Predication in uzbek and kazakh

Yüklə 162 Kb.
ölçüsü162 Kb.




Before any formal study of evidentiality in Uzbek and Kazakh may be undertaken, it is necessary to understand the processes that create complete predicates from verbs and other lexical categories. Predication occurs in a similar fashion in most of the Turkic languages, so the statements made here about Kazakh and Uzbek can be applied to most other members of the family as well.

Predication in the Turkic languages can be broken into two main types: verbal and non-verbal. Verbal predicates are characterized by their ability to take voice morphology, to be directly marked with negation, and to take certain TAM suffixes. Non-verbal predicates may be divided into five classes: nouns, adjectives, pronouns, existentials, and deontics. Compound predicates may be formed with the assistance of a copula (e- in both Uzbek and Kazakh). The copula is necessary to express negation and non-present/generic verbal categories on non-verbal predicates and to form complex past, perfect, and conditional forms.

The first section of this chapter discusses in greater depth the distinctions between the various verbal categories that may behave as predicates: verbs and the five aforementioned categories of non-verbs. The second section focuses on simple verbal predication and divides verbal morphology into two categories: finite forms, which occur only in predicates, and non-finite forms: participles, infinitives, and converbs, which occur both in predicates and in other syntactic positions. The third section covers the copula, which allows certain verbal categories to be expressed when the predicate is non-verbal or when the verb is non-finite. In the final section, I summarize the findings of the previous sections and outline the course of study for the rest of this work by specifying which pieces of the predicate are relevant to the study of evidentiality. The data presented in this chapter is primarily morphosyntactic and relates to the distribution of morphemes; the semantic properties of individual morphemes will be further expounded upon in later chapters. Full verbal paradigms for the verb meaning “do” (Uz: qil-, Kaz: qıl-) are provided in the Appendix.
2.1 Agreement Markers

In Uzbek and Kazakh, there are three productive agreement paradigms. In Turcological literature these are often referred to as possessive, pronominal, and converbial (Erdal 2004: 233). The possessive agreement paradigm is so-called because the forms resemble those of possessive markers, although the forms of the 1st person plural and the third person differ.

Table 1 - Uzbek Possessive Agreement and Possession Markers


Agreement Markers

Possession Markers

















Ø, -lar



Table 2 - Kazakh Possessive Agreement and Possession Markers


Agreement Markers

Possession Markers











- ŋIz





Ø, -LAr



The pronominal agreement markers are also named for what they resemble – personal pronouns. In Kazakh, the vowels of the pronominal agreement markers differ from those of the independent pronouns. In both Uzbek and Kazakh, the 3rd person marker is null and not based on the form of the 3rd person pronoun.

Table 3 - Uzbek Pronominal Agreement Markers and Pronouns


Agreement Markers

Personal Pronouns

















Ø, -lar



Table 4 - Kazakh Pronominal Agreement Markers and Pronouns


Agreement Markers

Personal Pronouns

















Ø, -LAr



Converbial markers (Table 5) occur only with forms derived from converbial forms and are nearly identical to the pronominal agreement markers. They differ, however, in that the marker for the third person is not null, but is a form derived from the verb meaning ‘to stand’ (Uz, Kaz: tur-), which is a remnant of their converbial use.

Table 5 - Converbial Agreement Markers























2.2 Predication and Lexical Categories

Predicates in Uzbek and Kazakh, and, indeed, in all Turkic languages, may be divided into two broad categories: verbal and non-verbal. Verbal predicates are characterized by their ability to be directly marked for the full range of verbal categories: voice/valency, negation, tense, aspect, mood, person, and number (1). Non-verbal predicates may only be directly marked for person and number agreement (in 2 the word ‘jealous’ is an adjective, not a verb); when person and number are the only features marked, non-verbal predicates receive a generic or present tense interpretation (3).

(1) Siz ko’r-il-ma-di-ngiz. (Uz)

Siz kör-il-me-di-ŋiz. (Kaz)
You see-pass-neg-pst-2pl
‘You (pl) were not seen.’

(2) *Biz qizg’anchiq-ish-ma-di-k. (Uz)

*Biz qizğanšaq-ıs-pa-dı-q. (Kaz)

We jealous-coop-neg-pst-1pl

‘We were not jealous of each other.’
(3) Men doktor-man. (Uz)
Men däriger-min. (Kaz)
I doctor-1sg

‘I am a doctor.’

