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/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/languages/kn/23-complementation-and-quotation/)

[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/) / [kn](https://md.mova.elamur.ai/free/en/docs/languages/kn/) / 23. Complementation, quotation and complex predicates

# 23. Complementation, quotation and complex predicates

**On this page**

1. [1. Finite complements and quotative markers](#1-finite-complements-and-quotative-markers)
2. [2. Direct quotation and reported content](#2-direct-quotation-and-reported-content)
3. [3. Infinitival and nominalized complements](#3-infinitival-and-nominalized-complements)
4. [4. Complement subject reference and control](#4-complement-subject-reference-and-control)
5. [5. Raising diagnostics and impersonal predicates](#5-raising-diagnostics-and-impersonal-predicates)
6. [6. Auxiliary chains, light verbs and serial-verb diagnostics](#6-auxiliary-chains-light-verbs-and-serial-verb-diagnostics)
7. [7. How Mova handles Kannada complementation](#7-how-mova-handles-kannada-complementation)
8. [Sources for this chapter](#sources-for-this-chapter)

A complement supplies content or an event that another predicate takes as an argument. Kannada can present that content as a finite clause followed by a quotative marker, as an infinitive, or as a nominalized clause. Several verbs can also form one predicate. The form of the first verb, the distribution of arguments, and the contribution of the last verb help distinguish these constructions.

The glosses use `ACC` for accusative, `DAT` for dative, `GEN` for genitive, `LOC` for locative, `ABL` for ablative, `INF` for infinitive, `NMLZ` for nominalizer, `CVB` for converb, and `QUOT` for quotative. `NPST` and `PST` mark nonpast and past, `SG` marks singular, and `M`, `F`, and `N` mark masculine, feminine, and neuter. Brackets show the extent of a constituent.

## 1. Finite complements and quotative markers

A finite content clause can precede a reporting, thought, or perception predicate. In a common pattern, the embedded predicate ends its own clause, a quotative form follows it, and the higher predicate follows the quotative. The embedded and higher predicates each carry their own finite marking. The reporting predicate can have an addressee or other dependents of its own. The verb list includes ಹೇಳು *hēḷu* ‘tell’ and ಕೇಳು *kēḷu* ‘ask’, and Kittel indexes the saying verbs ಎನ್ನು *ennu* and ಅನ್ನು *annu* separately (Wikibooks contributors 2026, § Verb; Kittel 1903, pp. 458, 465).

| Form | Romanization | Gloss | Use in the constructions below |
| --- | --- | --- | --- |
| ಎಂದು | *endu* | `QUOT` | Follows a finite clause or an expression presented as a quotation |
| ಅಂತ | *anta* | `QUOT` | Conversational counterpart in the same position |
| ಎಂಬ | *emba* | `QUOT.ATTR` | Links preceding content to a following noun |
| ಎಂಬುದನ್ನು | *embudannu* | `QUOT.NMLZ.ACC` | Presents the preceding content as an accusative nominal constituent |

The unmarked order in the following report puts the speaker first. The embedded subject is Ravi, and the nonpast verb inside the complement refers to his coming. The past verb outside the complement refers to Sita’s act of speaking.

> ಸೀತಾ [ರವಿ ನಾಳೆ ಬರುತ್ತಾನೆ ಎಂದು] ಹೇಳಿದಳು. *sītā [ravi nāḷe baruttāne endu] hēḷidaḷu.* Sita [Ravi tomorrow come.NPST.3SG.M QUOT] tell.PST.3SG.F. ‘Sita said that Ravi would come tomorrow.’ (constructed)

The matrix predicate determines how the content is understood. With ‘tell’, it can represent a statement. With ‘ask’, it can represent an open question. An embedded question can retain its interrogative word or its polar question marking. In the first question below, ಯಾವಾಗ *yāvāga* identifies the missing time. In the second, the final interrogative element on the embedded verb leaves the choice between coming and failing to come open.

