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/jv/kaw/19-complementation/)

[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/) / [jv](https://md.mova.elamur.ai/free/en/docs/languages/jv/) / [kaw](https://md.mova.elamur.ai/free/en/docs/languages/jv/kaw/) / 19. Complement clauses, reported content and predicate sequences

# 19. Complement clauses, reported content and predicate sequences

A complement expresses content or an event that a predicate takes as an argument. Old Javanese can also present words as a direct quotation, name the content with a noun phrase, or place two predicates together with a participant understood in both. The distinctions depend on what each predicate contributes, how its arguments are expressed, and whose point of view governs the passage. The examples use romanized Old Javanese. `IRR` glosses the dependent or nonasserted uses of *-a* illustrated here, `PASS` marks passive voice, `POSS` marks a possessor, and `LOC` marks a locative expression.

## Speech, cognition and perception predicates

A speech expression can identify a speaker, a recipient and what was said. These are separate roles. A word for the recipient may stand beside the speech verb even when the words spoken follow it or remain implicit. A speech noun may have a possessor identifying the speaker. A noun naming the utterance can itself serve as an argument of a speech verb.

| Short phrase | Gloss | Translation |
| --- | --- | --- |
| “warah sira” | tell 3SG | ‘Tell her.’ |
| “maṅunyakĕn stuti” | utter praise | ‘Utter praises.’ |

The first phrase is the end of a directive in which the person addressed is to return and tell someone else. In that passage *sira* is the recipient of the telling. The second phrase has a noun for the material uttered. Its meaning by itself gives no clause expressing the words of the praise. Neither the order of the pronoun in the first example nor the position of the noun in the second gives a rule for finding the beginning of a reported clause. (Van der Tuuk 1897, pp. 9–10, 21–23)

Cognition predicates relate a person to an object of knowledge, a proposition or a judgment. The expression for the knower may occur after the predicate. For instance, the beginning of a passage about the gods has the short phrase “wruh saṅ hyaṅ”, glossed `know HON deity`, ‘the deity knew’. The phrase identifies the knower. The following narrative is needed to delimit what that participant knew. The dictionary also distinguishes conjecture from knowledge in its discussion of *maṅuha* ‘conjecture’ and *paṅuhā* ‘conjecture, supposition’. A translation should preserve that lexical difference even where the content of the two predicates could take the same syntactic form. (Van der Tuuk 1897, pp. 9–10, 14–15)

| Form | Gloss | Translation |
| --- | --- | --- |
| *Wruh sira yan datĕṅ saṅ ratu.* (constructed) | know 3SG LINK arrive HON ruler | ‘He knew that the ruler had arrived.’ |

In this reading *sira* is the knower and *saṅ ratu* is the participant in the arrival. The arrival supplies the content of the knowledge. The past and anterior readings in the English translation depend on the narrative context.

Perception expressions call for a distinction between a stimulus and a perceived event. A thing heard can be named directly. A clause can specify what happened, including a participant and a predicate of its own. Hearing can also be a way of learning that something occurred. The distinction between direct perception and receipt of information depends on the sense of the predicate in context and on what the surrounding text says about access to the event. An arrival reported by a messenger can be the content of what someone hears, while a person's voice can be the stimulus actually heard.

| Form | Gloss | Translation |
| --- | --- | --- |
| *Aṅrĕṅö sira swara.* (constructed) | hear 3SG sound | ‘She heard a sound.’ |
| *Aṅrĕṅö sira yan datĕṅ saṅ ratu.* (constructed) | hear 3SG LINK arrive HON ruler | ‘She heard that the ruler had arrived.’ |

The examples distinguish the two readings by supplying different dependents. With a noun phrase and a following predicate, a further reading can arise in which the noun identifies a perceived participant and the predicate says what that participant was doing. In such a passage, the noun's role must be tested in both predications. A clause meaning ‘she heard that the ruler had arrived’ by itself entails no direct perception of the ruler.

Speech, cognition and perception are semantic classes. Within each class, the lexical sense and construction determine whether a phrase names a participant, gives propositional content, introduces words spoken in a separate voice, or leaves the content to be recovered from context.