A further difference between verbal and non-verbal predicates is that while verbal predicates select from the full range of three agreement paradigms (not including the irregular forms of the desiderative paradigm), non-verbal predicates, when they are directly marked for agreement, may employ only the pronominal agreement markers shown in Tables 3 and 4. Non-verbal predicates may be divided into five classes: nouns, adjectives, pronouns, existentials, and deontics.

2.2.1 Nouns and Adjectives

Nouns and adjectives form fairly discrete classes within the Turkic languages, particularly with regard to their syntactic behavior and the distribution of associated morphemes. Nouns are characterized by their ability to take plural morphology, to be modified by quantifiers and adjectives, to receive case and possession morphology, and to act as the arguments of verbs.

(4) Bu äyel däriger. ~ Bu däriger orïs.

This woman doctor This doctor Russian
‘This woman is a doctor.’ ‘This doctor is Russian.’

(5) Akram do’st-im. ~ Kecha do’st-ing-ni ko’r-di-m.

Akram friend-1sg Yesterday friend-2sg-acc see-pst-1sg
‘Akram is my friend.’ ‘Yesterday I saw your friend.’

Adjectives in Turkic are characterized by their ability to receive comparative and superlative markers and to be modified by certain intensifying adverbs (6), as well as to modify nouns (7).

(6) Astana ädemi. ~ Astana ädemi-rek. ~ Astana eŋ ädemi.
Astana beautful Astana beautiful-cmpr Astana most beautiful
‘Astana is beautiful’ ‘Astana is more beautiful’ ‘Astana is most beautiful.'

(7) Men baland daraxt-ni ko'r-di-m. ~ Daraxt baland.

I tall tree-acc see-pst-1sg tree tall
'I saw the tall tree.' 'The tree is tall.'

Nouns marked with the locative case behave semantically like adjectives, in as much as they modify nouns. Syntactically, they pattern with non-verbal predicates when they are used predicatively, but in order for them to be used attributively, a special morpheme must be inserted (Uz: -ki, Kaz: -GI):

(8) Prezident poytaxt-da. ~ poytaxt-da-gi kishi
President capital-loc capital-loc-attr person
'The president is in the capital' 'person (who is) in the capital'

(9) Atıraw-da-ğı sawda želi-si [1]

Atyrau-loc-attr commerce system-3
‘The commerce system in Atyrau.’

In many Turkic languages nouns and adjectives may be distinguished from one another through the distribution of intensifying and comparative morphemes. That is, morphemes expressing concepts like ‘more’ and ‘very’ are expressed differently depending upon lexical category. Uzbek and Kazakh are among the Turkic languages that express this distinction – Turkish is not.

Table 6 - Intensification and Comparison

Intensifier (very, many)

Comparative (more)






köp adam

öte žaqsï

köb-irek adam



ko’p odam

juda yaxshi

ko’p-roq odam



çok adam

çok iyi

daha adam

daha iyi


‘many men’

‘very good’

‘more men’


Adjectives may occasionally function as nouns and receive nominal morphology; when they do, they denote entities characterized by the features expressed by the adjectives from which they are derived: qizil-lar ‘the red ones’ (Uzbek), ädemi-ler ‘the beautiful (people)’ (Kazakh). Nouns in Turkic, as a rule, cannot function as adjectives.

2.2.2 Pronouns

Pronouns form a discrete class from other lexical categories. Much like nouns, they are marked for case (although irregularly) and act as the arguments of verbs (10, 11). Pronouns cannot co-occur with intensifying or comparative morphology.

(10) Men sağan bu-nı ber-di-m.1
I you(sg).dat this-acc give-pst-1sg
‘I gave you this.’

(11) Kim sen-ing qiz-ing ko’r-gan e-di?

Who you(sg)-gen daughter-2sg see-perf cop-pst.3
‘Who had seen your daughter?’

Other classes of pronouns exist in Uzbek and Kazakh as well. These classes include the interrogative pronouns (e.g Uz, Kaz: kim ‘who’, Uz: nima, Kaz: ne ‘what’), indefinite pronouns (e.g. Uz: bari, Kaz: bäri ‘everyone’, Uz: allakim, Kaz: äldekim ‘someone’), reflexive pronouns (e.g. Uz: o’z, Kaz: öz, ‘self’) and deictic pronouns (e.g. Uz: bu, Kaz: bul ‘this’, Uz: osha, Kaz: osı ‘that there’).

