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/ko/okm/15-predicate-inflection/)

[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/) / [Korean](https://md.mova.elamur.ai/free/en/docs/languages/ko/) / [okm](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/) / 15. Predicate inflection and the ending system

# 15. Predicate inflection and the ending system

**On this page**

1. [Predicate stems and the order of suffixes](#predicate-stems-and-the-order-of-suffixes)
2. [Tense, aspect and auxiliaries](#tense-aspect-and-auxiliaries)
3. [Mood, modality and evidential interpretations](#mood-modality-and-evidential-interpretations)
4. [Person-sensitive forms and their distribution](#person-sensitive-forms-and-their-distribution)
5. [Prefinal honorific markers and final speech levels](#prefinal-honorific-markers-and-final-speech-levels)
6. [Declarative, interrogative and directive endings](#declarative-interrogative-and-directive-endings)
7. [Connective, adnominal and nominalizing forms](#connective-adnominal-and-nominalizing-forms)
8. [Disputed analyses and contextual readings](#disputed-analyses-and-contextual-readings)
9. [How Mova handles predicate inflection](#how-mova-handles-predicate-inflection)
10. [Sources for this chapter](#sources-for-this-chapter)

Middle Korean verbs, descriptive predicates and the copula take endings that mark how a predication fits into a sentence. A predicate can end an utterance, link one clause to another, modify a noun or form a nominal expression. Markers closer to the stem can contribute time, aspect, modality and honorification. Some endings are sensitive to the person of the subject or to the role of a noun modified by the clause. The 1459 vernacular preface to the *Hunminjeongeum* illustrates several of these distinctions within a single passage (text) 1459, § preface).

## Predicate stems and the order of suffixes

### Stems and predicate classes

The stem carries the lexical meaning of the predicate. The ending complex specifies its grammatical use. In a simple form, the stem and the final ending are adjacent. In a more elaborate form, one or more markers intervene. A **prefinal marker** is a suffix that occupies this intervening position and needs a following ending. The term **ending** also covers suffixes that close a dependent clause, such as an adnominal ending before a noun.

Verbal stems include predicates of action, motion, change and existence. Descriptive stems predicate a property or state and can themselves head a clause. They inflect directly, as in the preface’s “하니라” *hanira* `be.many-ASSERT` ‘there are many’. A nominal predicate normally has a copular element, seen in “ᄯᆞᄅᆞ미니라” *stʌrʌminira* `only.matter-COP-ASSERT` ‘that is all’. The first form contains the descriptive stem 하- *ha-* ‘be many’. The second contains the dependent noun ᄯᆞᄅᆞᆷ *stʌrʌm* and copular 이- *i-*. Both have the ending -니라 *-nira* (text) 1459, § preface).

| Predicate type | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Action or motion | 가다 | *kata* | `go-DECL` | ‘goes’ (constructed) |
| Descriptive | 하니라 | *hanira* | `be.many-ASSERT` | ‘there are many’ (text) 1459, § preface) |
| Existential | 이셔도 | *isyəto* | `exist-CONCESSIVE` | ‘although there is’ (text) 1459, § preface) |
| Copular | ᄯᆞᄅᆞ미니라 | *stʌrʌminira* | `only.matter-COP-ASSERT` | ‘that is all’ (text) 1459, § preface) |

The romanization here follows the written forms. It uses *ʌ* for ㆍ, *ə* for ㅓ, *ɨ* for ㅡ, *ŋ* for old ㆁ, *z* for ㅿ and *β* for ㅸ. In *lq*, the *q* records the written ㆆ of the combined coda ᇙ. The transcription of that spelling does not specify a pronunciation for the ending in every context. Hyphens in forms cited as suffixes show where they attach. Hyphens in glosses separate grammatical contributions, including contributions expressed together on the surface.

### Suffix order

The basic order is **stem–prefinal material–clause ending**. A suffix deriving a new predicate belongs inside the stem for this purpose. It precedes the inflection of the derived predicate. A lexical stem followed by a connective and a second predicate forms a larger construction with two inflectable stems. For example, in an aspectual construction with an auxiliary, the first verb bears the connective and the auxiliary bears the ending appropriate to the whole construction.