## The expression and omission of complement markers

A linker at the left edge of a clause can mark a boundary while leaving the relation between clauses to be inferred. On a content reading, the clause answers what was said, thought, known or learned. On a conditional reading, it gives a circumstance under which another predication would hold. *LINK* in the glosses keeps these functions distinct from the choice of English translation.

| Form | Gloss | Translation |
| --- | --- | --- |
| *Wruh sira yan tan datĕṅ saṅ ratu.* (constructed) | know 3SG LINK NEG arrive HON ruler | ‘He knew that the ruler had not arrived.’ |
| “yan tan datĕṅa” | LINK NEG arrive-IRR | ‘If [he] had not come.’ |

In the constructed example the negated arrival is the content of the knowledge. In the second example the fragment belongs to a passage about the speaker nearly dying if a king had failed to come. The fragment gives the unrealized condition. Thus *yan* alone cannot establish that the clause is a complement. The near-death predication and the conditional interpretation of the arrival are decisive in the attested passage. (Van der Tuuk 1897, pp. 8–9)

For a candidate content clause, identify the predicate on each side of the boundary and assign the expressed participants within each clause. A speech, cognition or perception predicate favors a content reading when the following predication supplies an argument for its particular sense. A following predication may instead set the time or condition for the first event, even when the first predicate is one that can take content in other contexts. Where both readings fit, tense in an English translation or the choice between English ‘that’ and ‘when’ cannot decide the Old Javanese construction.

The verbal ending *-a* is a separate kind of marking. In “tan datĕṅa”, glossed `NEG arrive-IRR`, ‘had not come’ in the conditional context, *-a* belongs to the arrival predicate inside the clause. The same source passage has “matya”, glossed `die-IRR`, ‘would die’, in the near-death predication. The ending itself therefore does not identify the boundary of an embedded clause or specify its relation to another predicate. A linker, an ending and the semantic relation each need their own analysis. (Van der Tuuk 1897, pp. 8–9)

Three superficially similar cases arise where there is no overt content linker. A full predication may be adjacent to a predicate whose content it supplies. A quotation may stand next to its speech attribution. A speech or cognition predicate may leave its content implicit while the surrounding clauses continue the narrative. In the first case there must be a recoverable predication occupying the content role. In the second, the quoted span has the speech act and perspective of the quoted speaker. In the third, the context provides the content without any clause occupying that role at the point in question.

| Form | Gloss | Translation |
| --- | --- | --- |
| *Wruh sira.* (constructed) | know 3SG | ‘She knew.’ |
| *Wruh sira. Datĕṅ saṅ ratu.* (constructed) | know 3SG arrive HON ruler | ‘She knew. The ruler arrived.’ |

The first example can look back to a proposition already given. In the second, the intended reading has a new narrative assertion. The juxtaposition illustrates why an analysis with an unexpressed linker needs positive evidence that the second predication supplies the content of the first. The evidence available here does not establish a general rule for dropping *yan* or a complete inventory of markers and their alternatives. The dictionary also records substantial differences among manuscript readings, so a linker supplied only by an editorial reading must be distinguished from one visible in the witness. (Van der Tuuk 1897, §§ 5, 7–8)

## Direct quotation and embedded reported content

Direct quotation represents the words of a speaker or thinker from within that person's perspective. It can retain an interjection, address term or directive as part of the represented utterance. The attribution identifies the speaker or the response without turning those words into an assertion by the narrator. Its position can follow the opening of the quoted speech.

| Short phrase | Gloss | Translation |
| --- | --- | --- |
| “héh liṅ” | INTJ speech | ‘“Héh!” were the words …’ |
| “héh sahur nika” | INTJ reply GEN.THAT | ‘“Héh!” was that one's reply.’ |

The dictionary cites the first interjection as the words of someone in pain and the second as a reply. The gloss `GEN.THAT` renders the expression identifying whose reply it was. In both fragments the interjection is the represented utterance and the adjacent speech expression attributes it. A short fragment of this kind can establish the order of the interjection and attribution. The turn as a whole must be delimited from the text around it. (Van der Tuuk 1897, pp. 8–9)

