Mova0.6 pre-release

the flag of Georgia 25. Reference, narrative and dialogue

Old Georgian texts identify participants through names, descriptions, demonstratives, personal pronouns and the person marking of the finite verb. The choice among these resources depends on which participants are already accessible and what the clause contributes to the discourse. Narrative and dialogue also require the reader to track changes in time, viewpoint and speaker. In this chapter, a referent can be a person, an object or, where the expression permits it, an event or a whole situation.

Introducing and tracking participants

A name can identify somebody assumed to be known or introduce that person into the current episode. A common noun phrase can introduce a participant by role or description, such as a disciple or a person with an illness. A bare noun phrase can also pick out somebody already known. The absence of a demonstrative leaves the reading of the phrase to the predicate and the surrounding discourse.

An existential clause can establish a participant before that participant does anything in the story. An event clause can introduce a participant through the event itself. In the following passage the noun in the first clause establishes a referent. The next verb supplies an action and leaves the actor to be recovered from that clause.

იყო კაცი. მოვიდა.
iqʼo kʼaci. movida.
be.PST.3SG man.NOM. come.AOR.3SG.
“There was a man. He came.” (constructed)

A participant can be identifiable on first mention through an expression that relates them to somebody already known. This is an inference from the relationship, distinct from coreference. A daughter and her father, for example, remain different referents even if the mention of one gives the reader access to the other. In the account of Jairus, his request establishes his daughter as the person whose condition motivates the journey. Subsequent talk about the child, including by speakers other than Jairus, tracks the same participant (Blake 1929, § Mark 5:22–24, 35–43).

Once established, a participant can be tracked across several clauses by verb forms alone. Full noun phrases can recur within such a sequence. A repeated name may restore a participant after an interruption, and a fresh description may identify a known participant in a role that matters at that point. The participants in the account of Ashot and Febronia include the ruler, the woman whom Gregory takes to Mere and Febronia, who receives her there. As the scene moves between Gregory’s intervention and Ashot’s encounter with Febronia, the relevant participants and their roles change with the action (Marr 1911, § on Ashot and Febronia).

Discourse accessibility can differ from mention to mention. A participant may remain central to the episode while another is the immediate actor. A participant with no overt expression in one clause can become the object of a full description in the next. Conversely, an overt noun phrase can maintain an established topic through a stretch in which several other people act or speak.

Names, demonstratives, personal pronouns and zero anaphora

Names and descriptions

A name or descriptive noun phrase is useful when the reader needs enough lexical information to choose among plausible referents. Names also occur where the identity is already clear, for example at a new episode or a new speaking turn. Descriptions contribute their ordinary lexical meaning even when they refer back to a named person. Whether a title and a name pick out the same person is a question of reference in the passage.

Demonstratives and personal pronouns

The table gives independent personal forms of the first and second persons and the nominative singular head forms of the three demonstrative series. The person contrast is anchored to the current speech situation. Demonstrative distance can be understood in the physical scene, through association with a speech participant, or through the place a referent occupies in the discourse.

FormRomanizationGlossDiscourse use
მეme1SGThe current speaker
შენšen2SGOne current addressee
ჩუენčuen1PLThe speaker together with others
თქუენtkuen2PLThe current addressees as a plural group
ესეeseDEM.PROX.NOM.SGPoints to what is near the speaker or presented as close at hand
ეგეegeDEM.MED.NOM.SGPoints to what is associated with the addressee or presented at intermediate distance
იგიigiDEM.DIST.NOM.SGPoints to what is presented as distant or identifiable from the context, and can serve as a third-person pronoun

The demonstrative series can occur with a noun or independently. A postnominal demonstrative can mark the noun phrase as identifiable with little emphasis on physical distance. The same form used independently can resume an established participant. Its case belongs to its own clause. A change in case from the antecedent to the anaphor is expected when their syntactic roles differ. Neither the demonstrative series nor the third-person verb form identifies a participant as male or female. Gender in an English translation comes from the referent.

