along with Kyrgyz, Karakalpak, Nogai, and Kipchak Uzbek (Johanson 2021, 22). Kazakh has
three main dialects: Northern, Western and Southern (Amanolov 1997).
For a general introduction to Kipchak languages see Golden (2014) and Öner (1998), and for a
study on the Kipchak lexicon see Győrfi (2014). Kipchak languages are distinguished by a number of
Figure 1.1: The Turkic family tree (simplified)
17
phonological peculiarities, such as the lenition of some Old Turkic velar fricatives into a voiced labial-
velar approximant (e.g. ‘mountain’ taɣ
taw) (Johanson 2021, 382) and the devoicing of plosives in
initial position (e.g. ‘four’ dört
tört) (Berta 1998, 302). The lexicon of Kazakh is broadly character-
ized by the relative lack of Arabic and Persian borrowings, as opposed to languages with a more con-
siderable set of such borrowings, like Turkish and Uzbek. Rather, due to its 300-year long contact with
Russian, Kazakh has a considerable number of Russian borrowings. Kazakhstan can be called a fully
bilingual country given that almost the entire population can speak Russian, while most of the population
can also speak Kazakh. From a morphosyntactic point of view, Kazakh and the Kipchak languages share
most of their morphosyntactic expressions with other Turkic languages, with some notable differences,
such as the lack of the evidential past tense -MI and the extensive use of the suffix -A/y as an independ-
ent tense-aspect marker; in other Turkic languages -A/y is mainly used in dependent clauses.
1.3 Data collection and sources
Two different methods of data collection were used in this study. Elicited data was collected
through online fieldwork with native speakers based in Kazakhstan (§1.3.1). A “corpus” of naturalistic
AVCs was also extracted from the Sketch Engine corpus of Kazakh (Kilgarriff et al. 2004) 1.3.2).
1.3.1 Fieldwork methodology
The majority of the fieldwork data was collected between November 2020 and January 2022. In
this period approximately 57 hours of recorded elicitation sessions were collected from five consultants.
The elicitation sessions were conducted online due to the Covid-19 pandemic. The youngest consultant
was 28 years old, while the oldest was 36 years old in 2022. All the participants are native speakers of
Kazakh. They all used Kazakh at home with their families before moving out as adults. As an additional
criterion, all of the consultants still use Kazakh on a daily basis. These criteria of enrollment were nec-
essary for participation as a language consultant in the research since Kazakhstan is a fully bilingual
country with areas where Russian is dominant. The participants are all from the southern region of the
country. It has been claimed that one of the three dialects encompasses the southern regions, but at the
same time, it is also recognized that differences between dialects are limited to some lexical and phono-
logical peculiarities (Amanolov 1997). This approach provides a more accurate and more up-to-date
analysis, which is the key contribution of this work. This data was supported by some supplementary
written notes from 2013-2014, when I spent a year in Kazakhstan. Unless otherwise stated, all Kazakh
data presented was collected by the author, and reviewed by the consultants.
The fieldwork methodology mainly consists of elicitation, following the relevant literature on
tests and other parameters to consider in semantic fieldwork (Cover 2015; Cover and Tonhauser 2015;
Bowern 2008). In addition to elicitation, consultants were asked to perform tasks on the corpus (dis-
cussed in §1.3.2), for example, to double-check my initial classification of lexical verbs according to
18
transitivity. Before such a task, the consultants were appropriately trained. To minimize any mistakes
stemming from misunderstandings, mistranslation, or misinterpretation, elicited examples were speci-
fied with contextual information. These are provided with the examples discussed, when necessary.
Occasionally, the direct comments and intuitions of the consultants are also included to support the
analysis.
1.3.2 Corpus data
The dataset referred to as “the corpus” is a collection of AVCs obtained from the Sketch Engine
corpus of Kazakh (Kilgarriff et al. 2004)
2
comprising over 139 million words of written Kazakh. This
corpus of AVCs contains a total of 8353 naturalistic tokens, with each of the 18 auxiliary verbs repre-
sented by approximately 450 tokens each. As opposed to the consultants’ judgments, the corpus data is
not necessarily of the spoken variety, and it is not restricted geographically or temporally.
Throughout this work, descriptive analyses generated from the corpus, i.e. ratios of occurrences
of AVCs against parameters including syntactic dependence, negation, converb type, and lexical verb
selection are used to illustrate the natural distribution of AVCs in written Kazakh. Within the corpus of
AVCs, lexical verbs are categorized according to transitivity and several other lexical semantic factors,
such as the expression of motion, internal change, mental activity, and many more.
1.4 Introduction to frameworks of TAM
Expressing tense, aspect, and modality (TAM) is the main function of AVCs and TAM suffixes.
This section explains the theory, terminology, and definitions that will be used.
Tense is understood as the grammaticalized expression of the temporal location of an eventuality,
3
following Comrie (1985, 9). Even though it has been argued that all languages express tense (and some
syntactic frameworks assume a Tense node as a language universal (Chomsky 1986, 1995; Pollock
1989; Stowell 2007), some languages have no overt morphology to express tense. Languages argued to
be of this type include Guaraní (Tonhauser 2011) and Yukatek (Bohnemeyer 2002; 2009), i.e. tenseless
languages. I will argue that Kazakh has grammatical ways to distinguish four values: (i) past, (ii) non-
past, (iii) present, and (iv) future.
2
Sketch Engine is a collection of corpora of over 90 languages. It utilizes online sources and provides
various ways to search strings and shows their original context.
3
The literature usually makes the distinction between events and states as two fundamental types that can
be predicated (cf. §1.4.1). The term eventuality includes both types (Bach 1986).
19
Aspect refers to the structure of the eventuality. The internal structure of a given predicate is
referred to as lexical aspect, sometimes called actionality or Aktionsart 1.4.1). While lexical aspect is
thought of as an internal, inherent description of the eventuality, grammatical categories can also distin-
guish the aspectuality of the event externally, which is called grammatical aspect or viewpoint aspect
1.4.2). Lexical and grammatical aspects interact, which will be addressed in §1.4.3.
Modality is considered a separate module of grammatical expression that encodes ways a speaker
can categorize the uttered proposition, e.g. how probable it is, or whether it is necessary or obligatory
1.4.5). The subsections below further explain TAM in the description of Kazakh grammar.
1.4.1 Lexical aspect
Lexical aspect concerns the inherent temporal contour of an eventuality. The temporal contour is
generally expressed by verbal predicates and their arguments (Smith 1997, 17). In a highly influential
work, Vendler (1957) distinguished four types of eventualities that differ in lexical aspect, namely states,
activities, achievements, and accomplishments. His typology has been interpreted to consist of three
intersecting distinctions: (i) dynamic ~ stative, (ii) durative ~ punctual and (iii) unbounded ~ bounded
(Mourelatos 1981, 201202). Dynamic events entail some change over time, while statives do not. Du-
ratives entail that the event is extended over time, while punctuals describe an instantaneous change.
Bounded events entail a natural endpoint and a result state, while unbounded events do not.
The combinations yield the following types:
1. states (e.g. be Polish) [stative durative unbounded]
2. activities (e.g. dance) [dynamic durative unbounded]
3. achievements (e.g. shatter) [dynamic punctual bounded]
4. accomplishments (e.g. read the book) [dynamic durative bounded]
While the dynamic ~ stative and durative ~ punctual contrasts appear to be sufficient to describe
Kazakh data, boundedness needs to be further elaborated. In the above, boundedness (sometimes re-
ferred to as telicity) was defined as the entailment of a natural endpoint and a result state. Shatter and
read the book are great examples because in shatter the natural endpoint is when glass turns from one
piece into many, and the result state is where the glass is in many pieces. In read the book the natural
endpoint is reading the last word in the book, and the result state is having read the entire book. However,
there is another type of boundedness that neither entails a natural endpoint, nor a result state. In this
case, there is only temporal boundedness as the event is perceived as completed and yet, the natural
endpoint has not been reached. The typical example is the po- verbal prefix in Slavic languages that
express ‘to do something for a little while’ (Breu 1994, 28) an activity predicate that has been termi-
nated, as shown in (1).