Person, address and deixis provide useful evidence when punctuation leaves the boundary open. First person in a quotation refers to the quoted speaker. A second-person form in the same span is interpreted with respect to that speaker's addressee. A command or question can be part of a turn even when the narrator is simply telling what happened. In embedded reported content, a predication supplies the content of the reporting or cognition predicate. Its person forms can be resolved in the narrator's discourse. An embedded report can preserve some of a speaker's perspective, so a single person form is insufficient when both speakers share the same referent.

| Form | Gloss | Translation |
| --- | --- | --- |
| *“Tan mulih aku,” liṅ saṅ ratu.* (constructed) | NEG return.home 1SG speech HON ruler | ‘“I am not going home,” said the ruler.’ |
| *Wruh sira yan tan mulih saṅ ratu.* (constructed) | know 3SG LINK NEG return.home HON ruler | ‘She knew that the ruler was not going home.’ |

In the first example *aku* refers to the ruler through the speech attribution. In the second, *saṅ ratu* is the subject of the predication that supplies the content of *wruh*. The quotation marks in the first example display the intended boundary. Lexical attribution, turn structure and the interpretation of person and place can locate the boundary in a witness where punctuation is inconclusive.

The speech role of a nearby noun phrase needs to be established independently of the quote boundary. A recipient belongs to the predicate of speaking. A vocative belongs to the utterance in which someone is addressed. A speaker named in an attribution can also be referred to inside the quotation by a first-person pronoun. When a quoted character introduces someone else's words, the second attribution and the local interpretation of person determine the inner span. An attribution may also interrupt a continuing utterance, so a switch back into the narrator's voice can be temporary.

Lexical content, reported content and its presentation as direct speech can occur together. A noun meaning ‘reply’ classifies the speech event. The adjacent quoted words give its wording. A report can convey the same content with an embedded predicate and identify a different person as the source of the information. These distinctions are particularly useful where the text gives a reaction to speech but leaves the speech itself implicit.

## Nominalized and dependent complements

A nominal expression can name a thought, an utterance or an event and occupy a position available to noun phrases. Possession gives one diagnostic. A person expressed as the possessor of a thought can be its originator, while the person involved in the thought's content can be someone else. A content noun can refer back to a full predication in the preceding text.

| Short phrase | Gloss | Translation |
| --- | --- | --- |
| “paṅuhānireṅ twas” | conjecture-3.POSS-LOC heart | ‘The conjecture in his heart.’ |
| “maṅunyakĕn stuti” | utter praise | ‘Utter praises.’ |

In the first fragment the possessive expression identifies whose conjecture it is. The locative sequence fused into the romanized form contributes ‘in the heart’. The phrase names the conjecture without spelling out what was conjectured. In the second fragment the noun *stuti* names the material uttered. The examples show how a noun phrase can convey a content role without containing a separate predication. (Van der Tuuk 1897, pp. 9–10, 21–23)

A lexical noun for a kind of content and a nominalization of a particular predication require different evidence. For a nominalization, look for nominal behavior in the larger phrase and for a predicate or predicate-derived form responsible for its event or state interpretation. Its participants may still be understood. It may express one through possession, while others are implicit or expressed in dependents of the nominalization. A lexical noun can have a similar event meaning with fewer grounds for assigning it internal clause structure. A verbal origin or a translation using English ‘the fact that’ settles neither the construction nor the scope of any dependents.

A dependent predicate may present something commanded, intended or expected. In the following phrase the first predicate expresses dispatch, and the second specifies the role in which the dispatched person is to serve.

| Short phrase | Gloss | Translation |
| --- | --- | --- |
| “inutus jarumana” | PASS-dispatch love.messenger-IRR | ‘Sent to serve as a love messenger.’ |

The dictionary places this example among uses of *-a* on nouns that can designate a role or intended use. *Jarumana* assigns a prospective role to the participant dispatched by *inutus*. It therefore illustrates dependence with a shared participant and a nominal base in the dependent predicate. The translation uses an English infinitive to express the role. The Old Javanese form shows *-a* on the role word. The same dictionary discussion includes contemplated eventualities and intended uses, so the ending can contribute to readings outside the complement construction as well. (Van der Tuuk 1897, pp. 8–9)