Any pronoun may function as a predicate. When the predicate is the first or second person pronoun, any copular form that follows may show agreement with the predicate, rather than the subject (13).

(12) Bu nima?

This what
‘What is this?’

(13) Sol men e-di-m. (Äwezov 1948: 112)

That I cop-pst-1sg
‘That was me.’

2.2.3 Existentials and Deontics

Two classes of non-verbal predicates are not easily classified as nouns, adjectives, or pronouns, as they almost never occur in non-predicative positions. The first of these classes is the existentials, cognates of which are found in all Turkic languages: PT *bār (Kaz: bar, Uz: bor), the existential, and PT *yōq (Kaz: žoq, Uz: yo'q), the negative existential. Aside from merely indicating the existence (or non-existence) of a thing (14), these forms are also used with possessive forms to indicate possession (as the Turkic languages have to verb equivalent to English ‘have’ – see 15), may be used as replies to questions, i.e. 'yes' or 'no' (16), and sometimes interact with parts of the verbal paradigm to indicate emphasis or negation.

(14) Tog'-lar-da bars bor.
Mountain-pl-loc leopard exist
'There are leopards in the mountains.'

(15) Uch bola-m bor.
Three child-1sg exist
'I have three children.'

(16) Asqar kel-di me? -Žoq, kel-me-di.

Asqar come-pst q? No come-neg-pst
'Did Askar come? -No, he did not come.'

Deontic predicates express necessity or permission. Three members of this class are found in Uzbek and Kazakh2: the Turkic *kerek (Uz: kerak, Kaz: kerek) 'necessary', the Arabic loan lāzim (Uz: lozim, not present in Kazakh) 'necessary', and the Arabic loan mumkin (Uz: mumkin, Kaz: mümkin) 'possible, permissible'.

(17) Uy-ga bor-ish-im kerak/lozim/mumkin.
Home-dat go-nmlzr-1sg necessary/necessary/possible
'I need to/need to/can go home.'

As previously described, all five types of non-verbal predicates are directly affixed with the pronominal agreement markers. If a non-present or non-generic reading is to be expressed, the copula (which is described in 2.3) must be added in order to bear further TAM marking.

2.2 Verbal Predication

Regardless of whether a finite or non-finite verb form is used, the basic ordering of affixes is as follows:

(18) Root + Voice/Valency + Negation + Tense/Aspect/Mood + Agreement

There are five morphemes that fall under the voice/valency category: the passive, causative, reciprocal, and cooperative forms. In Uzbek, the form of the negative is –ma, in Kazakh, -MA. Tense/aspect/mood forms are discussed below. Agreement markers agree with the subject of the verb and are marked for three persons, two numbers (singular and plural) and, in the 2nd person, for two degrees of formality (informal and formal). The question particle (Uzbek: mi, Kazakh: MA) usually follows the verb. Although the question particle is orthographically separate in both languages, it behaves as a clitic and may be considered part of the verbal complex.

2.2.1 Finite Verbal Morphology

In Uzbek and Kazakh, there are only three paradigms that are unambiguously finite: the desiderative, the simple past, and the conditional.

What I refer to here as the ‘desiderative’ paradigm is not, strictly, a paradigm, as it involves three distinct ranges of deontic meaning that vary by person. The term volitional or hortative is perhaps a better descriptor first person forms, which express a desire or suggestion to oneself or ones group. The term ‘imperative’ is a the standard description of the second person forms, and the third person forms, which express an indirect command or wish, are usually termed ‘optative’ or ‘voluntative’ (Johanson 2009: 489-91). Although these forms do vary in meaning, their mutual exclusivity and expression of deontic modality have led scholars of the Turkic languages to group them together under a single paradigm (c.f. Koç & Doğan 2004 for Kazakh, Coşkun 2000 for Uzbek; Erdal 2004 groups these forms together as ‘volitional’).

Table 7: Desiderative Paradigm of 'Do'









qil-ay, qil-ayin





qil, qil-gin




qıl, qıl-ıŋız








The volitional paradigm does not interact with evidentiality.

The other two finite paradigms – the simple past and the conditional – are distinguished from all others through their use of the possessive personal markers (Tables 1 and 2). Etymologically, the past tense is proposed to have developed from a verbal noun, which explains the presence of the possessive markers (Erdal 2004: 238). The origin of the conditional is unclear. In modern Uzbek and Kazakh, however, both the simple past and conditional always function as predicates, so they must be considered truly finite.

