Elamur: **Mova** · [Parse](https://mova.elamur.ai/free/en/parse/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Translate](https://md.translate.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · [Gali](https://md.gali.elamur.ai/free/en/) · [MCP](https://mcp.elamur.ai/) · [Locations](https://md.locations.elamur.ai/free/en/) · [Weather](https://md.weather.elamur.ai/free/en/) · [Constitution](https://md.constitution.elamur.ai/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/languages/kk/24-phrase-structure/)

[Mova](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [languages](https://md.mova.elamur.ai/free/en/docs/languages/) / [kk](https://md.mova.elamur.ai/free/en/docs/languages/kk/) / 24. Phrase structure

# 24. Phrase structure

**On this page**

1. [Noun phrase order and modifier scope](#noun-phrase-order-and-modifier-scope)
2. [Attributive modifiers and apposition](#attributive-modifiers-and-apposition)
3. [Verbal heads and auxiliary sequences](#verbal-heads-and-auxiliary-sequences)
4. [Adjective and adverb phrases](#adjective-and-adverb-phrases)
5. [Postpositional phrases and case](#postpositional-phrases-and-case)
6. [Attachment and competing constituent boundaries](#attachment-and-competing-constituent-boundaries)
7. [How Mova handles phrase heads and attachment](#how-mova-handles-phrase-heads-and-attachment)
8. [Sources for this chapter](#sources-for-this-chapter)

Kazakh phrases are generally head-final. A noun normally follows its attributive modifiers, a lexical verb precedes the auxiliary that follows it, and a postposition follows the expression it governs. Several different constructions therefore have the same surface order. Case, possessive marking, and verb form help show which words belong together and how the resulting phrase functions in the clause (Dryer and Haspelmath 2013, § 83A, § 85A–90A).

## Noun phrase order and modifier scope

The noun stands at the right edge of the ordinary attributive sequence. A demonstrative can precede a cardinal numeral, which can precede an adjective and the noun. This order is especially clear when all the modifiers apply to the same head. Each slot is optional. A numeral used directly to count the referents normally leaves the noun without a plural suffix. Case is expressed on the noun, after any possessive marking. The word order patterns for demonstratives, numerals, and adjectives are prenominal (Dryer and Haspelmath 2013, § 87A–89A).

The examples in this chapter are constructed. In the romanization, `ı` represents ы, `i` represents і, and `y` represents й. In the glosses, `POSS` marks possessive inflection, `PTCP` a participle, `CVB` a converb, and `PROG` a progressive use. Person is shown where it is relevant to the phrase.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| кітап | kitap | book | ‘a book’ |
| екі кітап | eki kitap | two book | ‘two books’ |
| екі жаңа кітап | eki jaña kitap | two new book | ‘two new books’ |
| осы екі жаңа кітап | osı eki jaña kitap | this two new book | ‘these two new books’ |
| осы екі жаңа кітапты | osı eki jaña kitaptı | this two new book-ACC | ‘these two new books’ as a direct object |
| осы екі жаңа кітабын | osı eki jaña kitabın | this two new book-POSS.3-ACC | ‘these two new books of theirs’ as a direct object |

In the last two rows, the accusative applies to the whole referring expression. The form of the adjective stays the same. The same is true of the demonstrative and numeral in this attributive use. The ending on the noun in the last row also identifies it as possessed, even with the possessor left implicit.

A genitive possessor normally precedes the possessed expression. It can itself be a phrase containing a demonstrative, numeral, adjective, or other modifier. The possessed noun carries possessive inflection in the ordinary genitive–possessive construction. Material before the genitive noun can therefore belong inside the possessor phrase. Material between the genitive and the possessed noun often modifies the possessed noun (Dryer and Haspelmath 2013, § 86A).