Different prefinal markers occupy different positions. When object and subject honorification occur together on one predicate, object honorification is closer to the stem. Subject honorification normally precedes the material marking event presentation or prospectivity, as in the order stem–시–ᄂᆞ–ending. Addressee marking in the high declarative and interrogative series is close to the end of the predicate. The directive series has its own endings, and the endings of dependent clauses govern which prefinal markers they can take.

| Function in the construction | Position | Typical material | Condition on its use |
| --- | --- | --- | --- |
| Predicate derivation | Inside the inflecting stem | A causative or passive stem-forming suffix | The derived stem supplies the meaning and valency for inflection. |
| Honorification of a nonsubject | After the stem and before subject honorification where both occur | -ᄉᆞᆸ-, -ᄌᆞᆸ-, -ᅀᆞᆸ- *-sʌp-, -cʌp-, -zʌp-* | The construction licenses deference to a participant other than the subject. |
| Subject honorification | After the stem or its object-honorific extension | -시-, -샤- *-si-, -sya-* | The predicate presents its subject honorifically. The following ending helps determine the form of the marker. |
| Event presentation, retrospection or prospectivity | In the prefinal complex | -ᄂᆞ-, -더-, -리- *-nʌ-, -tə-, -ri-* | These markers have distinct distributions. The lexical predicate class and the ending constrain the combinations. |
| Person-sensitive marking | In the prefinal complex or fused with neighboring material | -오-, -우- *-o-, -u-* | Its distribution depends on the clause type and, in adnominal clauses, on the relation of the head noun to the clause. |
| Addressee honorification in high statements and questions | Toward the end of the inflection | -ᅌᅵ-, -ᅌᅵᆺ- *-ŋi-, -ŋis-* | The form depends on the final series. |
| Clause ending | At the outside of the predicate inflection | Declarative, interrogative, directive, connective, adnominal or nominalizing endings | The clause’s place in the sentence and its communicative function determine the series. |

The table gives the relative positions of recognizable material. It does not license filling every row in a single form. In particular, person-sensitive marking interacts with neighboring vowels, and some final formations are best recognized as whole endings before a finer segmentation is attempted.

The preface’s “ᄆᆡᇰᄀᆞ노니” *mʌyŋkʌnoni* `make-NPST.VOL.PERS-CONN` ‘since I make [them]’ brings together a form of the stem ᄆᆡᇰᄀᆞᆯ- *mʌyŋkʌl-* ‘make’, the form -노- *-no-* associated with -ᄂᆞ- and -오-, and connective -니 *-ni*. The stem-final ㄹ is absent before the following ㄴ. The form occurs after an overt first-person subject earlier in the passage (text) 1459, § preface).

### Identifying boundaries

Three facts matter when identifying the suffixes of a written form. First, the choice of an ending can be phonologically conditioned. A consonant stem may take an intervening vowel in a form whose vowel-stem counterpart has none. Second, a stem can alternate at the boundary, including the loss of final ㄹ before certain consonant-initial suffixes. Third, adjacent suffixes can contract, so that a single written vowel represents part of the contribution of each.

| Boundary pattern | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Vowel stem before adnominal ㄹ | 갈 | *kal* | `go-PROSP.ADN` | ‘who will go’ (constructed) |
| Consonant stem before prospective adnominal -(으)ㄹ | 먹을 | *məkɨl* | `eat-PROSP.ADN` | ‘which someone will eat’ (constructed) |
| ㄹ-stem before ㄴ-initial material | ᄆᆡᇰᄀᆞ노니 | *mʌyŋkʌnoni* | `make-NPST.VOL.PERS-CONN` | ‘since I make [them]’ (text) 1459, § preface) |
| Final consonant of a nominalizer before a vowel-initial particle | ᄡᅮ메 | *psume* | `use-NMLZ-LOC` | ‘in using [them]’ (text) 1459, § preface) |

In the last form, the written 메 *me* contains the consonant of the nominalizer and the vowel of the particle. The spelling crosses a morphological boundary inside a syllable block. The connection between this form and its base ᄡᅳ- *psɨ-* is understood through the nominalization and the following particle, which together occur in a phrase about daily use (text) 1459, § preface).