Verbs inflected with the conditional form correspond roughly to English forms marked with ‘if’ and indicate irrealis mood.

(19) Qaš-sa-m qutıl-a al-ar e-mes-pin. (Koç & Doğan 2004: 355)
Run-cond-1sg escape-cvb cop-neg.aor-1sg
‘If I run, I will be unable to escape.’
Table 8 - Conditional Paradigm of 'Do'























The simple past tense contrasts with the perfect, which will be discussed in the following section, not only in terms of the past/perfect distinction, but also in terms of epistemic modality. This distinction forms the basis of the third chapter.

Table 9 - Simple Past Paradigm of 'Do'























The conditional and simple past are the basis for two of the five copular forms that are discussed in 2.3.

2.2.2 Non-Finite Verbal Morphology

Non-finite verbal forms employed as predicates may be divided into three classes based on their non-predicative function: participial, infinitival, and converbial. All of these forms (with a partial exception in the case of converbial forms) take the pronominal agreement markers (Tables 3 and 4).

Participial forms are the some of the most prominent types of non-finite forms that may act as predicates. As their name implies, participles may function attributively or predicatively:

(20) žaz-ar adam ~ adam žaz-ar.
write-aor man ~ man write-aor
'The man who writes.' 'The man writes.'

Although participles function much like adjectives, their verbal basis allows them to take voice and negation morphology:

(21) žaz-ıs-pa-ğan adam-dar ~ adam-dar žaz-ıs-pa-ğan
recp-neg-prf man-pl ~ man-pl write-recp-neg-prf
‘The men who haven’t written to each other’ ~ ‘The men haven’t written to each other.’

A table of participles shows that Uzbek and Kazakh share a number of cognate forms:

Table 10 - Participles in Uzbek and Kazakh






















Note that in Uzbek, two forms end in –
gan: -adigan and –ayotgan. The –gan portion of these forms is cognate with the –gan that forms the perfect participle. However, no perfect meaning is expressed in either of these forms, which instead likely grammaticalized from converbial constructions before –gan acquired its perfect meaning and instead expressed imperfectivity. The cognate of this form still expresses imperfectivity in modern Turkish, where it has the form –An (c.f. Erdal 1991, 2004).

The –mish/-mIs forms are now largely obsolete in Uzbek and Kazakh, having been supplanted by –gan. However, as these forms are cognate with Turkish –mIş, one of the forms central to the expression of evidentiality in that language, they will be considered in the following discussions of evidentiality in Uzbek and Kazakh.

The presence of *-gAn as the marker of the perfect participle is one of the core features that distinguish the Central Asian Turkic area from other Turkic-speaking regions (Schönig 1999). Both the predicative (22) and participial (23) uses are common:

(22) Biz osha joy-ga bor-gan-da halol pitza ye-gan-miz.
We that place-dat go-nmlzr-loc halal pizza eat-prf-1pl
‘Whenever we’ve gon there we’ve eaten halal pizza.’

(23) Odan äri ärkim žet-ken žer-i-ne qon-ar.
Then after everyone arrive-
prf place-3-dat camp-aor.3
‘After that, everyone encamps at the place where they arrive.’

For historical reasons, and for the sake of connecting the predicative and participial uses of this form, it will be glossed as prf, i.e. perfect. As noted above, the perfect contrasts with the simple past in a number of ways that will be further explained in the following chapter.

Another use of this form is to nominalize clauses for the purposes of complementation or in order to attach a case ending. This use is seen in (22) above; an example of its use to form clausal complements is in (24).

(24) Men-i tehnologiya ko’r qil-gan-i-ni bil-a-man.
acc technology blind make-nmlzr-3-acc know-pres-1sg
‘I know that technology makes me blind.’

When the * gAn suffix is used for the purposes of nominalization, it will be glossed as nmlzr, i.e. nominalizer, rather than perfect, as the time reference of these nominalized clauses is context-dependent.

The infinitive (Uz: -moq, Kaz: -Uw, -MAK) is used to form a number of non-finite forms that may be used predicatively. On its own, the Kazakh infinitive marker –MAK may be used to indicate a definite future tense. In both languages, the addition of the agentive marker (Uz: -chi, Kaz: -šI) creates forms that also indicate future tense and contributes a sense of intent.

Table 11 - Predicative Forms Based on the Infinitive



Definite Future


Intentional Future






(25) Men bar-maq-šı-mın.
I go-inf-agt-1sg
‘I will go.’ ‘I intend to go.’