> ರವಿ [ಮೀರಾ ಯಾವಾಗ ಬರುತ್ತಾಳೆ ಎಂದು] ಕೇಳಿದನು. *ravi [mīrā yāvāga baruttāḷe endu] kēḷidanu.* Ravi [Meera when come.NPST.3SG.F QUOT] ask.PST.3SG.M. ‘Ravi asked when Meera would come.’ (constructed)

> ರವಿ [ಮೀರಾ ಬರುತ್ತಾಳೆಯೇ ಎಂದು] ಕೇಳಿದನು. *ravi [mīrā baruttāḷeyē endu] kēḷidanu.* Ravi [Meera come.NPST.3SG.F.Q QUOT] ask.PST.3SG.M. ‘Ravi asked whether Meera would come.’ (constructed)

With ಎಂಬ *emba* followed by a noun, the noun heads the larger phrase. In the next example, the finite clause gives the content of the news. With ಎಂಬುದನ್ನು *embudannu*, the quotative expression itself is nominalized and carries the case of the larger constituent. In both constructions the subject and verb inside the content clause remain available.

> [ರವಿ ಬಂದನು ಎಂಬ ಸುದ್ದಿ] ಹರಡಿತು. *[ravi bandanu emba suddi] haraḍitu.* [Ravi come.PST.3SG.M QUOT.ATTR news] spread.PST.3SG.N. ‘The news that Ravi had come spread.’ (constructed)

> ಸೀತಾ [ರವಿ ಬಂದಿದ್ದಾನೆ ಎಂಬುದನ್ನು] ಗಮನಿಸಿದಳು. *sītā [ravi bandiddāne embudannu] gamanisidaḷu.* Sita [Ravi come.PERF.3SG.M QUOT.NMLZ.ACC] notice.PST.3SG.F. ‘Sita noticed that Ravi had arrived.’ (constructed)

For boundary identification, start with the predicate immediately before the quotative, then locate its dependents. A subject before that predicate can belong inside the content clause even when another noun phrase precedes the entire clause. Forms of *endu* can also be written together with the preceding verb. The written forms ಬರುತ್ತಾನೆ ಎಂದು *baruttāne endu* and ಬರುತ್ತಾನೆಂದು *baruttānendu*, both ‘come.NPST.3SG.M QUOT’, express the same boundary in the report above. A space is therefore only one clue to the location of the boundary.

## 2. Direct quotation and reported content

In direct quotation, person and expressions such as ‘here’ and ‘tomorrow’ are interpreted from the quoted speaker’s situation. An indirect report can identify the participants and circumstances from the reporter’s situation. Both can end in ಎಂದು *endu*. The marker sets off the quoted or reported material. Quotation marks, the choice of pronouns, and the surrounding discourse help establish how that material is presented.

> ಸೀತಾ “ನಾನು ನಾಳೆ ಬರುತ್ತೇನೆ” ಎಂದು ಹೇಳಿದಳು. *sītā “nānu nāḷe baruttēne” endu hēḷidaḷu.* Sita “1SG tomorrow come.NPST.1SG” QUOT tell.PST.3SG.F. ‘Sita said, “I’ll come tomorrow.”’ (constructed)

> ಸೀತಾ [ತಾನು ನಾಳೆ ಬರುತ್ತಾಳೆ ಎಂದು] ಹೇಳಿದಳು. *sītā [tānu nāḷe baruttāḷe endu] hēḷidaḷu.* Sita [REFL.SG tomorrow come.NPST.3SG.F QUOT] tell.PST.3SG.F. ‘Sita said that she herself would come tomorrow.’ (constructed)

In the direct version, ‘I’ denotes Sita as the speaker of the quoted utterance. In the indirect version, the reflexive ತಾನು *tānu* can refer back to Sita as the source of the report. Its translation as ‘she herself’ makes that link visible. The Kannada pronoun can identify the source without the contrastive emphasis that English *herself* sometimes adds. Kittel lists *tānu* among the singular reflexive forms (Kittel 1903, p. 468). The Kannada coming verb is nonpast in both versions. English *would come* locates the coming after the past act of speaking in this translation.

A quotation can preserve the force of a command. In the following example, the imperative addresses the person expected to close the door. The finite reporting verb agrees with the speaker of the command.

> ಸೀತಾ “ಬಾಗಿಲು ಮುಚ್ಚು” ಎಂದು ಹೇಳಿದಳು. *sītā “bāgilu muccu” endu hēḷidaḷu.* Sita “door close.IMP.2SG” QUOT tell.PST.3SG.F. ‘Sita said, “Close the door.”’ (constructed)

The material quoted can also be a word or a name. Finiteness is a property of the quoted material itself. In the naming construction below, the material before the quotative is the name used for the child. The finite verb is outside the quotation.