მოვიდა კაცი. კაცი იგი დაჯდა.
movida kʼaci. kʼaci igi daǰda.
come.AOR.3SG man.NOM. man.NOM DEM.DIST.NOM.SG sit.down.AOR.3SG.
“A man came. The man sat down.” (constructed)

მოვიდა კაცი. იგი დაჯდა.
movida kʼaci. igi daǰda.
come.AOR.3SG man.NOM. DEM.DIST.NOM.SG sit.down.AOR.3SG.
“A man came. He sat down.” (constructed)

A demonstrative can point to something visible without a preceding verbal mention. A demonstrative that refers back to what has just been said can also take a larger stretch of discourse as its antecedent, such as a reported event. In that use, resolving reference requires identifying the relevant situation rather than finding a noun with matching features.

Independent first- and second-person pronouns can identify a speaker or addressee explicitly where that identity matters to the utterance. The verb may already mark the same person. An overt pronoun can support contrast, an assertion about the speaker in particular, or a change in the person under discussion. Its presence alone does not determine which of these readings applies.

Zero anaphora

A finite verb can be used without an independent noun phrase or pronoun for a participant recoverable in context. Person marking can make a speaker or addressee recoverable within a speech turn. In the third person, the verb can give grammatical information about an argument while leaving its identity to the discourse. The roles available for recovery depend on the verb’s valency and the actual person and number distinctions of the form.

პეტრე მოვიდა. დაჯდა.
pʼetʼre movida. daǰda.
Peter.NOM come.AOR.3SG. sit.down.AOR.3SG.
“Peter came. He sat down.” (constructed)

The same effect can occur with an object. In the next example the subject of the second verb is read as the man, and its contextually supplied object as the disciple. The construction also allows the actor or the object to be identified differently if the preceding discourse supplies a suitable referent.

მოვიდა მოწაფე. კაცმან იხილა.
movida mocʼape. kʼac-man ixila.
come.AOR.3SG disciple.NOM. man-ERG see.AOR.3SG.
“A disciple came. The man saw him.” (constructed)

Zero anaphora concerns a referential participant understood in the clause. To recognize it, first establish that the verb in that use makes the relevant role available. A verb used without a referential participant in a given role supplies no antecedent to recover for that role. An omitted argument with an indefinite reading can also lack a particular discourse antecedent.

Changes of referent and ambiguous antecedents

An explicit expression can signal a shift, but expressions of every degree of explicitness can occur during continued reference. An overt pronoun is compatible with keeping the current participant. A zero is compatible with a shift when the predicate, the structure of the exchange or the scene makes the new participant recoverable. A name or description gives the clearest indication of a shift when two plausible antecedents share the features expressed by a pronoun or verb.

პეტრე მოვიდა. იოვანეცა მოვიდა. იგი დაჯდა.
pʼetʼre movida. iovane-ca movida. igi daǰda.
Peter.NOM come.AOR.3SG. John.NOM-ADD come.AOR.3SG. DEM.DIST.NOM.SG sit.down.AOR.3SG.
“Peter came. John came too. He sat down.” (constructed)

John is the last person mentioned, and both men satisfy the grammatical requirements of the final clause. The sequence in isolation leaves room for either reading. In a longer passage, a preceding topic, the location of the participants or a subsequent clarification can decide between them. Repeating one of the names in the last clause would settle the choice directly. Omitting the demonstrative would still leave two possible third-person actors.

To work out an antecedent, identify the argument role and any person or number evidence provided by the current clause. Then consider who can fill that role given the predicate, the current episode and any speech in progress. Case marks the function of an overt expression in its own construction. An earlier nominative can correspond to a later ergative, dative or unexpressed argument. With a singular third-person verb and two possible singular participants, linear recency is only one consideration. Where the accessible participants and the action leave two readings open, the passage remains ambiguous.

The span relevant to an antecedent can extend across an interruption. In the episode of Jairus and the woman who touches Jesus’ garment, the journey prompted by Jairus is interrupted by another participant’s action and dialogue. The arrival of messengers associated with Jairus restores the first situation. The child can then be identified from the resumed situation despite the intervening attention to the woman (Blake 1929, § Mark 5:22–43).