20
(1) Polish (Slavic, IE; elicited example)
Context: What did you do before noon?
po-prac-owa-ł-em
DELIM
4
-work-IPFV-PST[3SG.M]-1SG
‘I worked for a little while / a little bit.’
The example in (1) does not entail a natural endpoint, but it is temporally bounded. According to
Croft (2012, 7980) the notion of boundedness should be broken down into a subtype that entails a
natural endpoint and a result state, and one that only expresses temporal boundedness. The former is
called q(ualitative)-boundedness, and the latter t(emporal)-boundedness. This distinction is relevant to
understanding the Kazakh data as it will be argued that some AVCs restrict the predicate to profile the
event phase where the natural endpoint is transgressed (CVB.B + iber ‘send’ in §4.3) or the result state
(CVB.B + qal ‘stay’ in §5.4) in a predicate, while others only entail temporal boundedness (CVB.A + tur
‘stand’ in §3.6.2 and CVB.B + bol in §4.4).
5
The term profile, adapted from Croft (2012, 55) will be used
to specify what phase an expression denotes, i.e. inherently includes in its meaning, as opposed to im-
plications.
Vendler’s typology is useful for accounting for the distribution of lexical verbs and grammatical
elements, such as grammatical aspect expressions (see §1.4.2). Given its popularity in studies on aspect,
it has been reviewed multiple times and more elaborate systems have been proposed with the addition
of new features (e.g. Dahl 1985; Bach 1986, Dowty 1991; Smith 1997). For instance, Botne (2003)
examines the achievement predicate ‘die’ in 18 languages, which is usually thought of as an event de-
scribing a change-of-state and that is conceptualized as instantaneous. However, while in Vendler’s
typology an achievement is a type of the highest granularity, Botne proposes further complexity by
further detailing event phases, i.e. the subparts of an eventuality, of an achievement. First, he breaks
down the possible event structure of the concept ‘die’ and proposes a phase preceding the core denota-
tion of ‘die’, i.e. in ‘Berik is dying’ the verb die profiles a durative phase which leads to the loss of life
but does not profile death. This is referred to as onset. The core meaning, which in the case of ‘die’ is a
change-of-state phase is labeled nucleus. Botne argues that in achievements, the nucleus is a punctual
phase, while in accomplishments it is durative. Lastly, the phase following the nucleus is coda, which
in this case, is a stative phase of being dead. In this work this typology will be generalized, i.e. activity
4
Given that this semantic category is expressed in Kazakh (§3.6.2), the verbal prefix po- is glossed delim-
itative consistently with the label of the AVC CVB.A + tur ‘stand’.
5
An AVC consists of an auxiliary verb lexeme and the converbial form of the lexical verb. Whenever I
refer to an AVC, I will use the following notation: [CONVERB A or B] + auxiliary lexeme ‘original meaning of the
auxiliary’, e.g. CVB.A + tur ‘stand’.
21
predicates such as ‘dance’ will be taken to denote a durative, dynamic nucleus without an entailed onset
or coda.
Botne (2003) argues that achievements can be classified into four types based on whether they
profile a certain part of the aspectual contour of an event. His typology is shown below.
(2) a. Acute: nucleus alone
b. Inceptive: nucleus plus onset
c. Resultative: nucleus plus coda
d. Transitional: nucleus plus onset and coda
In Chapter 4, I argue that these distinctions are helpful in accounting for differences in the distri-
bution of perfective AVCs. Specifically, I demonstrate that the AVC CVB.B + ket ‘leave’ (§4.2.2) is only
compatible with Botne’s resultative achievements, and the auxiliary iber ‘send’ 4.3) which consist-
ently forces compatible predicates to be read as an acute achievement.
A further elaboration of Vendler’s typology is proposed by Croft (2012). He argues that classify-
ing aspectual categories into the four well-known categories (such as states and achievements) is flawed
as the granularity of such categories is unavoidably too low. Consequently, he proposes a way of visu-
alizing eventualities that essentially breaks down the types of eventualities into as many categories as
the analyst deems necessary. For instance, the following two examples (Croft 2012, 58) show how sta-
tive predicates can differ in terms of actional phases they lexically encode by contrasting She is French
with The window is shattered, as illustrated in (3).
(3) a. ‘She is French.’
q
t
The two figures in (3) show the change in t(ime) in the x axis and q(uality) in the y axis. While
time is easily encoded as a one-dimensional parameter, quality is a complex notion that renders all
change, action, and motion into one dimension, based on the granularity required to describe an even-
tuality. The predicate ‘to be French’ is represented as a property that does not change over time (thus
the line is flat), and the arrow specifies that this property continues to apply. In (3)b the predicate is ‘to
be shattered’, which implies three actional phases: two states, and an event that can be characterized as
a change in quality. The solid line represents the actional phase that is profiled, i.e. the phase being
expressed by the sentence. To the left of the solid line, the dotted lines represent the event of the breaking
q
t
b. ‘The window is shattered.’
22
of the window (vertical line, rapid change in quality), as well as the state where the window was intact
(horizontal dotted line, no change in quality). The line is dotted, because the sentence does not profile
these actional phases, they are only inferred. Croft labels (3)a as a Permanent, Inherent state, while (3)b
is a Permanent, Acquired state. Croft’s system is helpful in visualizing aspectual specifications, espe-
cially because it makes a clear distinction between parts of the eventuality that an expression profiles,
and parts that are implied, but not profiled. This way of representing the phases of eventualities will be
used for visualization where necessary.
1.4.2 Grammatical aspect
While lexical aspect categories are used to describe the inherent temporal structure of eventuali-
ties, grammatical aspect categories distinguish “different ways of viewing the internal temporal constit-
uency of a situation” (Comrie 1976, 3). The main categories of grammatical aspect are perfective and
imperfective, generally acknowledged in most works on the topic. Following Comrie’s (1976, 16) defi-
nition: “perfectivity indicates the view of a situation as a single whole, without distinction of the various
separate phases that make up that situation; while the imperfective pays essential attention to the internal
structure of the situation.”
Slavic languages are frequently cited as having a rather clear morphological distinction between
perfective and imperfective verbs. In these languages, most verbs can be paired where one is perfective
and the other one is imperfective, and the morphological realization can vary between the addition of a
verbal prefix, an infix, or suppletion. Note, however, that Slavic perfectivity is an overly complicated
topic, so the following examples should be taken as one of the rather clear distinctions (see a discussion
in Croft (2012, 11024)).
(4) Context: The speaker is being asked whether his friend had explained how to set up the new
computer. The utterance is considered a complete response to the question.
Polish (Slavic, IE; elicited examples)
a. wy-tłumaczy-ł
OUT-explain-PST[3SG.M]
‘He has explained it [and I understood it].’
b. tłumaczy-ł
explain- PST[3SG.M]
‘He was explaining it [but I did not understand it].’
The example in (4)a. features a perfective verb form realized by the presence of the verbal prefix
wy- ‘out’. It entails both temporal and qualitative boundedness as the reading entails that the explanation
was complete, the speaker understood the explanation, and entailed by this qualitative boundedness, the
23
explaining event is temporally bounded as well. In (4)b., the imperfective verb form denotes a tempo-
rally bounded but qualitatively unbounded event conveying the concept of incompleteness. That is, the
explaining event is viewed as a temporally closed interval. However, the explanation is read as unsuc-
cessful, and no qualitative boundary is specified, i.e. it is unknown how far the explanation went, thus
this eventuality is arguably q-unbounded.
The distribution of grammatical aspect expressions depends on the lexical aspect of the predicate.
For example, stative predicates typically do not license imperfective verb forms (#John is knowing
French) (Cover and Tonhauser 2015, 332 citing Vendler 1957; Comrie 1976).