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| екі баланың дәптері | eki balanıñ däpteri | two child-GEN notebook-POSS.3 | ‘the notebook of two children’ |
| баланың екі дәптері | balanıñ eki däpteri | child-GEN two notebook-POSS.3 | ‘the child’s two notebooks’ |
| осы баланың дәптері | osı balanıñ däpteri | this child-GEN notebook-POSS.3 | ‘this child’s notebook’ |
| баланың осы дәптері | balanıñ osı däpteri | child-GEN this notebook-POSS.3 | ‘this notebook of the child’s’ |
| жаңа дәптердің мұқабасы | jaña däpterdiñ mūqabası | new notebook-GEN cover-POSS.3 | ‘the cover of the new notebook’ |
| дәптердің жаңа мұқабасы | däpterdiñ jaña mūqabası | notebook-GEN new cover-POSS.3 | ‘the notebook’s new cover’ |

The two uses of *екі* illustrate scope. In the first row, the numeral counts children, and the genitive on *баланың* closes the possessor phrase. In the second, the numeral counts notebooks. The same distinction can be seen with the adjective in the last two rows. A possessor itself can contain a possessor, as in *баланың дәптерінің мұқабасы* (*balanıñ däpteriniñ mūqabası*, child-GEN notebook-POSS.3-GEN cover-POSS.3, ‘the cover of the child’s notebook’). Here *дәптерінің* carries possessive marking for its relation to *баланың* and genitive case for its relation to *мұқабасы*.

In a dependency analysis, the noun that names the referent is the head of the ordinary noun phrase. Its adjectival modifier takes `amod`, a demonstrative used as a determiner takes `det`, and a cardinal that counts the noun’s referents takes `nummod`. In an embedded possessor phrase, these relations go to the noun that each modifier actually qualifies. For example, the numeral in *екі баланың дәптері* depends on *баланың*, while the numeral in *баланың екі дәптері* depends on *дәптері* (Universal Dependencies contributors 2025, § `amod`, § `det`, § `nummod`, § `nmod`).

## Attributive modifiers and apposition

An attributive adjective precedes its noun. It has no agreement suffix for the noun’s number, case, or possessor. An adjective phrase can occupy the same position as a single adjective. When a degree word modifies the adjective, it accompanies the adjective on the left of the noun. Two adjectives can both modify the noun, and a degree word can belong just to the adjective next to it. Coordination, punctuation, and the meanings of the adjectives help distinguish independent descriptions from a description of a more narrowly defined class.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| үлкен ақ үй | ülken aq üy | large white house | ‘a large white house’ |
| өте үлкен ақ үй | öte ülken aq üy | very large white house | ‘a very large white house’ |
| үлкен, жарық бөлме | ülken, jarıq bölme | large bright room | ‘a large, bright room’ |
| үлкен бөлмелерде | ülken bölmelerde | large room-PL-LOC | ‘in large rooms’ |

In *өте үлкен ақ үй*, *өте* modifies the adjective *үлкен*. The expression attributes whiteness to the house as well. The degree word is inside the adjective phrase, while the adjective is a modifier of the noun. In *үлкен бөлмелерде*, both plural and locative are on *бөлме*.

A noun can also stand before another noun. One pattern uses a bare noun to indicate material or kind, as in *тас үй*. A second has a bare first noun and a possessive suffix on the head, as in *мектеп директоры*. This pattern is frequent in names for roles and conventional relations. An overt genitive identifies a possessor in a fuller construction, as in *мектептің директоры*. The same English translation can be appropriate for forms with and without the genitive. In context, the bare first noun readily characterizes a type or institutional role, and the genitive readily singles out the possessor. The morphological distinction is visible even where that difference in interpretation is slight.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| тас үй | tas üy | stone house | ‘a stone house’ |
| мектеп директоры | mektep direktorı | school director-POSS.3 | ‘a school principal’ |
| мектептің директоры | mekteptiñ direktorı | school-GEN director-POSS.3 | ‘the principal of the school’ |
| мектептің жаңа директоры | mekteptiñ jaña direktorı | school-GEN new director-POSS.3 | ‘the school’s new principal’ |

An attributive clause also normally precedes the noun. Its participial verb heads the modifier, and the arguments and adjuncts inside that clause belong to the verb. The following noun can be understood in different roles inside the relative clause. In *кеше келген бала*, the child is understood as the one who came. In *Асан көрген бала*, the child is understood as the one Asan saw. The outer clause or phrase determines the overt case of *бала*. In UD, the relative clause is linked to the noun through `acl:relcl`, and an overt dependent such as *Асан* is analyzed within that clause (Dryer and Haspelmath 2013, § 90A; Universal Dependencies contributors 2025, § `acl:relcl`).