The distinction between a selected dependent predicate and a purpose clause turns on what the higher predicate or construction asks the dependent to specify. In a dispatch construction the dependent can specify the task or role assigned to the person sent. A purpose clause can state the reason for an otherwise complete action and can have a different intended actor. A sequence can admit both descriptions when the assigned task is also the purpose of the dispatch. Additional evidence can come from contrasts with other dependents of the same predicate and from whether the surrounding text assigns the lower event to a participant of the higher predicate.

A dependent predicate can be identified without positing a single Old Javanese infinitive paradigm. The relevant evidence is the form of the dependent, its ability to express its own participants and its relation to the higher event. An unexpressed participant is compatible with syntactic dependence and with ordinary recovery of reference from context. The use of *-a* on a role noun, and its occurrence in an unrealized condition, give separate reasons to keep mood or intended status distinct from the question of how a complement attaches. (Van der Tuuk 1897, pp. 8–9)

## Shared arguments, control diagnostics and ambiguous predicate sequences

Argument sharing describes a reading in which a participant of one predicate is also understood with another. Control is a stronger analysis. One predicate assigns a role to an argument, and the dependent predication requires that same participant in a role of its own. With a directive, the source of the order and the person to carry it out are distinct roles. With an intention, the person who forms the intention may also be the intended actor. The lexical construction determines which interpretation is available.

In “inutus jarumana”, glossed `PASS-dispatch love.messenger-IRR`, ‘sent to serve as a love messenger’, the dispatched person is the prospective messenger. The sender and the person dispatched have different roles in the higher event. Passive voice affects how those roles are expressed. The intended messenger is still the person sent. This gives evidence for the shared participant in this construction. It does not establish how every passive orders its arguments or which other predicates require the same sharing. (Van der Tuuk 1897, pp. 8–9)

| Candidate analysis | Relation between the predicates | Evidence to check |
| --- | --- | --- |
| Control by the higher subject | The higher subject has a role with both predicates. | Check that the higher predicate selects that participant in its own right and that the dependent requires that participant for the reading at issue. |
| Control by another higher argument | A recipient, patient or other higher argument has a role with the dependent predicate. | Compare the semantic roles under the attested voice. Ask who is directed, enabled or sent and who performs the action or takes the role. |
| Independent content clause | The lower predicate has its own participant structure, with an overt or contextually understood subject. | Look for passages with separate participants. An overt lower subject with a different referent demonstrates that the reading in that passage allows independence. |
| Raising | The participant belongs semantically to the lower predication and is expressed in a higher syntactic position. | Establish the higher position and test whether the higher predicate independently selects a participant in that position. |
| Ellipsis across linked predications | An argument omitted from one predication is recovered from discourse. | Check whether another salient participant can fill the role and whether the predications can carry separate assertions, commands or temporal settings. |

A raising analysis requires evidence about both syntactic position and semantic selection. An apparent subject near the higher predicate might instead belong to the lower predicate or be a topic of the entire sequence. Useful comparisons include a construction with the participant expressed inside the lower clause, and lower predicates with different kinds of subjects under the same higher predicate. An analysis is stronger if the higher predicate contributes a judgment about the whole situation while imposing no independent participant role on that phrase. The phrases cited here supply no sequence in which those conditions together establish raising.

The distribution of negation or modal marking can help delimit the predications. In a report, negation can belong to the event reported or to the predicate of reporting. In a sequence of actions, scope over one action can support a distinct predication. Scope over the whole sequence can be compatible with shared clause structure or with a single operator applying to coordinated events. The presence of one expressed subject, one negative or one modal form therefore has to be assessed alongside its interpretation and the range of alternative construals.