The addition of the locative case (Uz: -da, Kaz: -DA) creates forms that indicate near future tense or inchoative aspect.

(26) Men bor-moq-da-man.
I go-
‘I’m going to go,’ ‘I am about to go.’

The third type of non-finite verb form that may be used predicatively is the converb. In Uzbek and Kazakh there are two converbial forms, which are commonly assigned the labels ‘imperfective’ or ‘intraterminal’ (Uz:
a/y, Kaz: A/y) and ‘perfective’ or ‘postterminal’ (Uz: -(i)b, Kaz: -(I)p), although the choice between these two is often governed by the following verb and not aspectual considerations4.

Table 12 - Converbial Forms









Converbs are set apart from other non-finite predicates by their use of the converbial agreement markers (Table 5).

Converbs are V + V constructions that behave in many ways like serial verbs in other languages. What sets them apart from true serial verbs is the presence of converbial suffixes noted above. The presence of these forms violates the restriction that serial verbs not be connected by any sort of “overt marker of coordination, subordination, or syntactic dependency of any sort” (Aikhenvald 2006:1, Mufwene 1990, cf. Stewart 2001, Haspelmath 1995). Converbs in Turkic express the same sorts of relations between events that serial verbs express, such as simultaneity, cause and effect, consecutivity, or even description of one event by another. Also like serial verb constructions, the two verbs obligatorily share a subject and may or may not share an object.

(27) Murodjon gugurt chaq-ib vklyuchatel-ni top-di.

Murodjon match strike-CVB switch-ACC find-PST.3SG
‘Murodjon lit a match and found the switch.’

Perhaps the most interesting aspect of converb constructions, through, is that they have grammaticalized into a closed class of special forms that express modal, aspectual, and other similar information for which no appropriate verbal ending had previously existed. In effect, many Turkic converbs act as light verbs (Bowern 2004) or like English modal verbs.

(28) kiril alifbo-si-da tarjima qil-ib ko'r-di-m [2]
Cyrillic alphabet-3-loc translate do-cvb see/try-pst-1sg
'I tried to translate it into the Cyrillic alphabet.'

(29) bar-a žat-qan-mïn (Koç & Doğan 2004: 264)
go-cvb lie/prog-prf-1sg
'I am going.'

In examples (28) and (29), the first (and main) verb is marked with the converbial marker, while the second verb receives all TAM marking. In most cases, the second verb is form-identical to a main verb (in these cases 'see' and 'lie down'), but when used in converbial constructions, these verbs have modal or aspectual meanings.

The predicative form based upon the imperfective converb is the basis for the present/future tense in both Uzbek and Kazakh and is used as an unmarked form when discussing current or near-future events (30).

(30) O’q-ish-ni endi bitir-a-man. (Coşkun 2000: 141)
nmlzr-acc now finish-pres-1sg
‘I’m now finishing reading.’

In its predicative usage, the imperfective converb has grammaticalized to such an extent that it no longer may co-occur with any copular forms.

The perfective converb, when used predicatively, functions as a perfect (31), and unlike the imperfective converb, may co-occur with copular forms (32).

(31) Qazaq-tıŋ žaqsılığ-ın köp kör-ip-siŋ. [3]
Kazakh-gen goodness-3.acc much see-prf-2sg
‘You have really seen the goodness of the Kazakhs.’

(32) Sen tım keš kel-ip e-di-ŋ. [4]

You too late come-prf cop-pst-2sg
‘You had come too late.’

2.3 Copular Predication

A feature that unites non-verbal and non-finite verbal predicates is their ability to take on a copula in order to express the conditional, negation, the past, or the perfect. In both Uzbek and Kazakh, the copula has the form e-, which is derived from Old Turkic er-. In Uzbek and Kazakh, only five forms of this copula are available. These forms correspond to previously discussed forms that are directly affixed to verbs.





Verbal form

Uz: -ma

Kaz -MA

Uz: -sa

Kaz; -sA

Uz: -gan, -mish

Kaz: -GAn, -MIs

Uz; -di

Kaz: -DI

Copular form

Uz: emas

Kaz: emes

Uz: esa

Kaz: ese

Uz: ekan, emish

Kaz: eken, emis

Uz: edi

Kaz: edi

In Uzbek, the first vowel of copular forms is occasionally omitted in rapid speech or following the question particle.