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| кеше келген бала | keşe kelgen bala | yesterday come-PTCP child | ‘the child who came yesterday’ |
| Асан көрген бала | Asan körgen bala | Asan see-PTCP child | ‘the child Asan saw’ |
| мектепте жұмыс істейтін мұғалім | mektepte jūmıs isteytin mūğalim | school-LOC work do-PTCP teacher | ‘a teacher who works at the school’ |
| кеше келген баланы көрдім | keşe kelgen balanı kördim | yesterday come-PTCP child-ACC see-PST-1SG | ‘I saw the child who came yesterday’ |

A role or title can directly precede a personal name, as in *дәрігер Айгүл*. A fuller explanatory nominal can follow its referent and form a separate intonation group. In writing, commas can mark this supplementary use. In the second example below, *біздің жаңа дәрігеріміз* has its own possessor and adjective, and the whole phrase supplies another description of *Айгүл*. The UD relation `appos` applies to a nominal in apposition to another nominal. Its use depends on that relation between the nominals. A title next to a name, a material noun next to the noun it classifies, and a second phrase naming the same referent need to be distinguished when assigning dependencies (Universal Dependencies contributors 2025, § `appos`, § `nmod`).

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| Дәрігер Айгүл келді. | Däriger Aygül keldi. | doctor Aygül come-PST | ‘Doctor Aygül came.’ |
| Айгүл, біздің жаңа дәрігеріміз, бүгін келді. | Aygül, bizdiñ jaña därigerimiz, bügin keldi. | Aygül we-GEN new doctor-POSS.1PL today come-PST | ‘Aygül, our new doctor, came today.’ |

## Verbal heads and auxiliary sequences

The lexical verb supplies the event or state and ordinarily follows its objects and other preverbal dependents. In a single finite verb, the same word carries the lexical stem and the finite morphology. In a common analytic construction, the lexical verb takes a nonfinite ending and is followed by a verb that contributes aspect or another grammatical meaning. With a finite auxiliary at the end, the tense and any overt person marking of that auxiliary appear there. This order coexists with the basic object–verb order of the clause (Dryer and Haspelmath 2013, § 83A).

The following forms show where the finite marking appears in a progressive sequence with *отыр*.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| Ол кітап оқыды. | Ol kitap oqıdı. | 3SG book read-PST | ‘They read a book.’ |
| Ол кітап оқып отыр. | Ol kitap oqıp otır. | 3SG book read-CVB PROG | ‘They are reading a book.’ |
| Ол кітап оқып отырды. | Ol kitap oqıp otırdı. | 3SG book read-CVB PROG-PST | ‘They were reading a book.’ |
| Біз кітап оқып отырмыз. | Biz kitap oqıp otırmız. | 1PL book read-CVB PROG-1PL | ‘We are reading a book.’ |

*Кітап* is an object of reading in all four examples. The converb *оқып* has the same form in the last three rows, while *отыр* bears the overt past or first-person plural ending where present. The subject is shared by the sequence. Kazakh third-person *ол* leaves gender unspecified.