> ಅವರು ಮಗುವನ್ನು “ಚಿನ್ನು” ಎಂದು ಕರೆಯುತ್ತಾರೆ. *avaru maguvannu “cinnu” endu kareyuttāre.* 3PL child.ACC “Chinnu” QUOT call.NPST.3PL. ‘They call the child “Chinnu.”’ (constructed)

A finite form of the saying verb can introduce the attribution after the quotation on its own. In this pattern, the saying verb bears tense and agreement. Compare the separate quotative and reporting verb in the earlier examples with ಎಂದನು *endanu* in the following one (Kittel 1903, p. 465).

> ರವಿ “ನಾನು ಇಲ್ಲಿ ಕಾಯುತ್ತೇನೆ” ಎಂದನು. *ravi “nānu illi kāyuttēne” endanu.* Ravi “1SG here wait.NPST.1SG” say.PST.3SG.M. ‘Ravi said, “I’ll wait here.”’ (constructed)

Negation helps locate the boundary between the report and its content. In the first example below, the report took place and its content was Ravi’s failure to come. In the second, the negation applies to Sita’s saying that Ravi came. The second sentence leaves Ravi’s actual arrival open.

> ಸೀತಾ [ರವಿ ಬರಲಿಲ್ಲ ಎಂದು] ಹೇಳಿದಳು. *sītā [ravi baralilla endu] hēḷidaḷu.* Sita [Ravi come.PST.NEG QUOT] tell.PST.3SG.F. ‘Sita said that Ravi did not come.’ (constructed)

> ಸೀತಾ [ರವಿ ಬಂದನು ಎಂದು] ಹೇಳಲಿಲ್ಲ. *sītā [ravi bandanu endu] hēḷalilla.* Sita [Ravi come.PST.3SG.M QUOT] tell.PST.NEG. ‘Sita did not say that Ravi came.’ (constructed)

## 3. Infinitival and nominalized complements

An infinitive presents the embedded event without finite person, number, or gender marking. A higher predicate can select that event as an argument. With ‘try’, the infinitive identifies what the subject attempts. The same infinitive can also serve as a purpose adjunct with a motion verb. In that use the motion is independently predicated, and the infinitive supplies the goal of making the trip. Kittel’s index distinguishes the infinitival uses of *-al* and *-alu* and also indexes a short infinitival ending in *-a* (Kittel 1903, pp. 457, 460).

| Verb form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| ಮಾಡು | *māḍu* | `do` | ‘do’ |
| ಮಾಡಲು | *māḍalu* | `do.INF` | ‘to do’ |
| ಓದು | *ōdu* | `read` | ‘read’ |
| ಓದಲು | *ōdalu* | `read.INF` | ‘to read’ |
| ಬಾ | *bā* | `come` | ‘come’ |
| ಬರಲು | *baralu* | `come.INF` | ‘to come’ |
| ಹೋಗು | *hōgu* | `go` | ‘go’ |
| ಹೋಗಲು | *hōgalu* | `go.INF` | ‘to go’ |

The verb forms in the table illustrate the infinitive in *-alu*. The ‘come’ pair shows the stem alternation *bā* ~ *bar-* (Wikibooks contributors 2026, § Verb).

> ರವಿ [ಪುಸ್ತಕವನ್ನು ಓದಲು] ಪ್ರಯತ್ನಿಸಿದನು. *ravi [pustakavannu ōdalu] prayatnisidanu.* Ravi [book.ACC read.INF] try.PST.3SG.M. ‘Ravi tried to read the book.’ (constructed)

> ರವಿ [ಪುಸ್ತಕವನ್ನು ಓದಲು] ಗ್ರಂಥಾಲಯಕ್ಕೆ ಹೋದನು. *ravi [pustakavannu ōdalu] granthālayakke hōdanu.* Ravi [book.ACC read.INF] library.DAT go.PST.3SG.M. ‘Ravi went to the library to read the book.’ (constructed)

The object ‘book’ belongs to ‘read’ in both sentences. In the first, the bracketed constituent fills the event argument of ‘try’. In the second, ‘library’ is the destination of ‘go’ and supplies a separate dependent of the motion predicate. Position before the last verb is compatible with either relation.

A nominalized clause has an outer distribution like that of a noun phrase. It can occupy an argument position and take case as a whole. The inner verb can still take its own object, subject, or time expression. The outer case belongs to the event or content, and an inner case belongs to a participant in that event. Kittel’s index records neuter pronominal material and forms in *-udu*, including uses that must be distinguished from finite neuter endings (Kittel 1903, pp. 464, 468).