The dictionary gives a short sequence with a command to return and tell someone.

| Short phrase | Gloss | Translation |
| --- | --- | --- |
| “laku waluy warah sira” | go return tell 3SG | ‘Go back and tell her.’ |

The addressee of the directive is understood as the mover and the teller. *Sira* is the person to be told. Motion and return together specify the movement, and the telling can be understood as the next action or as the purpose of the movement. The phrase supplies clear evidence of juxtaposed predicates and a shared actor. The same span alone gives little evidence for a unique division between motion with direction, a purpose dependent, and successive directives. An English translation with ‘and’ or ‘to’ chooses a rendering of the relation without establishing an Old Javanese conjunction or complement marker. (Van der Tuuk 1897, pp. 9–10)

A serial verb analysis treats verbs as predicates within one clause, typically with some shared arguments or clause-level operators. To distinguish that analysis from an adjacent complement or asyndetic coordination, first identify each verb and its arguments. Then test whether one predicate fills a role selected by the other, whether either event is independently asserted or commanded, and whether the scope of polarity, modality and temporal expressions can be established. A directional verb can make the path of a motion more specific. A purpose predicate introduces the intended action. Two neighboring commands can each have directive force even with a single expressed or understood addressee. The mere adjacency of verbs and continuity of the actor are compatible with all three structures.

Where the readings remain open, retain the evidence the text supplies. A rendering such as ‘go back and tell her’ can record the shared actor and the recipient even when the clause boundary is uncertain. A spelling variant, a parallel passage or an overt participant in another witness can resolve a boundary that a short phrase leaves undetermined. The dictionary's references to variant manuscript readings make the textual locus material to such comparisons. (Van der Tuuk 1897, §§ 8, 15)

## How Mova handles complementation and predicate sequences

`tools/mova-parse` uses `tools/langseg` to assign languages to sentences and spans. Its second level uses a model for the language assigned to a span. Identification depends on the scripts and lexicons that the segmenter supports. Its Latin-script distinction covers English and French.

There is no second-level model for `kaw`. `tools/mova-parse` reports the missing model in the `# parse` comment. Tokens delivered for an unparsed span have `_` in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL, and `Parse=None` in MISC. Mova does not tokenize, tag, lemmatize or parse Old Javanese itself. It holds no annotated Old Javanese treebank and has no measured baseline for `kaw`.

Squid reads the grammatical points in this chapter as claims. Formalizable claims wait in its queue for implementation. No rule from this chapter is implemented, so the output gives no analysis of complement marking, quotation boundaries, nominalized or dependent complements, or argument sharing in predicate sequences. Tokenization, tagging and a Universal Dependencies treebank baseline are the first needs. `prag` and `world` read English trees and have no Old Javanese parse to read.

## Sources for this chapter

- Humboldt, Wilhelm von. 1838. Über die Kawi-Sprache auf der Insel Java, nebst einer Einleitung über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwickelung des Menschengeschlechts. Vol. 2. Berlin: Königliche Akademie der Wissenschaften, in Commission bei Ferdinand Dümmler. — public domain; not held in the library, to verify
- Brandes, Jan Laurens Andries. 1913. Oud-Javaansche oorkonden: Nagelaten transcripties van wijlen Dr. J. L. A. Brandes. Edited by N. J. Krom. Verhandelingen van het Bataviaasch Genootschap van Kunsten en Wetenschappen 60(1). Batavia: Albrecht & Co. The Hague: Martinus Nijhoff. — public domain; not held in the library, to verify
- Herman Neubronner van der Tuuk; J. L. A. Brandes (editor); D. A. Rinkes (editor) 1897. *Kawi-Balineesch-Nederlandsch Woordenboek, deel 1 (Batavia 1897)* — Public domain (EU life+70: the author Herman Neubronner van der Tuuk died 1894, the editors Jan Laurens Andries Brandes died 1905 and Douwe Adolf Rinkes died 1954)
- [← previous 18. Predicate types, argument structure and clause order](https://md.mova.elamur.ai/free/en/docs/languages/jv/kaw/18-simple-clause/)
- [next → 20. Coordination and adverbial subordination](https://md.mova.elamur.ai/free/en/docs/languages/jv/kaw/20-clause-linkage/)
- [↑ back to Old Javanese (Kawi) (kaw) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/jv/kaw/)

> 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)