6
It is crucial to highlight
that lexical aspect is not provided by the lexical verb per se, but the predicate in a given context. While
‘know’ is easily associated with a stative meaning that truly rules out any change, verbs like ‘love’
feature more flexibility. The following examples are elicited.
(5) a. I am loving this veggie burger.
b. #I am loving my father.
In (5)a, the object of the sentence is a burger that the speaker is trying the first time. The progres-
sive (imperfective) ‘be -ing’ is accepted because the context specifies that there is a kind of a change —
the speaker realizes that they love the burger. Consequently, the predicate is not considered stative, but
dynamic and thus the progressive is accepted. In (5)b, however, the object is the speaker’s father and
under normal circumstances, loving one’s father is perceived as a stative predicate that is not associated
with any change or temporal confinement. In this case, the imperfective ‘be -ing’ is not accepted (Cover
and Tonhauser 2015, 333). Another restriction is observed when the imperfective combines with sem-
elfactive (atelic and instantaneous) predicates. When a semelfactive verb, such as ‘cough’ is marked for
the imperfective, as in ‘Sam was coughing’ the predicate must be read as a pluractional, i.e. it cannot
express only a single unit of cough (Smith 1997, 18081).
Perfective markers are generally less restricted in terms of combining with different lexical aspect
classes in English (Smith 1997, 6771). In other languages, such as Mandarin, perfective markers are
restricted with stative predicates. Similar to the example in (5), the Mandarin perfective marker le is
licensed with statives that allow for a dynamic interpretation (sick ~ get sick), but other statives that are
less commonly conceptualized in a potentially dynamic interpretation (#intelligent ~ become intelli-
gent), cannot combine with le.
6
Two types of infelicity are specified throughout. The asterisk (*) stands for structural infelicity, such as
in ‘*Jack are guy a nice.’, while the number sign (#) stands for semantic infelicity. In (5)a, the construction ‘am lov-
ing’ is infelicitous because of the semantic class membership of ‘love’, and the object ‘father’. Otherwise ‘am V-
ing’ is structurally acceptable (Carnie 2011, 1516).
24
(6) a. Mali bing-le (Mandarin Chinese, Sinitic (Smith 1997, 70))
Mali sick-PFV
‘Mali got sick.’
b. #Mali congming-le
Mali intelligent-PFV
Intended: ‘Mali became intelligent.’
Just as lexical aspect, grammatical aspect has been argued to have a multitude of subcategories,
such as imperfective expressions of event plurality (e.g. the habitual) or the many types of perfectives,
such as the resultative, expressing a state caused by a previous event, or the completive, expressing the
completion of a task. The subtypes that are relevant to account for the Kazakh data will be introduced
in §1.4.3.
1.4.3 The neo-Reichenbachian framework and its limits
While there are several frameworks for linguists to formalize notions within tense and aspect, this
thesis will mainly use a neo-Reichenbachian framework, based on Reichenbach (1947) as well as the
modifications in Klein (1994) and Kamp and Reyle (1993). As this framework provides a simple way
to describe certain Kazakh TAM expressions, it will be used throughout. However, as will be shown,
the neo-Reichenbachian framework is not sufficient in formalizing every type of aspectual denotation,
so some parts of the analysis will not use it.
The neo-Reichenbachian framework is highly effective in describing and distinguishing tense and
grammatical aspect. Tense is defined as “a grammaticalized linguistic form that serves to restrict the
temporal location of the time being talked about” (Cover 2015, 234). The formalization is based on
Utterance Time (UT), i.e. the actual time a given sentence is uttered, and Topic Time (TT), that is, the
time interval that the utterance is about.
7
Tense is formalized as the relation of these two, such that in
past tense the topic time precedes utterance time (UT > TT), in present the topic time is equal to or
included in utterance time (UT TT), and in future tense the topic time follows utterance time (UT <
TT). Cross-linguistically a typical tensed expression may specify values such as PAST, PRESENT, and
FUTURE, as shown in (7). Since Kazakh encodes the nonpast tense (§2.2), this relation is shown in (7)d.
7
Cover uses a modified Reichenbachian framework, and one of the differences lies in terminology. In
Cover and the present work, Utterance Time (UT) is used instead of Reichenbach’s Speech time (S), Event Time
(ET) instead of Event time (E) and Topic Time (TT) instead of Reference time (R).
25
(7) a. I watched Star Wars [yesterday]. (UT > TT) PAST
b. I am watching Star Wars [right now]. (UT TT) PRESENT
c. I will watch Star Wars [tomorrow]. (UT < TT) FUTURE
d. I will watch Star Wars. / I live in Almaty. (UT TT) NONPAST
Aspect is connected to tense in that it specifies temporal restrictions, but instead of specifying
when an event occurred, it provides information on how we view the inherent temporal structure (lexical
aspect) of the predicate. Recall that as Comrie (1976, 3) says, aspect provides different ways of viewing
the internal temporal constituency of a situation”. The most widely studied aspect values are the perfec-
tive and the imperfective aspects. The perfective expresses an event as a whole, without attempting to
focus on its internal structure. In contrast, the imperfective focuses on the internal temporal structure of
the event. Most languages contrast more aspectual categories, such as the habitual and the progressive
(see e.g. Comrie 1976; Dahl 1985; Bybee, Perkins, and Pagliuca 1994; Heine 1993). The habitual is not
always analyzed as an aspect, but as a tense (see e.g. Givón 2001, 286). The cornerstone of the problem
is that the habitual is not a specification of the relation to Utterance Time, nor to Topic Time, but can
be interpreted as specifying the frequency of the eventuality, thus it is not purely a tense or aspect cate-
gory. The hierarchy of grammatical aspect (also known as viewpoint aspect) categories is generally
assumed to follow Comrie’s (1976, 25) typology presented in Figure 1.2
Figure 1.2: Typology of viewpoint (grammatical) aspect according to Comrie (1976)
The neo-Reichenbachian framework formalizes grammatical aspect categories the following way
(based on Comrie 1976; Cover and Tonhauser 2015). The perfective aspect views an eventuality as a
bounded entity with no internal temporal structure. In this framework, the description of the perfective
aspect requires reference to topic time (TT), utterance time (UT), and a third notion: eventuality time
(ET). ET is the temporal span of the eventuality, and it can be instantaneous as with Vendler’s accom-
plishments, or a long interval, such as in the case of activities (Cover and Tonhauser 2015, 308309).
As a formal definition of the perfective aspect, ET is included or equal to TT (ET TT). For example,
in Yesterday I watched Star Wars, the temporal span of the movie-watching event (ET) is placed entirely
26
within the boundaries of topic time (TT) made explicit by the temporal adverb yesterday, without any
further detail regarding how the watching happened temporally. ET can be the same interval as TT, for
example, specified by temporal adverbs as in I painted the room in two hours.
The imperfective aspect is quite the opposite of the perfective, defined as the eventuality time
including topic time (TT  ET). In At 2’o clock I was watching Star Wars, the temporal reference, or
topic time (TT), at 2’o clock, is entirely included in the time span denoted by the event of watching the
movie. In this case, English expresses imperfectivity with the construction [be -ing]. Cf. the same sen-
tence where the verb ‘watch’ stands in the simple past form: At 2’o clock I watched Star Wars. This
sentence implies that the watching of the movie began at 2’o clock, i.e. the event is bounded at its point
of initiation, as opposed to the imperfective sentence (was watching), which is not bounded.
8
The neo-Reichenbachian framework proves useful when a given category is analyzed as being
either tense or aspect. In §2.4, the participles -GAn and -(y)AtIn will be argued to denote the anterior
and prospective aspects, respectively. The anterior (sometimes called perfect) is defined as (ET<TT),
that is, the event happened before a given topic time, while the prospective is the opposite (TT<ET),
where event time follows topic time (Cover and Tonhauser 2015, 325). These categories are essentially
equal to relative tenses described elsewhere in the literature (e.g. Comrie 1985, 36). The formalization
allows us to illustrate that the aspect categories are orthogonal to tense, as for example, in the following
pair of sentences. Both in (8) and (9) the prospective aspect is shown, but in (8) in past reference, while
in (9) in future reference.