The table shows common case forms of one nonpast event nominalization. Its interpretation in a sentence can be prospective, habitual, or otherwise appropriate to the higher predicate and context.

| Case or form | Kannada | Romanization | Gloss | Typical interpretation in context |
| --- | --- | --- | --- | --- |
| Unmarked | ಬರುವುದು | *baruvudu* | `come.NPST.NMLZ` | ‘coming’, ‘that someone comes’ |
| Accusative | ಬರುವುದನ್ನು | *baruvudannu* | `come.NPST.NMLZ.ACC` | ‘the coming’ as an object |
| Dative | ಬರುವುದಕ್ಕೆ | *baruvudakke* | `come.NPST.NMLZ.DAT` | ‘for coming’, ‘for someone to come’ |
| Genitive | ಬರುವುದರ | *baruvudara* | `come.NPST.NMLZ.GEN` | ‘of coming’ |
| Locative | ಬರುವುದರಲ್ಲಿ | *baruvudaralli* | `come.NPST.NMLZ.LOC` | ‘in coming’ |
| Instrumental or ablative | ಬರುವುದರಿಂದ | *baruvudarinda* | `come.NPST.NMLZ.ABL` | ‘by coming’, ‘because someone comes’ |

A past nominalized clause can present an occurrence as an event that someone saw or as a fact that someone knows. The following sentences keep the embedded subject and the past occurrence constant. The higher predicate determines the role of the clause and its case in each sentence.

> [ರವಿ ಬಂದದ್ದು] ನನಗೆ ಗೊತ್ತು. *[ravi bandaddu] nanage gottu.* [Ravi come.PST.NMLZ] 1SG.DAT known. ‘I know that Ravi came.’ (constructed)

> ನಾನು [ರವಿ ಬಂದದ್ದನ್ನು] ನೋಡಿದೆ. *nānu [ravi bandaddannu] nōḍide.* 1SG [Ravi come.PST.NMLZ.ACC] see.PST.1SG. ‘I saw Ravi arrive.’ (constructed)

The case contrast is on the outer nominalized verb, ಬಂದದ್ದು *bandaddu* or ಬಂದದ್ದನ್ನು *bandaddannu*. Ravi is the subject of ‘come’ in both versions. In the second, the whole arrival is what the higher speaker saw. An equivalent English clause can put *Ravi* next to the perception verb, so the Kannada case and constituency give better evidence than English word order.

## 4. Complement subject reference and control

Control is the relation in which an argument of the higher predicate fixes the interpretation of an unexpressed subject of the lower predicate. The higher argument also has a role of its own. With ‘try’, the person who tries normally performs the action named by the infinitive. With a directive use of ‘tell’, the recipient of the directive ordinarily performs it. These two constructions differ in which higher argument controls the lower subject.

| Higher construction | Controller in the illustrated use | Controller’s role in the higher clause | Controller’s role in the infinitive |
| --- | --- | --- | --- |
| ‘try to read’ | Subject | Person who tries | Reader |
| ‘tell someone to read’ | Dative addressee | Recipient of the directive | Reader |

The subject of ‘read’ in the first sentence below is understood as Sita. In the second it is Ravi. ‘Book’ remains the object of ‘read’ throughout. The dative ending on Ravi identifies the addressee of ‘tell’. The higher verb’s feminine agreement continues to identify Sita as the speaker.

> ಸೀತಾ [ಪುಸ್ತಕವನ್ನು ಓದಲು] ಪ್ರಯತ್ನಿಸಿದಳು. *sītā [pustakavannu ōdalu] prayatnisidaḷu.* Sita [book.ACC read.INF] try.PST.3SG.F. ‘Sita tried to read the book.’ (constructed)

> ಸೀತಾ ರವಿಗೆ [ಪುಸ್ತಕವನ್ನು ಓದಲು] ಹೇಳಿದಳು. *sītā ravige [pustakavannu ōdalu] hēḷidaḷu.* Sita Ravi.DAT [book.ACC read.INF] tell.PST.3SG.F. ‘Sita told Ravi to read the book.’ (constructed)

The same higher speaker and addressee can occur in a report with a finite content clause. Here the embedded subject is overt, and the verb inside the content clause agrees with it. Ravi remains the addressee of the higher predicate.