A shift in topic and a shift in grammatical subject need separate judgments. A story can keep a participant at its center while messengers or interlocutors successively become subjects. At a speech boundary, the next speaker may be the addressee of the preceding turn. First-person expressions in the two turns will then refer to different people, even if the narrator continues to use third person for both.

Event sequence, background and discourse connectives

A succession of event clauses commonly invites a forward reading when it follows a single line of action. The reader can recover that order with no explicit time adverb. A connective can make continuation overt, and a time adverb can specify succession. The interpretation also depends on the verbs and on what the preceding clauses have already established. A clause may supply the circumstances of an event, explain it, or take the reader back to an earlier event.

მოვიდა კაცი და დაჯდა.
movida kʼaci da daǰda.
come.AOR.3SG man.NOM and sit.down.AOR.3SG.
“A man came and sat down.” (constructed)

Here coming and sitting are read in the order given. The two predicates and the shared actor support that interpretation. The coordinator itself also joins material with other temporal relationships.

FormRomanizationGlossContribution in discourse
დაdaandLinks material to what precedes. In a sequence of events it can carry the narration forward.
ხოლოxoloand, but, nowCan set a participant, event or turn against the preceding one, or mark a transition to it.
მერმეmermethen, afterwardPlaces the following event or phase after a time or event already in view.
მაშინmašinthen, at that timeLocates an event at a time supplied by the context. The context can support a reading of immediate continuation.
რამეთუrametufor, becauseCan introduce a ground or explanation for the preceding assertion or event.
რაჟამსražamswhenIntroduces a temporal clause that provides a reference time for another event or state.

The table gives common functions relevant to narrative. The English equivalent of a connective depends on the relation in the passage. In particular, a switch signaled by xolo can amount to a contrast between actions, a change of participant, or the next contribution to a conversation. The connective itself leaves the identity of the participant to the clause and its context.

Background includes settings, ongoing states and activities that hold when another event occurs. An imperfect form can present such a state or activity, and an aorist can present an event within that setting. Foreground and background are discourse functions. A completed event can occur in an explanatory passage, and a state can be central to what a passage asserts.

რაჟამს მოვიდა მოწაფე, კაცი იჯდა.
ražams movida mocʼape, kʼaci iǰda.
when come.AOR.3SG disciple.NOM man.NOM sit.IMPF.3SG.
“When the disciple came, the man was sitting.” (constructed)

The arrival fixes a reference time. The sitting is presented as in progress at that time. The order in which the predicates occur therefore need not be the order in which the situations began.

კაცი დაჯდა, რამეთუ სნეული იყო.
kʼaci daǰda, rametu sneuli iqʼo.
man.NOM sit.down.AOR.3SG for ill.NOM be.PST.3SG.
“The man sat down because he was ill.” (constructed)

The state gives a reason for the action, even though the account mentions it afterward. A larger retrospective passage can do the same work with a chain of events. The account of John’s death follows Herod’s reaction to reports about Jesus and supplies the earlier history that makes the reaction intelligible. Within the retrospective passage, successive clauses can again tell events in forward order (Blake 1929, § Mark 6:14–29).

Direct speech, reported speech and speaker transitions

Direct speech and its reference

A speech introduction can name the speaker, identify the recipient, or leave either to be supplied by the context. A reporting verb gives evidence about the direction of the exchange. A verb of answering also relates the new turn to something just said or asked. The content of the utterance can begin immediately after the introduction. In reading an edition, use the reporting construction, person forms and flow of the exchange to establish the boundary of the speech. The punctuation in the examples below marks the boundaries for the reader.

ჰრქუა მოწაფემან პეტრეს: მე ვხედავ.
hrkua mocʼape-man pʼetʼre-s: me vxedav.
say.AOR.3SG disciple-ERG Peter-DAT 1SG see.PRS.1SG.
“The disciple said to Peter, ‘I can see.’” (constructed)

The narrator refers to the disciple in the third person through the reporting clause. Inside the speech the disciple refers to himself in the first person, and Peter is the person to whom a second-person expression could refer. The same anchoring applies to demonstratives and to expressions whose interpretation depends on where or when the utterance is made. When the narrator resumes, the surrounding narrative again supplies the context for those expressions.