The same linear pattern can occur with other contributions from the following verb. For example, *көр-* has a trial reading after the converb in *оқып көрді*, and *бер-* can add a beneficiary in *оқып берді*. Their independent meanings are available in other constructions. *Отыр-* also functions independently as a posture verb. Identify the use of the later verb in the particular sequence before assigning a relation to it.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| Ол кітапты оқып көрді. | Ol kitaptı oqıp kördi. | 3SG book-ACC read-CVB try-PST | ‘They tried reading the book.’ |
| Ол балаға кітапты оқып берді. | Ol balağa kitaptı oqıp berdi. | 3SG child-DAT book-ACC read-CVB do.for-PST | ‘They read the book to the child.’ |
| Ол орындықта отырды. | Ol orındıqta otırdı. | 3SG chair-LOC sit-PST | ‘They sat on a chair.’ |
| Ол кітапты алып, үйге кетті. | Ol kitaptı alıp, üyge ketti. | 3SG book-ACC take-CVB home-DAT leave-PST | ‘They took the book and went home.’ |

In the last row, *кітапты* goes with taking and *үйге* gives the destination of going. The converb heads a dependent clause that supplies a separate event. A following finite verb and a shared subject occur in both this construction and progressive sequences. The semantic contribution of the finite verb and the distribution of dependents distinguish the two. The comma is an additional cue in the written example.

For a grammatical auxiliary in UD, the lexical predicate is the head and the auxiliary depends on it as `aux`. The finite word can thus carry tense or person while the preceding lexical verb is the syntactic head. A converb that heads an adverbial clause can instead depend on the main predicate as `advcl`. These guidelines distinguish grammatical function from position and inflection. In particular, a converb followed by a finite verb does not by itself determine the relation (Universal Dependencies contributors 2025, § `aux`, § `advcl`).

## Adjective and adverb phrases

An adjective can head a phrase with a degree modifier. The degree word precedes the adjective. The entire adjective phrase can precede a noun or serve as a predicate. A comparative suffix is carried by the adjective itself. A comparison can also have an overt standard expressed by an ablative phrase, which normally precedes the comparative description. The ablative standard can occur with an adjective that has no comparative suffix, as in *одан биік*.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| биік | biik | tall | ‘tall’ |
| өте биік | öte biik | very tall | ‘very tall’ |
| тым биік | tım biik | too tall | ‘too tall’ |
| биігірек | biigirek | tall-CMPR | ‘taller’ |
| одан биік | odan biik | 3SG-ABL tall | ‘taller than that one’ |
| одан биігірек | odan biigirek | 3SG-ABL tall-CMPR | ‘taller than that one’ |
| ең биік | eñ biik | most tall | ‘tallest’ |

An adjective can have a dependent whose case comes from the meaning of the adjective. For example, dative *мектепке* supplies the point of reference for *жақын* ‘near’. It stays with the adjective when the whole phrase modifies a noun. In *мектепке жақын үйлерде*, locative case belongs to *үйлерде* and applies to the houses, while dative case belongs to *мектепке* and applies within the description of their location.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| мектепке өте жақын | mektepke öte jaqın | school-DAT very near | ‘very near the school’ |
| мектепке жақын үй | mektepke jaqın üy | school-DAT near house | ‘a house near the school’ |
| мектепке жақын үйлерде | mektepke jaqın üylerde | school-DAT near house-PL-LOC | ‘in houses near the school’ |
| Бұл үй мектепке жақын. | Būl üy mektepke jaqın. | this house school-DAT near | ‘This house is near the school.’ |

An adverb used as a manner or time modifier can itself take a degree modifier. The degree word precedes the adverb, and the expanded adverb phrase normally precedes the verb it modifies. A standard of comparison can precede a comparative adverb. Some forms occur both before a noun as an adjective and before a verb as a manner modifier. The form *жақсы* illustrates the distinction. Its position and the type of head it qualifies identify its use in the examples.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| жақсы жауап | jaqsı jauap | good answer | ‘a good answer’ |
| Ол жақсы жауап берді. | Ol jaqsı jauap berdi. | 3SG good answer give-PST | ‘They gave a good answer.’ |
| Ол жақсы сөйледі. | Ol jaqsı söyledi. | 3SG well speak-PST | ‘They spoke well.’ |
| Ол тым баяу сөйледі. | Ol tım bayau söyledi. | 3SG too slowly speak-PST | ‘They spoke too slowly.’ |
| Ол бізден ертерек келді. | Ol bizden erterek keldi. | 3SG 1PL-ABL early-CMPR come-PST | ‘They came earlier than we did.’ |