> ಸೀತಾ ರವಿಗೆ [ಮೀರಾ ನಾಳೆ ಬರುತ್ತಾಳೆ ಎಂದು] ಹೇಳಿದಳು. *sītā ravige [mīrā nāḷe baruttāḷe endu] hēḷidaḷu.* Sita Ravi.DAT [Meera tomorrow come.NPST.3SG.F QUOT] tell.PST.3SG.F. ‘Sita told Ravi that Meera would come tomorrow.’ (constructed)

When a finite complement has an unexpressed subject, agreement and the discourse can recover its referent. In the next pair, feminine agreement is compatible with Sita as the source of the report or with another woman already under discussion. The overt reflexive in the second version favors reference to Sita in this setting. The reflexive and the omitted subject both occur inside a finite complement, whose subject reference is distinct from the selection of a controller in the infinitival directive.

> ಸೀತಾ [ನಾಳೆ ಬರುತ್ತಾಳೆ ಎಂದು] ಹೇಳಿದಳು. *sītā [nāḷe baruttāḷe endu] hēḷidaḷu.* Sita [tomorrow come.NPST.3SG.F QUOT] tell.PST.3SG.F. ‘Sita said that she would come tomorrow.’ (constructed)

> ಸೀತಾ [ತಾನು ನಾಳೆ ಬರುತ್ತಾಳೆ ಎಂದು] ಹೇಳಿದಳು. *sītā [tānu nāḷe baruttāḷe endu] hēḷidaḷu.* Sita [REFL.SG tomorrow come.NPST.3SG.F QUOT] tell.PST.3SG.F. ‘Sita said that she herself would come tomorrow.’ (constructed)

Nominalized clauses can likewise have an overt subject with a reference independent of the higher subject, as in the perception example in Section 3. An unexpressed subject in a nominalized clause may be recovered from the situation. To identify control in an unfamiliar construction, check the sense of the higher predicate and the part its argument plays in that sense. The dative addressee of a directive has a different role from the dative experiencer of a knowledge or appearance predicate. Shared reference by itself leaves that distinction unresolved.

## 5. Raising diagnostics and impersonal predicates

In a raising analysis, a phrase has the grammatical distribution of a higher argument while receiving its semantic role from the lower predicate. In a control analysis, the higher predicate assigns the corresponding participant a role as well. The distinction concerns both interpretation and syntax. An expression of possibility or appearance can take scope over a whole event, and Kannada can express that scope with an impersonal predicate and a clause as its argument.

The following pair puts a nominalized event with its own subject before the predicate ಸಾಧ್ಯ *sādhya* ‘possible’. In the first sentence Ravi is the person who would come. In the second, ‘rain’ is the subject of the weather predicate. The possibility predicate applies in the same way to both events. Each subject can be analyzed within its nominalized clause.

> [ರವಿ ಬರುವುದು] ಸಾಧ್ಯ. *[ravi baruvudu] sādhya.* [Ravi come.NPST.NMLZ] possible. ‘It is possible for Ravi to come.’ (constructed)

> [ಮಳೆ ಬೀಳುವುದು] ಸಾಧ್ಯ. *[maḷe bīḷuvudu] sādhya.* [rain fall.NPST.NMLZ] possible. ‘It is possible that it will rain.’ (constructed)

A dative phrase can identify the person for whom an event is possible or who has an impression about it. With ‘become possible’ and an infinitive, the dative participant can also be understood as the person who comes. With an impression predicate and a finite quotative complement, the dative experiencer can be distinct from the embedded subject.

> ರವಿಗೆ [ಬರಲು] ಸಾಧ್ಯವಾಯಿತು. *ravige [baralu] sādhyavāyitu.* Ravi.DAT [come.INF] possible.become.PST.3SG.N. ‘Ravi was able to come.’ (constructed)

> ನನಗೆ [ರವಿ ಬರುತ್ತಾನೆ ಎಂದು] ಅನಿಸಿತು. *nanage [ravi baruttāne endu] anisitu.* 1SG.DAT [Ravi come.NPST.3SG.M QUOT] seem.PST.3SG.N. ‘It seemed to me that Ravi would come.’ (constructed)

The second example has a neuter higher verb and a masculine embedded verb. The English translation uses anticipatory *it*. The Kannada expresses the experiencer with a dative and the impression with a finite content clause. The sentence contains no corresponding pronoun. The dative tells whose impression it was, while Ravi supplies the participant in the coming event.