## Tense, aspect and auxiliaries

### Event presentation and relative time

A Middle Korean predicate can receive a time interpretation from its ending, a prefinal marker, an auxiliary, a time expression or the surrounding account. These sources of information work together. A final declarative form can describe a current state or a regular occurrence without a separate present marker. The form of a dependent predicate also has to be read relative to the event or state with which it is construed.

| Form or construction | Main contribution | Conditions on interpretation |
| --- | --- | --- |
| Stem with no overt tense or aspect marker | The predicate is located by its ending and context. | Common with properties, states and nominal predication. With other predicates the ending and discourse supply the temporal reading. |
| -ᄂᆞ- *-nʌ-* | Presents an action or process as current, unfolding or habitual. | Characteristic of verbal predicates. In an adnominal clause the reference point may be supplied by the containing clause. |
| -더- *-tə-* | Presents a situation from a retrospective viewpoint. | The common readings involve recall or prior experience. The speaker’s relation to the event and the utterance type affect the reading. |
| -리- *-ri-* | Presents a prospective, intended or otherwise projected situation. | Prediction, intention and inferential readings depend on the predicate, subject and clause type. |
| -아/-어 plus existential 잇- or 이시- *is-*, historically also contracted formations with -앳-/-엣- *-ays-/-əys-* | Presents a state associated with a preceding event. | A resultative reading requires a plausible resulting state. A later generalized past reading must be established from the date and context. |
| Adnominal -(ᄋᆞ/으)ㄴ *-(ʌ/ɨ)n*, -ᄂᆞᆫ *-nʌn* and -(ᄋᆞ/으)ㄹ *-(ʌ/ɨ)l* | Relate the modifying predication to a nominal head and a reference point. | The ㄴ series has different values with action and descriptive stems. The ㄹ series includes prospective and modal uses. |

A bare descriptive form such as “하니라” *hanira* `be.many-ASSERT` ‘there are many’ describes a state valid in the situation under discussion. The preface also uses “ᄆᆡᇰᄀᆞ노니” *mʌyŋkʌnoni* `make-NPST.VOL.PERS-CONN` ‘since I make [them]’ for an act that supplies the grounds for the following purpose. The English rendering may present the making as already accomplished. The temporal relation between that act and the intended use helps determine the translation (text) 1459, § preface).

### The distribution of -ᄂᆞ-

The marker -ᄂᆞ- occurs with an action or process in such formations as finite -ᄂᆞ다 *-nʌta* and adnominal -ᄂᆞᆫ *-nʌn*. A present English translation can express an ongoing event, a habit or a general truth. Selection among those readings depends on time expressions, the kind of activity and the larger clause. A motion verb, for instance, may describe a journey under way or a regular route.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가ᄂᆞ다 | *kanʌta* | `go-NPST.VOL-DECL` | ‘goes’ or ‘is going’ (constructed) |
| 먹ᄂᆞᆫ 사ᄅᆞᆷ | *məknʌn sarʌm* | `eat-NPST.VOL.ADN person` | ‘a person who is eating’ or ‘a person who eats’ (constructed) |
| 큰 집 | *kɨn cip* | `be.big-ADN house` | ‘a large house’ (constructed) |

`NPST.VOL` in these glosses labels the verbal nonpast formation, including ordinary ongoing and habitual readings. It carries no claim that the subject acts voluntarily. The descriptive predication in the last row takes an adnominal ending in ㄴ directly. This is the same written coda as the anterior adnominal series with action verbs. The stem class is part of the interpretation.

### Retrospection and projection

A form in -더- frames the situation as something viewed from an earlier point. In a statement, the speaker is ordinarily the person whose earlier observation or experience supplies that point of view. Fictional narration and quotation can establish a viewpoint within the narrated or quoted situation. With a state, the remembered fact can remain true at the time of speech. With an event, the perspective can be on an event in progress or on an occurrence recalled as a whole, as context permits.

The prospective -리- has a different orientation. With a first-person subject and a controllable action, it can convey intention. With another subject, or with a predicate involving a state or an event outside anyone’s control, it readily conveys expectation or prediction. A question with prospective marking can ask what will happen, what someone is willing to do or what is likely. In a dependent construction, the reference point may be in the past of the narrator even when the projected event is later than that point.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가더라 | *katəra* | `go-RETRO-DECL` | ‘I recall that [someone] was going’ (constructed) |
| 가리라 | *karira* | `go-PROSP-DECL` | ‘[someone] will go’ or ‘I intend to go’ (constructed) |
| 먹을 것 | *məkɨl kəs* | `eat-PROSP.ADN thing` | ‘something to eat’ (constructed) |

The prospective form in the last row is adnominal. Its interpretation can concern suitability or availability even when the eating event is never realized. A finite translation with *will* would lose that property of the nominal expression.