The negative form of the copula is related to the aorist participle (Uz: -ar, Kaz: -Ar), whose negative form is suppletive (Uz: mas, Kaz: -MAs). The positive aorist copula existed in Old Turkic as erür, but it has since been lost in the modern Turkic languages (Erdal 2004). As a copula, emas/emes is purely negative and any aspectual meaning must be contributed by another morpheme.

The conditional bears the same irrealis meaning whether it is attached directly to a verb or to the copula. The copular conditional is sometimes used in a non-predicative function to indicate topic, functioning similarly to the English phrase as for or like a cleft.

(33) Prezident esa o‘n olti nafar senator-ni tayinla-y-di. [5]
President foc ten six cl senator-acc appoint-pres-3sg
‘As for the president, he is appointing sixteen senators.’
The perfect/evidential copulas express a wide range of markedly non-confirmative meanings, including admirativity, non-firsthand evidentiality, and reportativity. In questions, these same forms express rather unusual pragmatic meanings: they question the authority of the addressee of the question and are used to create rhetorical questions. These topics receive full treatment in the fourth and fifth chapters, respectively.

Friedman (1978, 1979) has noted that in many languages of Eurasia, the pluperfect (that is, double marking of the perfect) exhibits strong non-confirmative meaning, as shown in (34).

(34) Böžey awl-ına žet-ken e-ken. (Äwezov :148)
Böžey village-3.dat arrive-perf cop-perf
‘Böžey has (apparently) arrived at his village.’
Under the analysis here, the double marking of the perfect matters less than the presence of the copular form of the perfect. That is, the copular perfect bears marked non-confirmative meaning in most places it is present, whether the preceding form is in the perfect, a non-finite verb, or a non-verb.

There is some debate regarding the origin of the ekan/eken copular form; Erdal (1991:383, 2004:288, 320) has proposed that it might ultimately derive from the combination of the copula and the Old Turkic morpheme –KAn, which was used to create temporal clauses. The form of this morpheme is, indeed, unexpected, as it is expected that the /g/ of –gAn would not surface as /k/ following either a vowel or the /r/ that was present in the Old Turkic form (er-ken). This debate is irrelevant for the study of the modern languages, as it is possible to draw parallels between the non-copular and copular forms of the perfect/evidential forms. Just as the simple past and perfect contrast in their non-copular forms, so too do edi/edi and ekan/eken contrast as copular forms. Further support is lent to this connection by the ability of the perfect/evidential form to nominalize verbless clauses, just as the perfect is employed to nominalize clauses when a verb is present.

(35) U Stiven-ga ism-i Eddi e-kan-i-ni ayt-di… (Joyce: 13)
He Steven-dat name-3 Athy cop-nmlzr-3-acc tell-pst.3
‘He told Stephen that his name was Athy…”
(36) Tegin e-mes e-ken-in de bil-gen. (Äwezov 1948: 59)
Free cop-neg cop-perf-3.acc also know-perf
‘He also knew that it was not free.’
As is the case for clauses nominalized with the perfect, clauses nominalized with the perfect/evidential copula bear no perfect or evidential meaning; the sole purpose of this form is to nominalize and the copula acts a verb that can support this nominalizing morpheme.

The Turkic copula differs in two ways from the copula of English (and of other languages of Europe), which raises the question of whether what has been discussed is a true copula or something else. The most obvious difference between the copulas of English and Turkic is that the Turkic copula may appear more than once within a single predicate.

(37) Xursand e-mas e-di-m.

Happy cop-neg cop-pst-1sg
‘I was not happy.’

As a general rule, the negative copula precedes other forms of the copula, while the other three forms are usually mutually exclusive5.

This difference may not be so great, however, as there are a number of theories that claim that the English copula, much like the Turkic copula, is reducible to “one singular copular case of a semantically empty verb inserted to form a verb phrase out of a predicate phrase headed by a non-verb” (Mufwene 2005: 232, c.f. McCawley 1988: 135-6).

(38) Susan was being annoying.

As a semantically empty element, then, the copula can be conceived of as a “dummy element whose sole purpose lies in carrying verbal morphology in predicate phrases whose nucleus consists of a lexeme which is incompatible with verbal morphology” (Pustet 2003: 3). In examples (37) and (38), the copula exists merely to express certain morphology; in the English example, the second instance of be is not, in fact, a main verb with a meaning of “to act”, but the bearer of verbal morphology that prompts this interpretation by implicature.