(8) (TT<ET<UT) prospective in past
Berik-tiŋ keše attïɣ-atïn-ï-n bil-mey
Berik-GEN yesterday exercise-ATIN-3-ACC know-NEG.CVB
aldïn ala bekerge kino-ɣa bilet al-ɣan e-di-m.
in advance in.vain cinema-DAT ticket buy-GAN COP-PST-1SG
‘I had bought a ticket for the cinema in advance, in vain, because I did not know that Berik
would exercise yesterday.’
8
This interpretation is reported by native English speakers in the south of England.
27
(9) (UT<TT<ET) prospective in future
Context: Berik’s plane hasn’t taken off, but we know that it will be late.
Ušaɣ-ï kešik-pek, sondïqtan bügin keški concert-te
plane-3 be.late-FUT therefore today evening concert-LOC
Berik-tiŋ oyna-ytïn-ï belgisiz
Berik-GEN play-ATIN-3 uncertain
‘His [Berik] plane will be late, so it is uncertain whether Berik will play at the concert today
in the evening.’
We now turn to aspect categories that are not successfully described in the neo-Reichenbachian
framework, like event plurality. Such cases are usually described with the addition of descriptive meth-
ods (see e.g. Cover and Tonhauser 2015, 33037 and Bar-el 2015). Grammatical specification of plu-
ractionality (also known as event plurality) entails that the described event consists of multiple subev-
ents. The clearest example is the habitual aspect, which is classified as a subtype of imperfectivity
(Bybee, Perkins, and Pagliuca 1994, 151). In English, the habitual aspect can be expressed by the ‘used
to do’ construction as in ‘Peter used to swim (every morning).’ In this case, ‘swim’ is read as a recurrent
event and thus it cannot be interpreted as a single event, regardless of the presence of the adverb ‘every
morning’. The formalization of such events is not well established in the neo-Reichenbachian frame-
work, but some proposals for a solution are reviewed in Cover and Tonhauser (2015, 32426). For
example, Cable (2020, 25) assumes a ‘Habitual Imperfective Head’ and formalizes it as a combination
of modal (the Habitual part) and aspectual (Reichenbachian notation) components. According to his
proposal, the Habitual is an aspect operator whose topic time relates to the habits of a given person, thus
resolving vagueness by mapping the temporal characteristics of the habitual predicate to a world where
it exists.
9
In Kazakh, some TAM expressions entail pluractionality, and some others systematically imply
it. Among the consistently pluractional markers are the past habitual suffix -(y)AtIn 2.7), the habitual
aspect AVC CVB.B + ür ‘walk’ 3.7.1) as well as the pluractional AVC CVB.B + at ‘lie’ that only
combines with certain motion verbs 3.4.3-3.4.5). Other constructions imply pluractionality, such as
the nonpast suffix -(y)A which functions as a gnomic (also known as generic) aspect marker that is
frequently read as a pluractional. The gnomic aspect is understood to be expressing present tense refer-
ence, but topic time has a ‘large span’, thus compatible with statements of ‘general truths’. Additionally,
the gnomic is both compatible with dynamic and stative predicates (Bybee, Perkins, and Pagliuca 1994,
141).
9
Cable’s (2020, 25) formalization of the ‘Habitual Imperfective Head’ is as follows:
[[ IMPRVHAB ]]w,t,g = [λP<s, <ε,t>> : [ λtʼ : w’ HABIT(w,t’) . e . P*(w’)(e) & t’ T(e) ] ]
28
Among the AVCs, the continuative CVB.A + ber ‘give’ 6.3.1) expresses the persistent continu-
ation of a dynamic event, but it has a very strong tendency to express habituality where the event is
perceived as annoying for some affected party. This AVC is distributed similarly to the English AVC
‘keep -ing’ as in ‘People keep asking me when I am going to get married.’ More specific types of plu-
ractionality are the frequentative and the iterative. The frequentative is shown to express that an event
occurs with higher frequency than the habitual, while the iterative is defined as a pluractional that entails
that the subevents happen on one occasion, i.e. ‘read the letter over and over [in the time span of one
afternoon / #every day]’. In Kazakh, both are compatible with the AVC CVB.B + at ‘lie’ (§3.4.1-3.4.2);
the frequentative requires the presence of the nonpast suffix on the auxiliary verb.
Another class of AVCs where the neo-Reichenbachian formalism will not be used extensively is
that of the perfective AVCs (Chapter 4). These expressions all fit the neo-Reichenbachian definition of
perfectivity, i.e. (ET TT), but each AVC specifies either a non-aspectual distinction or the specifica-
tion of a phase within the event. An example of a non-aspectual specification is that of the propinquitive
(‘almost’) CVB.A + azda 4.7), which entails the non-realization of the event a feature that is
not part of neo-Reichenbachian analyses. For an aspectual issue, a clear example is the AVC CVB.B +
ket ‘leave’ (§4.2.2) which entails a persistent result phase. Differences between these cases will be vis-
ualized following Croft’s (2012) eventuality phase profiling system, introduced in §1.4.1.
Some AVCs, mostly in the said perfective class, profile and specify a particular phase in the event
structure. This raises theoretical questions as the empirical data suggest that what the AVC expresses is
not generally classified as grammatical aspect, but lexical aspect. See the following examples, where
the AVC CVB.B + ïq ‘exit’ is only compatible with accomplishment predicates, and it profiles the dy-
namic phase (leading to the change-of-state phase), while AVC CVB.B + ket ‘leave’ is also compatible
with accomplishment predicates, but it only profiles the change-of-state phase and the result state. In
(10) both AVCs are accepted because nothing requires the dynamic phase to be profiled. In (11), how-
ever, the adverb requires the dynamic phase to be profiled, and hence only CVB.B + ïq is felicitous.
(10) keše bayandama-
yesterday assignment-ACC
az-ïp [šïq--m / ket-ti-m]
write-CVB.B [AUX(ACCO)-PST-1SG / AUX(S.RES)-PST-1SG]
‘Yesterday I finished writing the assignment.’
29
(11) keše bayandama- [qïyïn bolsa da / tez]
yesterday assignment-ACC [even if it was difficult / quickly]
az-ïp [šïq--m / #ket-ti-m]
write-CVB.B [AUX(ACCO)-PST-1SG / #AUX(S.RES)-PST-1SG]
‘Yesterday I finished writing the assignment [even if it was difficult / quickly].’
Most theories assume that lexical aspect categories are inherent to the predicate and it is mostly
tied to the verbal lexeme (Vendler 1957). However, some authors argue that it is not quite possible to
distinguish what aspectual features are expressed by a lexeme itself, and by grammatical elements (Dahl
1985, 2627; Croft 2012), thus a strict division between values of grammatical and lexical aspect is not
supported cross-linguistically. Other works elaborate on the idea that lexical aspect categories are ex-
pressed by grammatical elements (for the theoretical base see Laca 2004, 2006; Cabredo Hofherr, Laca,
and Carvalho 2010, and for a recent analysis of Hungarian adapting this approach, see Kardos and Farkas
2022) and as shown throughout, the semantic content of some Kazakh constructions does not always
correspond to Vendlerian lexical aspect categories. Thus, my analysis is inspired by Laca’s analyses,
which assume a syntactic distinction between a higher aspect (expressing categories of grammatical
aspect) and a lower aspect (expressing lexical aspect categories). While I do not pursue the syntactic
analyses that were used to conclude that more than one type of aspect can be expressed grammatically,
I will argue that some semantic denotations expressed by auxiliary verb constructions in Kazakh belong
in the realm of grammatical aspect, while others are similar to traditional lexical aspect values. In this
regard, as I do not assume a fundamental, classificational difference between lexical and grammatical
aspect categories when it comes to expressions of AVCs, my work follows Croft (2012, 32), and it can
be categorized as a unidimensional approach, as opposed to bidimensional ones, which do assume two
distinct classes of aspect categories (Sasse 2002, 202203).