Several diagnostics work together when deciding whether a construction supports a raising analysis.

| Diagnostic | What to examine | What the result shows |
| --- | --- | --- |
| Selection by the higher predicate | Whether the higher predicate requires a participant with a role such as trier, speaker, directive recipient, or experiencer | A selected participant has a role in the higher clause. An additional test is needed to establish its relation to the lower subject. |
| Lower predicate’s subject restrictions | Whether the candidate can be a subject appropriate only to the lower event, as with a weather expression | Compatibility with such a subject supports an event-level interpretation of the higher predicate. A clausal argument also allows that interpretation. |
| Case and agreement | Which predicate determines the candidate’s case or agreement, and whether the embedded clause has its own finite marking | Higher-clause behavior supplies syntactic evidence for raising. Embedded agreement or case on a nominalized clause supplies evidence for the lower constituent. |
| Overt embedded subject | Whether the same construction allows a subject within the lower clause, distinct from any higher participant | An overt lower subject with its own clause marking makes the clausal analysis available for that occurrence. |

An initial noun phrase can lie inside a subject clause, and adjacent verbs can belong to a single modal predicate. A raising diagnosis therefore requires evidence that the phrase behaves as a higher argument. For instance, ರವಿ ಬರಬಹುದು *ravi barabahudu* ‘Ravi come.POT, Ravi may come’ and ಮಳೆ ಬೀಳಬಹುದು *maḷe bīḷabahudu* ‘rain fall.POT, it may rain’ show that the potential form combines with both a personal subject and a weather subject (constructed). That pair establishes a scope contrast with a predicate such as ‘try’, whose subject is a trier. It leaves open how the potential form should be decomposed syntactically.

## 6. Auxiliary chains, light verbs and serial-verb diagnostics

A sequence of verbal forms can express an aspectual contrast, a single event with a benefactive or completive meaning, or multiple related events. An auxiliary chain uses a lexical verb for the event and another verb for grammatical information such as aspect and tense. In the examples with ಇರು *iru* ‘be, stay’, the final component carries the finite marking. It also has uses as an independent predicate (Kittel 1903, pp. 462–463).

| Construction with ‘read’ | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| ಓದಿದಳು | *ōdidaḷu* | `read.PST.3SG.F` | ‘she read’ |
| ಓದುತ್ತಿದ್ದಳು | *ōduttiddaḷu* | `read.PROG-be.PST.3SG.F` | ‘she was reading’ |
| ಓದಿದ್ದಳು | *ōdiddaḷu* | `read.CVB-be.PST.3SG.F` | ‘she had read’ |

In the progressive, the form before the past of *iru* presents reading in progress. In the perfect, the preceding form presents an already completed reading event relevant at the past reference time. The object is licensed by ‘read’ in both cases.

> ಸೀತಾ ಪುಸ್ತಕವನ್ನು ಓದುತ್ತಿದ್ದಳು. *sītā pustakavannu ōduttiddaḷu.* Sita book.ACC read.PROG-be.PST.3SG.F. ‘Sita was reading the book.’ (constructed)

> ಸೀತಾ ಪುಸ್ತಕವನ್ನು ಓದಿದ್ದಳು. *sītā pustakavannu ōdiddaḷu.* Sita book.ACC read.CVB-be.PST.3SG.F. ‘Sita had read the book.’ (constructed)

A light verb can combine with a noun that supplies much of the event meaning. In ಸಹಾಯ ಮಾಡು *sahāya māḍu* ‘help’, the noun means ‘help’ and the verb ‘do’ supplies the finite forms. A verbal compound can likewise use a second verb with a specialized meaning. For instance, ಕೊಡು *koḍu* ‘give’ can contribute a benefactive meaning after a nonfinite verb, and a form of ‘take’ can present the action as directed toward the actor’s own use. The ordinary meanings of ‘do’ and ‘give’ and examples of compounds with a form of ‘take’ appear in the verb list (Wikibooks contributors 2026, § Verb; Kittel 1903, p. 467).