### Resultative and other auxiliary constructions

A lexical predicate can be followed by a second predicate that locates the result, continuation, completion, possibility or intended realization of the first. The form of the lexical predicate is selected by the construction. In the resultative pattern, -아/-어 attaches to the lexical stem and the existential predicate supplies the state predication. In a desire construction with -고져 *-kocyə*, the higher predicate takes the inflection for the clause. The same general division of labor is visible in long negation, with a predicate in -디 *-ti* followed by an inflecting negative predicate.

| First predicate | Following predicate | Contribution of the combination |
| --- | --- | --- |
| Stem plus -아/-어 *-a/-ə* | An existential stem | A condition holding as the result of the lexical event, when the event can produce such a condition. |
| Stem plus -고져 *-kocyə* | ᄒᆞ- *hʌ-* ‘do’ | A wish, intention or purpose involving the lexical event. |
| Stem plus -디 *-ti* | 아니ᄒᆞ- *anihʌ-* or 몯ᄒᆞ- *mothʌ-* | Negative predication or inability, according to the following predicate. |

The resultative pattern can be illustrated by 안자 잇다 *anca ista* `sit.down-CVB exist-DECL` ‘is seated’ (constructed). The lexical verb provides entry into a position. The existential predicate situates the position as a state. When an existential predicate inflects, its inflection belongs to that predicate even though the larger construction is translated as a single English verb phrase. Written formations in -앳-/-엣- reflect the close historical connection of the connective vowel and the existential element. They must be assessed in context for the extent to which the resultative reading is retained.

In the preface, “니르고져” *nirɨkocyə* `say-DESID` ‘wish to say’ supplies the lower predicate of a construction whose higher predicate appears as “호ᇙ” *holq* `do-O-PROSP.ADN` ‘[which someone] is minded to do’. The higher predicate modifies a dependent noun, making the whole expression refer to what people wish to say (text) 1459, § preface). The first ending identifies the desired action. The second identifies the relation of the higher clause to its nominal head.

## Mood, modality and evidential interpretations

### Clause mood and the speaker’s commitment

Final endings distinguish statements, questions and directives, with further contrasts inside the series. The form identifies the grammatical presentation of the utterance. The surrounding discourse identifies what the utterance is intended to accomplish. A declarative can announce an intention or make a prediction. A question can request information, seek confirmation or make a point for which the answer is taken to be evident. A directive can be a command, a request or an exhortation.

| Grammatical formation | Typical communicative use | Reading that requires context |
| --- | --- | --- |
| Declarative | Presents a proposition as an assertion. | A promise, warning or expression of resolve may be conveyed by a projected action. |
| Polar interrogative | Presents a proposition for confirmation or denial. | A confirmation question or rhetorical use depends on what interlocutors already accept. |
| Content interrogative | Asks for the value of a questioned expression. | With an obvious or unavailable answer, it can express surprise, challenge or emphasis. |
| Imperative | Directs an addressee toward an action. | The authority, entreaty or degree of insistence depends on the situation and the address forms. |
| Hortative | Proposes an action involving the addressee and the speaker or the speaker’s group. | The size and membership of the group are supplied by the discourse. |
| Prospective formation | Presents a projected state or event. | Intention, prediction or inference follows from the kind of subject, predicate and clause. |

A suffix in the interrogative series can keep its form inside quoted speech. The matrix verb that introduces or reports the speech then bears its own inflection. In passages without explicit quotation marks, the switch in clause mood, person and speech level can help identify the scope of the quotation. It works together with lexical cues from the reporting predicate and the structure of the surrounding passage.