1.4.4 Lars Johanson’s framework on tense and aspect
Lars Johanson and his colleagues, especially Éva Csató, have produced a great number of works
on Turkic languages with a special interest in TAM problems (see e.g. Csató, Johanson, and Karakoç
2019; Johanson 1999, 2000a, 2000b, 2003, 2004, 2021; Csató and Johanson 1994; Johanson and Csató
1998). Drawing mainly on evidence from Turkic languages, Johanson developed a cross-linguistic
framework for describing TAM expressions, including AVCs which he refers to as postverbial con-
structions. Johanson’s framework is based on the separation of viewpoint aspect (roughly equivalent to
grammatical aspect, e.g. perfective vs. imperfective) and actionality (roughly equivalent to lexical as-
pect, e.g. the four Vendler categories). That is, according to Sasse’s (2002) classification of theories of
aspect, his framework is strictly bidimensional by separating these two classes of features, as opposed
to this work, or that of Croft (2012), which can be considered as mainly unidimensional.
30
Johanson argues that there are three main types of viewpoint operators: (i) intraterminals, that
view eventualities within their temporal boundaries (roughly equivalent to imperfectivity) (12), (ii) post-
terminals, that view the eventuality after the transgression of its temporal boundaries (roughly equiva-
lent to perfectivity) (13) and (iii) adterminals which profile the attainment of a relevant boundary and
thus, it corresponds to the notion of perfectivity but with the additional criterion that the relevant bound-
ary is profiled (14). Adterminality is claimed to be demonstrable only in Slavic languages.
(12) Peter is watching Star Wars.
(13) Peter has watched Star Wars.
(14) Russian postroit‘build’ [+ ADTERMINAL] vs. stroit’ ‘build’ [- ADTERMINAL]
10
Apart from the viewpoint operators, Johanson assumes a number of other features in describing
eventualities. A key feature is actionality which is roughly equal to lexical aspect and he assumes the
main values transformative and nontransformative. This feature is equal to the bounded-unbounded
contrast in Vendlerian works.
11
Transformative predicates inherently express some change (“a natural turning point”), which is a
concept that can be viewed analogous to change-of-state verbs, e.g. ‘break’ or ‘grow’ (Rappaport Hovav
2002). Johanson uses subclasses to denote a transformation at the beginning of the eventuality (initio-
transformative, e.g. ‘hide’) and at the end of the eventuality (finitransformative, e.g. ‘die’) in this
regard, his framework is similar to that of Breu (1994), Sasse (1991) and Bickel (1997). Within the
feature of transformativity, he distinguishes subfeatures, such as MOMENTARY (which is equal to the
feature DURATIVITY that Vendlerians assume e.g. to distinguish between an achievement and accom-
plishment).
In addition, Johanson assumes a tense feature [± PAST], as well as a relation between Localization
point and Orientation point, which are equivalents of Utterance time and Topic time (discussed in
§1.4.3), respectively (Johanson 2000b, 35). In order to account for non-single-eventuality predications,
he introduces a pluractional feature he refers to as pluri-occasional [±]. One of the unique concepts in
Johanson’s framework is FOCALITY. Focality looks at whether a TAM specification profiles a relatively
large (low focal) or small (high focal) portion of the eventuality in a temporal regard. For example, he
argues that the English sentences ‘Peter has gone’ is low focal, while ‘Peter is gone’ is high focal
10
Johanson (2000b, 13537) does not provide a clear example in form of a sentence with context.
11
It must be mentioned Johanson explicitly says that transformativity is referred to as perfectivity in other
works, while nontransformativity is referred to as imperfectivity (Johanson 2004, 181). However, to remain con-
sistent with the definitions used here, I rather compare transformativity to boundedness.
31
(Johanson 2000b, 120). The concept of focality is discussed with respect to the distribution of the AVCs
CVB.B + tur 3.6.1) and CVB.B + ür 3.7.1).
I agree with much of Johanson’s analyses apart from two crucial points. Firstly, as I shall argue
in §3.4, FREQUENCY should be included as a parameter in a full account of the aspectual system of
Kazakh given that some structures are sensitive to it. Secondly, Johanson’s works tend to argue that
non-temporal-aspectual distinctions, such as intensity, suddenness, volition, or intention are side effects
of the base transformational vs. nontransformational opposition (e.g. Johanson and Csató 2018, 153). I
find this unlikely in the case of Modern Spoken Kazakh because these features are orthogonal to trans-
formativity and they are consistently expressed by particular AVCs (see throughout Chapter 4). In dia-
chrony, such an analysis might be enlightening, but in a synchronic study such as this one, it does not
fully explain the attested data.
1.4.5 Modality
Modality is one of the umbrella terminologies in linguistics for which various definitions have
been proposed. According to Bybee, Perkins, and Pagliuca (1994, 176) modality can be vaguely defined
as “the grammaticalization of speakers’ subjective attitudes and opinion. Similarly, in Abish (2014, 11)
modality is defined broadly, as “the expression of attitudes towards the proposition”. Portner (2009, 1)
uses a much narrower definition: “modality is the linguistic phenomenon whereby grammar allows one
to say things about, or on the basis of, situations which need not be real”. Yet another definition is
proposed by Cover and Tonhauser (2015, 234): “[modality] is a way of talking about possibility, neces-
sity, probability, and so on; languages often have both grammatical and lexical mechanisms for express-
ing modality.” This thesis will not attempt to define modality but instead, some grammatical expressions
will be categorized as modal expressions, following semantic denotations and labels generally associ-
ated with modality in linguistics.
Modality is often broken into three categories: dynamic, deontic and epistemic (Nuyts 2016).
Dynamic modal expressions concern notions such as ability, need, necessity and permission. Among
these participant-inherent expressions encode the needs, abilities etc. of the first argument, whereas
potentials describe these notions but when the source is not inherent to the first argument, but an external
force. In Kazakh, the abilitative AVC (§6.4.1) belongs to this category. Deontic modality encodes “an
indication of the degree of moral desirability of the state of affairs expressed in the utterance” (Nuyts
2016, 36), including expressions of permission, obligation and advice. The Kazakh counterexpectational
readings of the continuative CVB.A + ber ‘give’ 6.3.1) can be classified as such. §6.2 discusses the
AVC CVB.B + kör ‘see’ and analyzes it as the expression of the attemptive modality, which is best trans-
lated to English as ‘try to do’. Epistemic modal categories express an estimation of the likelihood that
the predicate is realized. Kazakh AVCs express some of these categories, such as high hypothetical
CVB.A + at ‘lie’ (§3.4.6-§3.4.7), the low hypothetical CVB.B + at ‘lie’ (§3.4.8).
32
Apart from modal AVCs, Kazakh has a past tense epistemic modal suffix used as in independent
clauses. -(I)p, labeled the mirative (§2.9), expresses that the speaker became aware of the predicate after
it had happened and thus it frequently conveys a reading of surprise. A similar modal category, the
counterexpectational is similar that it might be read as a surprising fact, but unlike the mirative, the
counterexpectational is only compatible with a context where some participant is bothered, annoyed by
the predicate, or finds it unethical. The also epistemic modal suffix -(A)r is labeled the aorist, and it
expresses a “neutral” epistemic degree, such as the English may or maybe 2.8).
The realis / irrealis dichotomy (the terminology follows Nikolaeva (2016) and Pinkster (2021,
315)) concerns the epistemic parameter of whether the speaker presents the predicate as one that may
actually have happened (realis), or one that is presented under the overt condition that it certainly did
not, or will not happen (irrealis). That is, the irrealis described imagined, hypothetical or potential states
of affair (Elliott 2000, 67). For a summary of this issue, see Nikolaeva (2016).