Reporting the content of speech

A speech report can give the content as a clause within the report. A content introducer such as vitarmed can stand at the transition from a speech verb to its content. Old Georgian can also use an introducer before words presented as direct speech. To determine how the reported material is presented, look at its person forms and other expressions that depend on the speech situation. Where all the relevant participants are third persons, the distinction may depend on a wider stretch of text.

თქუა, ვითარმედ ხედავს.
tkua, vitarmed xedavs.
say.AOR.3SG COMP see.PRS.3SG.
“He said that he could see.” (constructed)

On the translated reading, the report gives the speaker’s perception in the narrator’s third person. The person forms by themselves also allow the person who sees to be somebody else, if that person is accessible in the surrounding discourse. Third-person agreement supplies no identity relation between the subject of the speech verb and the subject of the content clause.

A narrator can also report that the act of speaking took place without giving the words, or identify the kind of contribution by describing it as a request, question or answer. A single scene can move among these possibilities. In the encounter with Bartimaeus, the narration describes attempts to silence him and also presents utterances by Bartimaeus, the people who summon him and Jesus. The different sources of speech matter for tracking both the command to come and the ensuing question and reply (Blake 1929, § Mark 10:46–52).

Speaker transitions

A new speech turn can be introduced with a full speaker expression and an overt addressee. Once the roles are established, the context can supply a speaker or recipient left unexpressed. An answer to a question supplies evidence for a turn change, especially in an exchange with two possible speakers. Where several participants are present, names, descriptions and the direction of a reporting verb give additional evidence. A connective alone does not determine who speaks.

ჰრქუა მოწაფემან პეტრეს: ვხედავ. ხოლო პეტრემან ჰრქუა მას: მეცა ვხედავ.
hrkua mocʼape-man pʼetʼre-s: vxedav. xolo pʼetʼre-man hrkua mas: me-ca vxedav.
say.AOR.3SG disciple-ERG Peter-DAT see.PRS.1SG. and Peter-ERG say.AOR.3SG 3SG.DAT 1SG-ADD see.PRS.1SG.
“The disciple said to Peter, ‘I can see.’ And Peter said to him, ‘I can see too.’” (constructed)

The first verb of seeing has no independent subject. Its first-person marking is understood relative to the disciple’s turn. Peter’s name in the second introduction identifies the new speaker, and the dative pronoun refers back to the disciple, now the addressee. The first-person form inside Peter’s speech therefore has a different referent. The additive particle relates Peter’s claim to the disciple’s preceding claim.

Brief answers can rely on the preceding turn for their predicate or other material. At Caesarea Philippi the question about Jesus’ identity establishes what the disciples are being asked to identify. Peter’s reply fills that place in the exchange. His role as speaker is separate from the identity being asserted in the reply (Blake 1929, § Mark 8:27–30). When speech is quoted within another speech, first and second person likewise have to be assigned within the relevant quotation. The immediately surrounding speech introduction and the content of the exchange help establish which utterance supplies that context.

How Mova handles reference and discourse relations

tools/mova-parse uses tools/langseg to assign languages to sentences and spans at level 1. tools/langseg classifies Georgian script as other. Level 2 requires a parser model for the language of a span, and no such model exists for oge. For a language identified without a model, tools/mova-parse reports the missing parse in the # parse comment. It leaves LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL as _ and sets Parse=None in MISC. Mova holds no annotated Old Georgian treebank and has no measured treebank baseline for the language.

Mova has no Old Georgian tokenization, tagging, lemmatization or parsing in this pipeline. Its frame, pragmatics and world components, including prag and world, read English trees and require a parsed input to have anything to read for another language. The rules in this chapter for referring expressions, omitted arguments, antecedents, discourse relations and speech turns have no implementation for oge.

Squid reads the manual and turns its grammatical points into claims. Formalizable claims wait in its queue for a place in Mova’s code. Queueing records a claim for possible implementation. The first work for a language at this stage concerns what a tokenizer, a tagger and a Universal Dependencies treebank baseline require.

Sources for this chapter

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/ka/oge/25-reference-and-discourse/