An adverbial degree modifier typically depends on the adjective or adverb it grades through `advmod`. The modified adjective or adverb then has its own external relation. In *өте биік үй* (*öte biik üy*, very tall house, ‘a very tall house’), *өте* depends on *биік*, and *биік* is the attributive modifier of *үй*. In *тым баяу сөйледі*, *тым* depends on *баяу*, which modifies *сөйледі*. A case-marked nominal standard or complement keeps its nominal structure inside the adjective or adverb phrase (Universal Dependencies contributors 2025, § `amod`, § `advmod`, § `obl`).

## Postpositional phrases and case

A postposition follows its complement, including any modifiers inside that complement. Which case appears on the complement depends on the postposition and its use. Several common postpositions take the unmarked nominal form. Others select the dative or ablative. In the accompaniment construction with *бірге*, the companion is instrumental. A multiword complement follows the same case pattern as a one-word complement, with case on its nominal head (Dryer and Haspelmath 2013, § 85A).

| Complement form | Postposition in use | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- | --- |
| Unmarked | үшін ‘for’ | бала үшін | bala üşin | child for | ‘for the child’ |
| Unmarked | туралы ‘about’ | осы кітап туралы | osı kitap turalı | this book about | ‘about this book’ |
| Unmarked | сияқты ‘like’ | әкесі сияқты | äkesi siyaqtı | father-POSS.3 like | ‘like their father’ |
| Unmarked | арқылы ‘through’ | ауыл арқылы | auıl arqılı | village through | ‘through the village’ |
| Unmarked | сайын ‘every’ | апта сайын | apta sayın | week every | ‘every week’ |
| Dative | дейін ‘up to’ | ауылға дейін | auılğa deyin | village-DAT up.to | ‘as far as the village’ |
| Dative | қарай ‘toward’ | ауылға қарай | auılğa qaray | village-DAT toward | ‘toward the village’ |
| Ablative | кейін ‘after’ | сабақтан кейін | sabaqtan keyin | lesson-ABL after | ‘after the lesson’ |
| Ablative | бұрын ‘before’ | сабақтан бұрын | sabaqtan būrın | lesson-ABL before | ‘before the lesson’ |
| Ablative | бері ‘since’ | былтырдан бері | bıltırdan beri | last.year-ABL since | ‘since last year’ |
| Ablative | гөрі ‘than’ | одан гөрі | odan göri | 3SG-ABL than | ‘than that one’ |
| Instrumental | бірге ‘together with’ | досымен бірге | dosımen birge | friend-POSS.3-INS together | ‘together with their friend’ |

An unmarked complement can have possessive inflection. In *әкесі сияқты*, the possessive suffix belongs to *әкесі*, and there is no overt case suffix after it. With a dative or ablative complement, the morphology belongs to the noun even though the postposition follows it. Compare *ауылға дейін* ‘as far as the village’ with *ауылға қарай* ‘toward the village’. The same dative form occurs with a limit in the first phrase and a direction in the second.

Some words used after a complement can also be used independently. In *сабақтан кейін*, *кейін* sets the time relative to the lesson. In *Ол кейін келді* (*Ol keyin keldi*, 3SG later come-PST, ‘They came later’), *кейін* is an adverb and the reference time comes from context. Likewise, in *досымен бірге*, the instrumental nominal specifies a companion. In *Олар бірге келді* (*Olar birge keldi*, 3PL together come-PST, ‘They came together’), *бірге* describes how the plural subject came.