In the Turkology literature, irrealis is sometimes referred to as counterfactual (Johanson 2021,
761), and in other works realis and irrealis have been called real condition and unreal condition, re-
spectively (Muhamedowa 2015, 5758). In §3.4.6-§3.4.8, conditional sentences will be analyzed. In
these cases, the protasis can be either realis, e.g. If it rains, the ground gets wet or irrealis, where the
protasis that explicitly presents an unreal event, prototypically with past reference, e.g. If I had married
Jessica, we would have three kids by now (for English conditionals see e.g. Huddleston 2002).
1.5 Kazakh TAM expressions: Morphology and Syntax
This section introduces the chief morphological and syntactic characteristics of TAM expressions
in Kazakh. The aim is to present the components of AVCs, as well as their status as grammatical mark-
ers. §1.5.1 and §1.5.2 introduce the morphology of suffixes in general, and §1.5.4 discusses multi-word
constructions. §1.5.4-§1.5.8 provide a detailed description of the components of AVCs, namely the con-
verbial forms of the lexical verb, and the status of the auxiliary verb. §1.5.9 briefly introduces AVCs
with multiple auxiliary verbs and lastly, §1.5.10 argues that Kazakh AVCs are periphrases. These details
are essential to the rest of the thesis.
1.5.1 Phonological properties of suffixes
In the Turkological tradition (e.g. Johanson and Csató 1998; Öner 1998; Erdal 2004), suffixes are
distinguished by their phonological characteristics. In Kazakh, as in most other Turkic languages, suf-
fixes obey consonant and vowel harmony rules (for a summary and the traditional analysis in Turkology,
see Johanson (2021, 24346)). These assimilations can be of different types, according to the initial
phoneme of the affix, and there are discrepancies due to an affix’s history (e.g. see §2.7 for an affix that
originates in an AVC). According to the vowel harmony rules in Kazakh, vowels can be classified into
33
two types (Type 1 and 2). Suffixes that belong to Type 1 feature either the vowel [a] or [e] the former
appears following a stem featuring back vowels, and the latter follows stems with front vowels. Such a
suffix is the conditional with the allomorphs [-sa] and [-se]. Type 2 suffixes function the same way, but
instead of [a] and [e], they feature the back vowel [ï] and the front vowel [i]. An example of such a
suffix is the second singular agreement suffix with the allomorphs [-sïŋ] and [-siŋ]. Written Kazakh
presents harmony along backness in certain vowels. In spoken Kazakh roundedness is also a contrasting
factor, as shown in Table 1.1
Table 1.1 Schematic rules of vowel harmony
Written Kazakh
Spoken Kazakh
Type 1
Type 2
Type 1
Type 2
Unrounded
Rounded
Unrounded
Rounded
Front
[e]
[i]
Front
[e]
[ö]
[i]
[ü]
Back
[a]
[ï]
Back
[a]
[ï]
[u]
Initial consonants of suffixes usually harmonize with the previous phoneme in voicing, and in
some cases, there is an additional allomorph that can be triggered by nasal consonants, sibilants, liquids
or glides. For instance, in the genitive case suffix -Nthe initial [n] harmonizes, as following vowels
and nasal consonants, it is realized as [n], following non-nasal voiced consonants as [d], and following
voiceless consonants as [t]. The examples in (15) show consonant harmony in the genitive case suffix
following a vocalic stem (a), a nasal consonant (b), a voiced sibilant (c), and a voiceless consonant (d).
The distributional generalization regarding the suffix -NIŋ is presented schematically in Table 1.2.
(15) a. qala-nïŋ city-GEN
b. ana-m-nïŋ mother-1SG-GEN
c. qaɣaz-dïŋ paper-GEN
d. qazaq-tïŋ Kazakh-GEN
Table 1.2 Consonant and vowel harmony of the genitive case suffix
Front V
Back V
Vowels and nasal Cs
[niŋ]
[nïŋ]
Sibilants, glides and liquids
[diŋ]
[dïŋ]
Voiceless Cs, [b], [v], [g], [d]
[tiŋ]
[tïŋ]
The representation of suffixes follows Turkological traditions. Firstly, all suffixes will be marked
with an initial dash (-abc). Secondly, as a general rule of description, alternating phonemes, i.e. those
34
that are within a harmony domain, are given in capitals, while non-harmonizing phonemes are in lower
case. For example, the genitive case suffix shown in Table 1.2 will be shown as -NIŋ because while the
initial two phonemes alternate according to the rules of harmony, the final [ŋ] always remains constant.
Thirdly, theme phonemes, i.e. initial phonemes of affixes that appear following certain final stem end-
ings, but do not appear following other final stem endings, are shown in brackets. To show an example
with a theme vowel, the converb B suffix will be shown as -(I)p, since following vowel-final stems, the
exponent is only [p] such as saqta-p ‘hide-CVB.B’. If the stem’s final phoneme is a consonant, the theme
vowel appears, which harmonizes with the stem’s backness, which results in the exponents [ïp] and [ip],
such as in bar-ïp ‘go-CVB.B’ and kel-ip ‘come-CVB.B’. Lastly, if a phoneme in a suffix cannot be for-
malized as an addition, such as the theme phonemes, but is rather in complementary distribution with
another phoneme, the alternating phonemes are both shown, separated with a slash. An example of this
is converb A, whose exponent following vowel-final stems is [y], but following consonants is either [a]
or [e], given that it harmonizes with the backness of the stem. This is shown as -y/A.
1.5.2 TAM suffixes
All Turkic languages predominantly rely on agglutination, whereby morphosyntactic features are
realized additively, as an ordered sequence of suffixes.
12
The most straightforward illustration might be
the so-called “longest word” in Kazakh.
13
(16) qanaɣat-ta-n-dïr-ïl-ma-ɣan-dïq-tar-ïŋ-ïz-dan
satisfaction-VBLZ-REFL-RECP-CAUS-PASS-NEG-GAN-PL-2-PL-ABL
‘[someone is] from those ones, whom you (plural) have not made [feel themselves] satisfied’
Suffixes usually have a number of allomorphs, and their distribution is governed by the backness
of the stem vowels and the quality of the last phoneme of the stem or the suffix they attach to (detailed
in §1.5.1). All derivations are expressed by suffixes, as well as possessive markers, person and number
agreement, noun cases, and some of the TAM expressions mentioned in this thesis. As detailed in Chap-
ter 2, TAM suffixes mostly appear as heads of an independent clause and they express a range of se-
mantic content. In the context of Kazakh, I understand independent verb forms (and clauses headed by
thereof) as ones bearing tense and person-number agreement, as well as clause-final position given the
very strict OV ordering of the language. Any other verb form is considered dependent. When heading
12
The typological classification of languages regarding their means of expressing morphosyntactic features
(i.e. agglutinative, isolating, flexive and incorporating) has been widely used since Schleicher (1850), and it was
challenged for its shortcomings, e.g. by Sapir (1921).
13
https://www.quora.com/What-is-the-longest-word-in-Kazakh, accessed on 29/11/2021. As most of the
“longest words”, this one is grammatical, but hardly ever used spontaneously.
35
an independent clause, TAM suffixes are tensed and agree with the subject, as shown in (17) with the
past tense suffix -DI.
(17) keše Berik-pen kezdes-ti-m
yesterday Berik-COM meet-PST-1SG
‘I met with Berik yesterday.’
As opposed to the exclusively independent -DI in (17), some TAM suffixes can appear as heads
of dependent clauses as well, and thus they are analogous to participles in English, such as suffix -ing
that marks (i) the progressive aspect with the auxiliary verb be (Peter is walking), and (ii) modifies
nominal phrases (The person walking over there is Peter). Such a TAM suffix in Kazakh is -(y)AtIn
2.3 and 2.7), which in independent clauses expresses past habituality (18), while in dependent clauses
it expresses the prospective aspect (19).