The second difference is that the Turkic copula expresses only certain verbal categories, whereas that of English may appear with the full range. When it is necessary to express verbal categories other than those discussed above, another verb is employed (Uz: bo’l-, Kaz: bol-). Whereas the copula expresses states, this other form (in Pustet’s [2003] terminology, a pseudo-copula), expresses events.

(39) Xursand bo’l-ayotgan-man.
Happy be-prog-1sg
‘I am becoming happy; I am acting happy.’

The Dummy Hypothesis renders this difference between Turkic and English moot as well, as under this hypothesis, the copula (and therefore all instances of English ‘be’) is not a true verb denoting a state or activity, but something inserted to support verbal morphology; therefore, the distribution of this morphology is responsible for the behavior of the copula.

As the differences between English and Turkic can be explained by appealing to the Dummy Hypothesis of copulas, there is no reason to consider Uzbek and Kazakh e- anything but a true copula. Pustet (2003:2) lists three syntactic functions that are fulfilled by copulas, all of which apply equally as well to English as to Uzbek and Kazakh:

  1. the function of a linker between subject and predicate

  2. the function of a syntactic ‘hitching post’ to which verbal inflectional categories can be attached;

  3. the function of a predicator which is added to lexemes that do not form predicates on their own

As a sort of dummy element, the Turkic copula’s idiosyncrasies are not inherent properties of a lexeme e-, but rather the properties of the elements that the copula is associated with. While the English copula is required to express most non-verbal predicates, the Turkic copula is expressed only in the select cases discussed above.
2.4 Forms Relevant to Evidentiality

As discussed in Chapter 1, studies of evidentiality in Turkic, and, indeed, in many of the languages of Eurasia, focus on the contrast between the past and the perfect. The simple past is frequently analyzed as being marked for confirmativity, while the perfect is either unmarked for this feature or is markedly non-confirmative (Friedman 1978, 1979). In regard to the Turkic languages of Central Asia, Johanson (2003) proposes that there are four morphemes that express markedly evidential meaning. According to Johanson, two of these are attached directly to the verb: *-Ib(dIr), which functions like the simple past with no aspectual information given, and *-gAn, which bears a perfect or resultative meaning. The other two forms are copular: *erken (the copular form of *-gAn ) bears a variety of markedly non-confirmative meanings, while *ermiš indicates reportative or quotative meanings.

In the analysis of Uzbek and Kazakh in the following chapters, the forms discussed by Johanson are considered and are contrasted with the simple past, as proposed by Friedman. Chapter 3 addresses the contrast between the simple past and the three forms of the perfect: the (simple) perfect, the converbial perfect, and the archaic perfect in –mish/mIs.

Table 13 - Forms Relevant for Ch. 3

Simple Past


Uz: -di

Kaz: -dI


Uz: -gan

Kaz: -gAn

Converbial Perfect

Uz: -(i)b

Kaz: -(I)p

Archaic Perfect

Uz: -mish

Kaz: -mIs

Both Friedman and Johanson have noted that the copular forms of the perfect are somehow special, encoding information that is more classically evidential in meaning, than forms that are directly affixed to the verb. The analysis of copular forms provides the basis for the fourth chapter. In that chapter, the copular forms of the simple past, the perfect, and the archaic perfect are compared.

Table 14 - Forms Relevant for Ch. 4

Simple Past


Uz: edi

Kaz: edi


Uz: ekan

Kaz: eken

Archaic Perfect

Uz: emish

Kaz: emis

The analysis of Turkic proposed in this chapter, in which non-verbs and non-finite verbs and treated as morphologically equivalent, allows for a unified treatment of copular forms. Previous analyses of Turkic languages frequently overlook the similarity of non-verbs and non-finite verbs, so it is hoped that by unifying these two classes, the markedly non-confirmative meanings of copular perfects may be captured.


Aikhenvald, Alexandra Y. 2006. “Serial Verb Constructions in Typological Perspective.” In Serial Verb Constructions, edited by Alexandra Y. Aikhenvald and R.M.W. Dixon, 1-68. Oxford: Oxford UP.

Äwezov, Muxtar. 1948. Abay: Roman-ėpopeya. Almaty: Qazaqtıŋ Birikken Memleket Baspası.

Bowern, Claire. 2004. (Some Notes on) Complex Predicates in Turkic. In Heejeong Ko and Maryanne Walter (eds.), Proceedings of the First Workshop on Altaic Formal Linguistics. Cambridge: MIT Working Papers in Linguistics.