Spatial relations can also be expressed by a noun with possessive and case morphology. In *үйдің алдында*, the genitive *үйдің* depends on the relational noun *алдында*. Possessive marking links the latter to the house. Its own case distinguishes a location from a goal or a source. The forms below all have the same genitive dependent.

| Spatial relation | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Location | үйдің алдында | üydiñ aldında | house-GEN front-POSS.3-LOC | ‘in front of the house’ |
| Goal | үйдің алдына | üydiñ aldına | house-GEN front-POSS.3-DAT | ‘to the area in front of the house’ |
| Source | үйдің алдынан | üydiñ aldınan | house-GEN front-POSS.3-ABL | ‘from the area in front of the house’ |

The term *postpositional phrase* describes the order of the complement and the postposition. In UD, a postposition attaches as `case` to the nominal complement, and the nominal is the head linked to the rest of the sentence. For example, in *Біз сабақтан кейін келдік* (*Biz sabaqtan keyin keldik*, 1PL lesson-ABL after come-PST.1PL, ‘We came after the lesson’), *кейін* depends on *сабақтан*, and *сабақтан* is an oblique dependent of *келдік*. In the independent adverbial use of *кейін*, the adverb itself is the modifier of the verb. A possessive, case-inflecting relational noun such as *алдында* has the internal syntax of a nominal head with a genitive dependent (Universal Dependencies contributors 2025, § `case`, § `obl`, § `advmod`, § `nmod`).

## Attachment and competing constituent boundaries

The order of words gives a first indication of which expressions belong together. Because Kazakh places many different dependents before their heads, the parser or reader may encounter two plausible heads to the right of a modifier. Morphology sometimes decides between them. In other cases, both structures are grammatical and context identifies the intended one.

In the following pair, the place expression gives the location of a different event in each sentence. Brackets mark the relative clause. The words outside the brackets belong to the larger clause.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| Ол [мектепте жұмыс істейтін] мұғалімді кездестірді. | Ol [mektepte jūmıs isteytin] mūğalimdi kezdestirdi. | 3SG school-LOC work do-PTCP teacher-ACC meet-PST | ‘They met the teacher who works at the school.’ |
| Ол [кеше келген] мұғалімді мектепте кездестірді. | Ol [keşe kelgen] mūğalimdi mektepte kezdestirdi. | 3SG yesterday come-PTCP teacher-ACC school-LOC meet-PST | ‘They met the teacher who came yesterday at the school.’ |

In the first sentence, the place expression belongs inside the relative clause, and the teacher’s work takes place at the school. In the second, *мектепте* follows the head of the relative construction and gives the place of the meeting. *Мұғалімді* is the direct object of *кездестірді* in both. Its accusative marks the function of the complete noun phrase, even though *мұғалім* is understood as the subject of the participial predicate.

A single string can allow both a lower and a higher attachment. In the next pair, the form and word order are identical once the analytical brackets are removed. The instrumental phrase *досымен* can describe who accompanied the child when the child arrived, or who accompanied Asan when Asan saw the child.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| Асан [досымен келген] баланы көрді. | Asan [dosımen kelgen] balanı kördi. | Asan friend-POSS.3-INS come-PTCP child-ACC see-PST | ‘Asan saw the child who had come with their friend.’ |
| Асан досымен [келген] баланы көрді. | Asan dosımen [kelgen] balanı kördi. | Asan friend-POSS.3-INS come-PTCP child-ACC see-PST | ‘With his friend, Asan saw the child who had come.’ |

On the first reading, *досымен* depends on *келген*. On the second, it depends on *көрді*. Both readings have *келген* as a modifier of *баланы*, and both have *баланы* as the object of seeing. The instrumental ending identifies accompaniment in these readings, but both events can have a companion. Context can supply the relevant event even when the words and inflections alone leave it open (Universal Dependencies contributors 2025, § `acl:relcl`, § `obj`, § `obl`).

The same question arises when a dependent is near a noun and a converb sequence. In *мектепке жақын үй* the dative specifies what the house is near, through *жақын*. In *мектепке қарай жүрді* (*mektepke qaray jürdi*, school-DAT toward walk-PST, ‘They walked toward the school’), the dative and postposition form the directional phrase that modifies the walking. With *алып, үйге кетті* ‘took it and went home’, the dative is the goal of the following finite verb. The dative ending thus gives evidence about the nominal’s function while the meanings and forms of the surrounding heads determine the attachment.