### Desire, purpose and possibility

The ending -고져 introduces a desired state of affairs. A following ᄒᆞ- can complete the expression and inflect for the construction that contains it. If a purpose clause identifies the intended consequence of an earlier event, the agent who has the purpose is determined by the clause structure and discourse. The subject of the intended event can be someone else. In the preface the resolve to provide the letters belongs to the first-person narrator, while the acts of learning and using them are assigned to others. The passage uses “ᄒᆞ고져” *hʌkocyə* `do-DESID` ‘wish to do’ in the final expression of purpose (text) 1459, § preface).

Prospective adnominals can delimit the available or intended uses of a thing. The comparative strength of willingness, possibility and obligation is often contributed by a noun or predicate outside the prospective form. A construction built on an expression for ability has that lexical contribution even when it uses the same prospective ending as an intended activity.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가고져 ᄒᆞ다 | *kakocyə hʌta* | `go-DESID do-DECL` | ‘wants to go’ (constructed) |
| 먹을 것 | *məkɨl kəs* | `eat-PROSP.ADN thing` | ‘something to eat’ (constructed) |
| 펴디 | *phyəti* | `express-COMP` | ‘express’, as the complement of a predicate of inability (text) 1459, § preface) |
| 몯ᄒᆞᇙ | *mothʌlq* | `be.unable-PROSP.ADN` | ‘who cannot [do so]’ (text) 1459, § preface) |

The last form qualifies people described as unable to express what they mean. In this setting the ㄹ formation has a dispositional or modal interpretation. The assertion about many such people is supplied by the higher descriptive predicate “하니라” *hanira* `be.many-ASSERT` ‘there are many’ (text) 1459, § preface).

### Evidential readings

The retrospective -더- is the main form in this set for which the source of knowledge regularly affects interpretation. A speaker who describes an earlier scene can present it as recalled experience. A speaker asking a retrospective question can seek the addressee’s recollection. Reporting a retrospective statement in direct speech can retain the original speaker’s vantage point. Analysing these uses requires keeping the time of the situation, the time of observation and the time of utterance distinct.

The relationship to evidence is different for prospective -리-. A prediction may be based on presently available signs, and a conjecture may concern something the speaker cannot directly verify. The inferential force in such readings comes from the projection and its context. A prospective form with a speaker’s controlled action can instead communicate intention on the basis of the speaker’s own plan.

| Construction | Interpretation available in context | Contextual question |
| --- | --- | --- |
| Retrospective statement | Recollection of a previously observed or experienced situation | Whose earlier vantage point does the statement express? |
| Retrospective question | Request for a report of what was experienced or recalled | Whose recollection is being sought? |
| Prospective statement | Expectation or inference | What grounds does the discourse provide for the projection? |
| Prospective with a controllable first-person action | Intention or resolve | Is the speaker presenting a plan or estimating what will happen? |
| Reported utterance | Information attributable to a quoted speaker | Which person supplies the perspective within the reported clause? |

There is no obligatory ending choice between direct and reported evidence across all clauses. An unmarked assertion can contain knowledge from different sources. An evidential reading can be recorded for a particular form and context without assigning that reading to every occurrence of the ending.

## Person-sensitive forms and their distribution

### Person, number and agreement

Most predicate forms are compatible with a subject of any person or number. The subject can be expressed by a noun or pronoun, or recovered from the discourse. Subject honorification makes a further distinction in how a referent is treated. Its presence depends on deference to that referent. Number is expressed where needed in nominal material, numerals or context, with no regular singular and plural conjugations for verbs.

| Subject | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| First person | 나도 가다 | *nato kata* | `1SG-ADD go-DECL` | ‘I go too’ (constructed) |
| Third person, singular reading | 그 사ᄅᆞᆷ도 가다 | *kɨ sarʌmto kata* | `that person-ADD go-DECL` | ‘that person goes too’ (constructed) |
| Third person, plural reading | 두 사ᄅᆞᆷ도 가다 | *tu sarʌmto kata* | `two person-ADD go-DECL` | ‘two people go too’ (constructed) |

Two environments with stronger person effects are forms containing -오/-우 and direct questions with a second-person subject. The effects have different causes and different distributions. A first or second person inferred from context can be grammatically relevant even when no overt pronoun accompanies the predicate.

### -오/-우 in statements and other clauses

In independent statements and some linked clauses, -오/-우 has a strong association with a first-person subject or a clause anchored in the speaker’s participation. The relevant person is that of the clause carrying the suffix. A following clause may change subject, and a preceding topic may continue over several clauses. Direct quotation supplies its own speaker and addressee. The first-person association can therefore point to a quoted speaker even when the full passage is narrated by someone else.