(18) bala kez-im-de Almatï-ɣa bar-atïn-bïz
child time-1SG-LOC Almaty-DAT go-ATIN-1PL
‘In my childhood, we used to go to Almaty.’
(19) Almatï-ɣa bar-atïn adam-dar sawalnama- toltïr-sïn
Almaty-DAT go-ATIN man-PL questionnaire-ACC fill-IMP[3]
‘Those people who are going to Almaty should fill the questionnaire.’
In addition to suffixal TAM expression, Kazakh has a large number of multiverb constructions in
its verbal paradigm. These multiverb constructions involve a lexical verb and an auxiliary verb. The
auxiliary verb may be a verbal lexeme that is used as any other lexical verb and has a lexical meaning.
However, in certain constructions, when following a lexical verb that stands in a converbial form, the
lexical meaning of the auxiliary is bleached and it expresses a grammatical category. Or, the multiverb
construction can involve an existential verb, such as bol ‘be, become’ or e ‘be’, which appear in copula
constructions and take a nominal or adjectival complement, instead of a verbal one. Before turning to
AVCs, lexical verb + existential verb constructions are briefly discussed as they will appear in combi-
nation with AVCs, e.g. in §2.6.
One of the existential verbs is bol ‘be’, which is used in both nominal and verbal predication. In
nominal predication, the existential verb (labeled as COP[ula]) appears in non-present reference, for ex-
ample in the past tense example in (20).
(20) 1990 ïl-ï bala bol--m
1990 year-3 child be-PST-1SG
‘In 1990 I was a child.’
36
In verbal predicates, existential verbs can combine with participles to express some TAM cate-
gory. They appear in various parts of the verbal paradigm, including the expression of anteriority (de-
tailed in §2.6) and the apodosis of irrealis conditional sentences (§3.4.7). In the following example, the
combination of the anterior marker -GAn and the future tense marked existential verb bol expresses
anteriority in future.
(21) üy-ge kel-gen-še äke-m uyqï--nan
home-DAT come-GAN-ADVZ father-1SG sleep-3-ABL
uyan-ɣan bol-a-
wake.up-GAN be-NPST-3SG
‘By the time I get home, my father will have woken up from his sleep.’
While existential verb constructions are vital in the Kazakh TAM system, they are not the main
focus of this work. Their common peculiarity with AVCs, the main concern of this thesis, is that they
are both multiverb constructions expressing a TAM contribution. Similar to English, existential verb
constructions feature some dependent verb form and an existential verb, while an AVC consists of one
of two converbs and an auxiliary verb, as detailed in the following section.
1.5.3 Auxiliary Verb Constructions from a cross-linguistic perspective
A few words are due before the discussion of Kazakh AVCs. Anderson (2006, 4) defines auxiliary
verbs as systematically bleached, grammaticalized items on the lexical verb
functional affix contin-
uum. This vague definition is motivated by the fact that auxiliary verbs and AVCs are extremely diverse
across languages, in terms of morphology, syntax and semantics. Cross-linguistically, a prototypical
auxiliary verb conveys only grammatical information, even if in an earlier variation of the language, the
auxiliary lexeme had a lexical meaning. Additionally, the auxiliary verb often requires the lexical verb
it combines with, to take some defined form. This is shown in (22), where in a., the English auxiliary
should contributes a modal category, governs the base form of the verb (i.e. *should helping, *should
helped) and while in Modern English, should is not used as a lexical verb, it is known that in Old English
it was a lexical verb sculan ‘to owe, be under obligation’ (Klein 1971, 667). In (22)b, the auxiliary verb
dii-ja ‘sit’ has a lexical meaning, but in combination with another lexical verb, it expresses the continu-
ative, similarly to the Kazakh AVC CVB.A + ber ‘give’ discussed in §6.3.1.
37
(22) a. I should help Peter.
b. Kayardild, Tangkic, Australia (Evans 1995, 312)
wuu-ja yurda-ya muyinkalan-ki, yiiwi-ja dii-j
put-ACT inside-LOC dinghy-LOC lie-ACT AUX(‘sit’)-ACT
yurda-y, warra-n-marri, barri-n-marri
inside-LOC go-N-PRIV crawl-N-PRIV
‘(I) put (the turtles) inside the dinghy, and (they) just stayed lying there, without
moving, without crawling around.’
For the purposes of this thesis, we should make a comparison between AVCs and two similar
categories, serial verbs and light verbs. AVCs and serial verb constructions (SVCs) are similar in that
both are generally treated as monoclausal complex predicates describing a “single event”. According to
Anderson, the key difference is that in SVCs, the verbal components of the construction either retain
their lexical meanings or allow for a reading where the lexical meanings of the components are compat-
ible with the expressed predicate.
(23) Engenni (Atlantic-Congo, Nigeria) (Lord 1993, 227)
ò kpei dhe me
he wash finish me
He finished washing me.
According to Kroeger (2004, 22729), SVCs gravitate toward particular functions, such as the
voice alternating benefactive, directional specification with motion verbs, resultatives and various at-
tributives. The following example can be analyzed as the serialized verb die expresses the completive
aspect but notice that the lexical content ‘die’ persists.
(24) Tok Pisin (Pacific Creole, PNG) (Kroeger 2004, 228)
ol i-sutim pik i-dai
they shoot pig die
‘They shot the pig dead.’
While Kroeger does not propose a definition of SVCs, he lists the prototypical characteristics.
The various syntactic tests in Kroeger’s paper include tests on scope of negation, constituency tests by
circumfixes, independent negation, the scope of TAM, Wh-extraction and relative clauses.
1. two or more morphologically independent, non-auxiliary verbs within the same clause
2. no overt markers of co- or subordination
3. single intonation contour
4. the SVC describes a “single event”
5. a single TAM and agreement marking, possibly distributed over the verbs
38
6. the verbs have at least one common argument
7. the SVC does not contain two overt NPs that refer to the same argument
8. the verbs refer to one grammatical subject
While SVCs can be formally identical to AVCs, semantic analysis provides a basis that, as shown
in this thesis, auxiliary verbs do not retain their original lexical meaning (see e.g. §3.5). Light verbs may
also be confused with auxiliaries, as just like SVCs, they can be formally identical (for a comprehensive
study see Butt 2010). The prototypical light verb, however, combines with a nominal and acts as a
verbalizer with the possibility to contribute additional grammatical categories, such as valence. Most
Turkic languages have light verbs, such as the Turkish et and ol. They most often combine with nominals
of foreign origin and et provides a transitive or intransitive verbal reading, while ol always serves as an
intransitive light verb (Johanson 2021, 57882). In (25), the Arabic origin noun tahliye ‘evacuation’
may combine with both light verbs.
(25) Turkish, Oghuz Turkic
a. tahliye ol-mak
evacuation LV-INF
‘to be evacuated’
b. tahliye et-mek
evacuation LV-INF
‘to evacuate’ (transitive)
Cross-linguistically, verb+verb constructions have been analyzed as containing a light verb, such
as in the following example.
(26) Urdu, Indo-Iranian (Butt 2010, 63)
naadyaa makaan banaa pa-ii
Nadya.F.NOM house.M.NOM make fall-PRF.F.SG
‘Nadya fell to building a house.’
Light verb and auxiliary verb constructions are best thought of as different phenomena that as
close to each other on the lexical verb
functional affix continuum and it might be impossible to es-
tablish a cross-linguistically valid boundary (Anderson 2006, 1617). As such, the boundary between
light verbs and auxiliary verbs should be drawn within a given langue (Butt 2010, 74). In Turkic the
distinction appears to be very clear light verbs take nonverbal complements and specify transitivity,
while auxiliary verbs take verbal complements and specify a plethora of TAM categories.