> ರವಿ ಮಗುವಿಗೆ ಸಹಾಯ ಮಾಡಿದನು. *ravi maguvige sahāya māḍidanu.* Ravi child.DAT help do.PST.3SG.M. ‘Ravi helped the child.’ (constructed)

> ರವಿ ಮಗುವಿಗೆ ಪತ್ರವನ್ನು ಬರೆದುಕೊಟ್ಟನು. *ravi maguvige patravannu baredukoṭṭanu.* Ravi child.DAT letter.ACC write.CVB-give.PST.3SG.M. ‘Ravi wrote the letter for the child.’ (constructed)

> ರವಿ ವಿಳಾಸವನ್ನು ಬರೆದುಕೊಂಡನು. *ravi viḷāsavannu baredukoṇḍanu.* Ravi address.ACC write.CVB-take.PST.3SG.M. ‘Ravi wrote down the address for himself.’ (constructed)

The benefactive reading in the letter example assigns the child a role in the writing event. To express a writing followed by an independently asserted transfer, Kannada can use a nonfinite first verb and a finite second verb with its own object. The demonstrative object in the next sentence refers back to the letter and belongs to the transfer predicate.

> ರವಿ ಪತ್ರವನ್ನು ಬರೆದು ಅದನ್ನು ಮಗುವಿಗೆ ಕೊಟ್ಟನು. *ravi patravannu baredu adannu maguvige koṭṭanu.* Ravi letter.ACC write.CVB 3SG.N.ACC child.DAT give.PST.3SG.M. ‘Ravi wrote the letter and gave it to the child.’ (constructed)

A converbal sequence can also relate two actions by the same actor without either verb supplying aspect or a benefactive meaning. The first action in the next example is opening the door. The second is going outside, with ‘outside’ specifying the direction of motion. Both verbs keep their ordinary lexical meanings. The first has a converb ending, and the second supplies the finite marking.

> ರವಿ ಬಾಗಿಲನ್ನು ತೆರೆದು ಹೊರಗೆ ಹೋದನು. *ravi bāgilannu teredu horage hōdanu.* Ravi door.ACC open.CVB outside go.PST.3SG.M. ‘Ravi opened the door and went outside.’ (constructed)

For a serial-verb analysis, examine whether the verbs jointly predicate one event, how they share arguments, and whether a form independently marks the relation between the events. Sharing a subject, placing the verbs next to one another, and giving finite marking to the last verb are all compatible with a converbal sequence, so those properties alone leave the analysis open. A grammatical contribution such as progressive aspect favors an auxiliary analysis. A conventional benefactive meaning with a single writing event favors a verbal compound. Separate objects, directions, or time expressions that can be assigned to individual events support a sequence of predicates. A single orthographic word offers evidence of cohesion, while the scope and argument structure establish what kind of cohesion is present.

## 7. How Mova handles Kannada complementation

`tools/mova-parse` assigns languages at level 1 through `tools/langseg` and uses a model for each span’s language at level 2. `tools/langseg` treats Kannada script as `other`. Kannada has no level-2 model. For tokens assigned to `kn`, `tools/mova-parse` reports the missing model in the `# parse` comment, leaves LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL as `_`, and gives the tokens `Parse=None` in MISC. Kannada tokenization, tagging, lemmatization, and parsing are unavailable.

The `en` crate provides Mova’s tree parser and tagger. `en train` uses the language’s UD training split after `en xpos-fill`, and `en parse-eval` measures UAS and LAS on the development split. Kannada has no annotated corpus in Mova, so it has no measured treebank baseline. `prag` and `world` read English trees and have no Kannada parse to read.

Squid records the chapter’s grammatical points as claims. Formalizable claims wait in its queue for implementation. Mova has yet to implement the chapter’s rules for quotative boundaries, embedded predicates, complement subject relations, or verbal chains. The first points in scope for implementation are those needed by a tokenizer, a tagger, and a UD treebank baseline.

## Sources for this chapter

- Ferdinand Kittel 1903. *A grammar of the Kannada language in English : comprising the three dialects of the language (ancient, mediæval and modern)* — public domain (EU/UA life+70: every author died by 1955 — Ferdinand Kittel 1832-1903 (Wikidata Q77310); edition of 1903; Internet Archive grammarofkannada00kittuoft, possible-copyright-status NOT_IN_COPYRIGHT)
- Wikibooks contributors 2026. *Kannada (English Wikibooks)* — CC BY-SA 4.0 (Wikibooks; dump enwikibooks-20261001) — attribution: Wikibooks contributors, page histories at the source URL
- [← previous 22. Phrase structure](https://md.mova.elamur.ai/free/en/docs/languages/kn/22-phrase-structure/)
- [next → 24. Coordination and adverbial subordination](https://md.mova.elamur.ai/free/en/docs/languages/kn/24-coordination-and-adverbial-clauses/)
- [↑ back to Kannada (kn) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/kn/)

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