The choice between *o* and *u* is connected with vowel harmony and with the shape of surrounding morphemes. A visible *o* in a syllable may belong to a stem or a separate ending. Identification of the marker depends on the base, the ending, the clause type and possible contraction. The preface’s “ᄆᆡᇰᄀᆞ노니” *mʌyŋkʌnoni* `make-NPST.VOL.PERS-CONN` ‘since I make [them]’ has -노- *-no-* in a first-person clause. The narrator has already identified the maker as the speaker (text) 1459, § preface).

| Form or environment | Distribution | Interpretation to establish |
| --- | --- | --- |
| -오- or -우- in a statement | Favored in a speaker-anchored clause, especially with first-person subject. | Identify the relevant speaker and check which clause has that subject. |
| -노- *-no-* in the verbal nonpast complex | Surface combination associated with -ᄂᆞ- and -오-. | Check the verbal stem and the ending following the contraction. |
| -오- or -우- in a question | A person interpretation can be oriented to the addressee in a question about the addressee’s own participation. | Distinguish the person being questioned from a participant being asked about. |
| An *o* or *u* element in an adnominal clause | Distribution depends on the relation between the nominal head and the predication. | Determine the role of the head and whether the relative clause has a separate subject. |
| The vowel of -옴/-움 *-om/-um* | Part of the form used for nominalization. | Identify nominal syntax before using the vowel as evidence of person. |

The first-person association is strongest as a description of a clause type and its perspective. It does not provide a complete set of person endings. Forms without the marker can have a first-person subject. A form with the same vowel sequence may have a different grammatical analysis.

### The adnominal distribution

In adnominal clauses, -오/-우 can occur when the head corresponds to an object or another nonsubject relation. Subject relatives have an ordinary formation without this addition. The head’s semantic contribution and the lower predicate’s valency are essential to the analysis. For example, a ‘person’ following a verb of seeing can be the seer or the person seen. An overt subject within the relative clause, or the presence of person-sensitive material in the adnominal ending, can assist in distinguishing the readings. With dependent nouns whose referent is an event, fact or content, the connection to a grammatical role may be less direct.

In the preface, “호ᇙ” *holq* `do-O-PROSP.ADN` ‘[which someone] is minded to do’ modifies 바 *pa* ‘matter, content’. The expression refers to what is desired. The people who have the desire are separately available in the context. Another prospective adnominal, “몯ᄒᆞᇙ” *mothʌlq* `be.unable-PROSP.ADN` ‘who cannot [do so]’, qualifies the people who are unable, who are the understood subjects of the inability. The two tokens both use the prospective adnominal and differ in the additional vowel associated with the construction of the first (text) 1459, § preface).

| Head relation | What the head supplies | Adnominal formation to check |
| --- | --- | --- |
| Subject | The person or thing carrying out the action or bearing the property | An ordinary subject-relative form, without the object-associated -오/-우. |
| Direct object | The entity acted on or the content of a verb of saying, knowing or desiring | An expanded form containing -오/-우 is characteristic where the construction licenses it. |
| Other nonsubject or dependent-noun head | A circumstance, content or fact construed through the relative clause | The actual head and the construction govern whether the expanded form occurs. |

A relative clause may itself report a first-person act. The subject of the act and the role of the head are separate facts. An analysis based only on the translation ‘I’ risks assigning the finite person reading to an adnominal form with a different distribution.

### Second-person interrogatives

A direct question with a second-person subject can use the special low-series ending -ㄴ다 *-nta*, with related forms containing other prefinal or prospective material. This series is especially important because its final 다 *ta* resembles the ordinary declarative ending. The construction must be identified from the whole form and the question context. A question addressed to someone can still have a third-person subject. The person of the addressee alone does not establish the condition for the second-person subject series.

| Subject and clause type | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Second-person subject question | 네 가ᄂᆞᆫ다 | *ney kanʌnta* | `2SG.NOM go-NPST.VOL-Q.2` | ‘Are you going?’ (constructed) |
| Second-person subject question with a questioned place | 네 어듸 가ᄂᆞᆫ다 | *ney ətɨy kanʌnta* | `2SG.NOM where go-NPST.VOL-Q.2` | ‘Where are you going?’ (constructed) |
| Ordinary verbal assertion | 가ᄂᆞ다 | *kanʌta* | `go-NPST.VOL-DECL` | ‘[someone] is going’ (constructed) |

The second-person ending can therefore occur in both a polar and a content question. The presence of an interrogative word and the surrounding discourse supply the distinction in those uses. The final form contributes the special subject-person distribution.