39
1.5.4 Auxiliary Verb Constructions in Kazakh
AVCs express various TAM categories, such as perfectivity, completeness, benefaction, and abil-
ity. This work aims to describe every single-auxiliary AVC in Modern Spoken Kazakh (Chapter 3-6)
with enough precision that allows us to account for their distribution. In cases of apparent synonymy, it
will be shown that there are, indeed, contrasting factors. The analysis will describe 18 auxiliary verbs.
Some of these auxiliary verbs combine with one or both of the mandatory converbial forms of the lexical
verb, yielding a total of 28 AVCs.
In the following, auxiliary verbs and AVCs are defined in the context of Kazakh. An auxiliary
verb is a member of a closed class of verbal lexemes that can convey lexical meaning as any normal
verb or grammatical meaning. When conveying a grammatical meaning, their lexical meaning is
bleached and they take a verbal complement, which may be a lexical verb or another auxiliary verb. The
verbal complement must stand in a converbial form. The converb + auxiliary verb(s) constitute an aux-
iliary verb construction (AVC). Given that an AVC may have multiple auxiliary verbs, it must be
stressed that each AVC must contain exactly one lexical verb. Auxiliary verbs arguably constitute a
synchronically closed class because (i) the number of lexemes that fit the definition is stable in a given
time and variety, (ii) auxiliaries can function as lexical verbs when not in an AVC, and (iii) across Turkic
varieties and in diachrony, the set of auxiliaries fluctuate, i.e. there must be mechanisms that allow the
inclusion and exclusion of novel members (see §7.3.1 for the formalization of this assumption).
An AVC consists of exactly one lexical verb that stands in a converbial form and at least one
auxiliary verb. If the AVC features one auxiliary verb, it is inflected as any main verb would, expressing
all the necessary morphosyntactic features and agreement exponents (such as tense and person-number
agreement as shown in (17) above). If there is more than one auxiliary verb, only the last one inflects as
main verbs, the previous one(s) stand in one of the converbial forms (stacked AVCs are detailed in
§1.5.9). The order is fixed as [lexical verb<auxiliary verb]. Nothing can intervene between the lexical
and the auxiliary verb. For each lexical verb, there are two converbial derivations in Kazakh, labeled as
converb A and converb B.
14
Given that in an AVC, structurally, only the lexical verb varies, all AVCs
discussed will be referred to as a combination of the type of converbial form and the auxiliary verb, i.e.
[CONVERB A or B + auxiliary stem ‘original meaning’] e.g. the benefactive AVC in §6.3.2 will be given
as CVB.B + ber ‘give’. The following is a simple example.
14
Apart from converb A and B, there is a third converb suffix -GAlI which may combine with the four
imperfective auxiliaries (Chapter 3) (Muhamedowa 2015, 54; 136; Ótott-Kovács 2015, 20). These constructions
denote a form of the prospective aspect. Since the limited distribution, the consensus of the literature and for spatial
considerations, I do not include constructions with -GAlI in the discussion.
40
(27) Aygerim Berik-pen söyles-ip atïr
Aygerim Berik-COM speak-CVB.B AUX(IPFV ‘lie’).PRS[3]
‘Aygerim is talking to Berik [right now and in any posture, i.e. not necessarily lying].’
The lexical verb (in the example above: ‘speak’) can either stand in the converb A, or the converb
B form, detailed below in §1.5.5 and §1.5.6, respectively. Converb selection is a complex topic as a
given auxiliary verb may consistently select only one converb, or both with completely different TAM
contributions, while with other auxiliary verbs, the converb can alternate and express boundedness
these will be detailed in the sections on each auxiliary verb. Notice in the glosses that the auxiliary is
specified for the TAM category it expresses (in the example above: IMPERFECTIVE [IPFV]), as well as its
original lexical meaning, ‘lie’. Every auxiliary verb originates in a regular verb that has a lexical mean-
ing, and most of them can also function as such, either on their own, as a single verb (28), or in an AVC,
in the role of the lexical verb (29). While the examples below clearly show that these lexemes can con-
tribute lexical content, the translation in the example above in (27) shows that when these lexemes func-
tion as an auxiliary, the lexical content is bleached, and the construction, including the auxiliary verb,
expresses a grammatical category. Semantic bleaching is understood as the grammaticalization process
whereby a lexeme partially or completely loses its lexical content, which is replaced by a grammatical
relation, such as a TAM value (Hopper and Traugott 1993, 9498; Anderson 2006, 56). This is apparent
in (29), which shows that in an AVC, the auxiliary and the lexical verb can be the same lexeme.
The semantic bleaching (i.e. the disappearance of the lexical meaning when the lexeme expresses
a grammatical category) is also
(28) Aygerim sol aq-ta ür
Aygerim that place-LOC walk.PRS[3]
‘Aygerim is walking in that direction.’
(29) Aygerim qazir Berik-pen ür-ip ür
Aygerim now Berik-COM walk-CVB.B AUX(IPFV ‘walk’).PRS[3]
‘Aygerim is dating [“walk with”] Berik nowadays.’
AVCs are arguably monoclausal as they pass commonly accepted tests, such as replacement,
clefting, relativization and passivization, as shown in (30)-0. Note that the four tests use four different
AVCs. For tests on constituency in complex predicates in general see Butt (1995; 2003), and for Turkish
see Çetinoğlu, Butt, and Oflazer (2008). Apart from these brief tests, it must be mentioned that AVCs
are commonly thought of as monoclausal objects across languages (Anderson 2006, 321).
41
In the replacement test, the AVC can be replaced by a single verb construction
(30) replacement test
a. kitap- oqï-p tasta-
book-ACC read-CVB.B AUX(H.ENER)-PST[3]
‘S/he read the book quickly.’
b. kitap- oqï-
book-ACC read-PST[3]
‘S/he read the book.’
In the clefting test, in (31)a. the AVC is nominalized and is a complement to the matrix verb get
offended. In (31)b. the same AVC appears but now as a nominalized equative predicate which is identi-
fied with the also nominalized predicate get offended.
(31) clefting test
a. söyle-y ber-u-i-ne ašulan--q
talk-CVB.A AUX(CONT)-INF-3SG-DAT get.offended-PST-1PL
‘We got offended that s/he kept talking.’
b. biz-diŋ ašulan-ɣan-ïmïz
1pl-GEN get.offended-GAN-1PL
söyle-y ber-u-i e-di
talk-CVB.A AUX(CONT)-INF-3SG COP-PST[3]
‘Why we got offended was that he kept talking.’
In the relativization test, the same AVC appears as an independent verbal predicate and as the
head of a relative clause, realized as the attributivizer function of the polyfunctional suffix -GAn.
(32) relativization test
a. Aygerim temeki eg-ip atïr
Aygerim cigarette pull-CVB.B AUX(IPFV ‘lie’).PRS[3]
‘Aygerim is smoking.’
b. temeki eg-ip at-qan qïz Aygerim
cigarette pull-CVB.B AUX(IPFV ‘lie’)-GAN girl Aygerim
‘The girl who is smoking is Aygerim.’
42
Lastly, in the passivization test the AVC appears first in the active voice, and then in the passive
voice. The passive marker -(I)l appears once, on the right edge of the AVC, which is an argument for
monoclausality.
passivization test
a. ana süt-ti bala-lar-ɣa i-kiz-ip qoy-
mother milk-ACC child-PL-DAT drink-CAUS-CVB.B AUX(IAM)-PST[3]
‘The mother has already made the kids drink milk.’
b. süt bala-lar-ɣa i-kiz-ip qoy-ïl-
milk child-PL-DAT drink-CAUS-CVB.B AUX(IAM)-PASS-PST[3]
‘The milk has already been drunk by the kids [as per the request of someone].’
By default, nothing can intervene between the lexical verb and the auxiliary verb of an AVC. The
only exception is the clitic ɣana ‘only’ (Muhamedowa 2015, 129). It is treated as a clitic because it
exhibits consonant harmony, but not vowel harmony (allomorphs: ɣana, qana depending on the voiced-
ness of the final phoneme of the host).