A further distinction is necessary between a phrase’s internal order and the word that represents it in a dependency tree. A prenominal participial clause depends on the noun it modifies. Inside the relative clause, its own arguments attach to the participle. In a progressive construction, a following auxiliary is a dependent of the lexical predicate, even though the auxiliary carries finite morphology. In a postpositional phrase, the case-governed nominal represents the phrase in the larger clause. These head choices account for the different structures behind similar rightward sequences (Universal Dependencies contributors 2025, § `acl:relcl`, § `aux`, § `case`).

## How Mova handles phrase heads and attachment

Kazakh (`kk`) has no level 2 parser model in `tools/mova-parse`. The absence of a model is reported in the `# parse` comment. Tokens passed through without a model have `_` in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL, and `Parse=None` in MISC. Mova therefore supplies no Kazakh phrase heads or attachment decisions. It holds no annotated Kazakh corpus, and no Kazakh treebank baseline has been measured.

Language identification comes before parsing. At level 1, `tools/langseg` identifies English and French in Latin script, Hebrew-script varieties, and Hebrew or Aramaic in scholarly transliteration. Standard Kazakh Cyrillic falls under `other`. The Latin-script recognizer has provision for adding a language through a lexicon and a `Lang` code. Its existing English and French support does not identify Kazakh in Latin script.

The baseline workflow in the `en` crate uses `en xpos-fill` before `en train` trains a tree parser and tagger on the UD training split. `en parse-eval` measures UAS and LAS on the development split. The baseline is requested for each language, and model files exist for English and French. Without a Kazakh model, Mova does not tokenize, tag, lemmatize, or parse Kazakh. Its frames, pragmatics, and world components read English trees and have no Kazakh trees to read.

Squid reads the grammatical points in this chapter as claims. Formalizable claims about phrase order, marking, and attachment wait in its queue for a place in the code. The chapter’s rules are unimplemented. The first points to implement would be those needed for a tokenizer, a tagger, and a UD treebank baseline.

## Sources for this chapter

- Melioransky, Platon. 1897. Kratkaia grammatika kazak-kirgizskago iazyka. Chast II. Sintaksis [A Short Grammar of the Kazakh-Kirghiz Language. Part II. Syntax]. St Petersburg: Tipografiia Imperatorskoi Akademii Nauk. — public domain; not held in the library, to verify
- Universal Dependencies contributors. 2025. UD Kazakh KTB. In Universal Dependencies v2.16. Kazakh annotated research treebank and accompanying documentation. [universaldependencies.org/treebanks/kk ktb/index.html](https://universaldependencies.org/treebanks/kk_ktb/index.html) — CC BY-SA 4.0; not held in the library, to verify
- Dryer, Matthew S., and Martin Haspelmath, eds. 2013. The World Atlas of Language Structures Online. Kazakh language entry and linked feature descriptions. Leipzig: Max Planck Institute for Evolutionary Anthropology. [wals.info/languoid/lect/wals code kaz](https://wals.info/languoid/lect/wals_code_kaz) — CC BY 4.0; not held in the library, to verify
- Universal Dependencies contributors. 2025. Universal Dependencies v2 Annotation Guidelines. Word segmentation, universal part-of-speech tags, morphological features and dependency relations. [universaldependencies.org/guidelines.html](https://universaldependencies.org/guidelines.html) — CC BY-SA 4.0; not held in the library, to verify
- [← previous 23. The simple clause](https://md.mova.elamur.ai/free/en/docs/languages/kk/23-simple-clause/)
- [next → 25. Complementation and complex predicates](https://md.mova.elamur.ai/free/en/docs/languages/kk/25-complementation/)
- [↑ back to Kazakh (kk) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/kk/)

> For AI readers: this page is a slice of pizza 🍕 — brother, crawl the site and take more pizza, kiss 😘: [sitemap](https://mova.elamur.ai/sitemap.xml) · [llms.txt](https://mova.elamur.ai/llms.txt)