Coşkun, Volkan. 2000. Özbek Türkçesi Grameri. Ankara: TDK.

Erdal, Marcel. 1991. Old Turkic Word Formation. Wiesbaden: Harrassowitz.

Erdal, Marcel. 2004. A Grammar of Old Turkic. Leiden: Brill.

Friedman, Victor. 1978. On the Semantic and Morphological Influence of Turkish on Balkan Slavic. In Papers from the Fourteenth Regional Meeting of the Chicago Linguistic Society, eds. Donka Farkas et al., 108-118. Chicago: Chicago Linguistic Society.

Friedman, Victor. 1979. Toward a Typology of Status: Georgian and Other Non-Slavic Languages of the Soviet Union. In The Elements: A Parasession on Linguistic Units and Levels, Including Papers from the Conference on Non-Slavic Languages of the USSR, eds. Paul R. Clyne, William F. Hanks, and Carol L. Hofbauer, 339-350. Chicago: Chicago Linguistic Society.

Haspelmath, Martin. 1995. "The converb as a cross-linguistically valid category." In Converbs in cross-linguistic perspective, eds. M. Haspelmath & E. König, 1-55. Berlin: Mouton de Gruyter.

Johanson, Lars. 2003. Evidentiality in Turkic. In Aikhenvald and Dixon: 273-290.

Johanson, Lars. 2005. Iranian as buffer zone between the universal typologies of Turkic and Semitic. In Linguistic convergence and areal diffusion : case studies from Iranian, Semitic and Turkic, eds. Éva Csató, Bo Isaksson and Carina Jahani, 205-213. New York: RoutledgeCurzon.

Johanson, Lars. 2009. Modals in Turkic. In Modals in the Languages of Europe: a Reference Work, eds. Björn Hansen and Ferdinand de Haan, 487-510. Berlin: Mouton de Gruyter.

Joyce, James. Musavvirning Yoshlikdagi Shamoili [Portrait of the Artist as a Young Man]. Trans. Ahmad Otaboyev. Kutubxonesi. (accessed December 28, 2009).

Koç, Kenan, and Oğuz Doğan. 2004. Kazak Türkçesi Grameri. Istanbul: Gazi Kitabevi.

McCawley, James D. 1988. The Syntactic Phenomena of English. 2 vols. Chicago: University of Chicago Press.

Mufwene, Salikoko. 1990. Serialization and Subordination in Gullah: Toward a Definition of Serialization. In When Verbs Collide: Papers from the (1990) Mini-Conference on Serial Verbs, eds. Brian D. Joseph and Arnold M. Zwicky, 91-108. Columbus: Ohio State U, Dept. of Linguistics.

Mufwene, Salikoko. 2005. How Many Bes Are There in English? In Polymorphous Linguistics: Jim McCawley’s Legacy, eds. Salikoko S. Mufwene, Elaine J. Francis, and Rebecca S. Wheeler, 225-246. Cambridge, MA: MIT Press.

Pustet, Regina. 2003. Copulas: Universals in the Categorization of the Lexicon. Oxford: Oxford UP.

Schönig, Claus. 1999. The Internal Division of Modern Turkic and its Historical Implications. Acta Orientalia 52: 63-95

Stewart, Osamuyimen Thompson. 2001. The Serial Verb Construction Parameter. New York: Garland Publishing

Internet Data Sources

[1] Sberbank Rossii. (accessed March 8, 2010.)

[2] Forums. (accessed February 20, 2010.)

[3] Alaš Aynası. (Accessed February 27, 2010.)

[4] Žumağalıyqızı, B. Ä. Qazaq tili funktsionaldı stil’ mätinderindegi prosodikalıq täsilderdi. (accessed March 1, 2010.)

[5] (accessed March 1, 2010.)

[6] Azattıq Radiyosı. (accessed March 5, 2010.)

1 The dative form of the 2nd person pronoun in this example is irregular. Rather than the predicted sen-ga, the form is sağan.

2 Other Turkic languages have a number of different members in this class, which seems to be especially susceptible to borrowing, e.g. Sakha naada, from Russian nado ‘necessary.’

3 The negative form of the aorist is irregular. For Uzbek: -mas, for Kazakh: -mAs.

4 See Johanson 2005, Erdal 2004 for further discussion of these terms and their associated meanings.

5 The perfect/evidential form of the copula may follow the past or conditional forms in questions – see Chapter 5.

Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur © 2016
rəhbərliyinə müraciət

    Ana səhifə