## Prefinal honorific markers and final speech levels

### Three relations expressed in the predicate

Honorification distinguishes the subject of the predication, a respected nonsubject participant and the addressee of an utterance. These roles can coincide in a particular sentence. A question asked of a respected person about that person’s action, for instance, can call for both subject and addressee honorification. A speaker reporting another person’s actions to a peer can apply subject honorification and use a familiar ending to the peer. The speech level of a quotation belongs to the participants in the quoted speech.

| Relation | Morphological locus | Typical exponent | Referent to identify |
| --- | --- | --- | --- |
| Subject honorification | Prefinal, after the stem or after the object-honorific extension | -시-/-샤- *-si-/-sya-* | The honored subject of that predicate, overt or inferred. |
| Object honorification | Prefinal, next to the stem in a sequence that also has subject honorification | The -ᄉᆞᆸ- *-sʌp-* series | A respected object, recipient, goal or comparable nonsubject participant licensed by the predicate. |
| Addressee honorification | In or close to the final ending | High -ᅌᅵ- *-ŋi-* in statements, high -ᅌᅵᆺ- *-ŋis-* in questions, and distinctive directive endings | The recipient of the utterance. |

### Subject honorification

The common forms are -시- and -샤-. The shape is determined by the ending that follows and by the resulting combination. -시- is the usual recognizable form before a consonant-initial continuation. -샤- occurs in combinations with vowel-initial material, including sequences in which suffix vowels are represented by a contracted form. The vowel-initial environment includes conventional formations with their own surface shape, so it is necessary to identify the complete ending.

| Continuation | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Consonant-initial declarative ending | 가시다 | *kasita* | `go-SHON-DECL` | ‘[the honored person] goes’ (constructed) |
| Verbal nonpast material followed by a final ending | 가시ᄂᆞ다 | *kasinʌta* | `go-SHON-NPST.VOL-DECL` | ‘[the honored person] is going’ (constructed) |
| A linked predicate with a contracted combination | 가샤 | *kasya* | `go-SHON-CVB` | ‘the honored person goes, and …’ (constructed) |

Subject honorification can occur in a dependent predicate when the subject referred to there is treated honorifically. Each predicate is interpreted with its own subject. In a sequence including a lexical predicate and an auxiliary, the choice of where honorific marking is expressed is part of the construction. Recovering the subject from a preceding clause may be necessary with either a simple or a complex predicate.

### Object honorification

The core forms of the object-honorific marker are -ᄉᆞᆸ-, -ᄌᆞᆸ- and -ᅀᆞᆸ-. The historical label **object honorification** includes respect directed to a recipient or goal in an appropriate predicate. Case alone does not establish whether this marking is due. The lexical relation between participants and the status attributed to the nonsubject matter.

| Stem boundary | Usual form before a consonant-initial continuation | Romanization | Form with ㅸ before vowel-initial material | Romanization |
| --- | --- | --- | --- | --- |
| After an obstruent in the ㄱ, ㅂ, ㅅ class | -ᄉᆞᆸ- | *-sʌp-* | -ᄉᆞᄫᅡ- and related vowel combinations | *-sʌβa-* |
| After a coronal in the ㄷ, ㅌ, ㅈ, ㅊ class | -ᄌᆞᆸ- | *-cʌp-* | -ᄌᆞᄫᅡ- and related vowel combinations | *-cʌβa-* |
| After a vowel or a sonorant such as ㄴ, ㅁ or ㄹ | -ᅀᆞᆸ- | *-zʌp-* | -ᅀᆞᄫᅡ- and related vowel combinations | *-zʌβa-* |

The vowel in the last two columns includes material of the following ending. Forms with ㅸ reflect the alternation at the boundary. The ordinary stem alternations have to be applied before inferring an analysis from the consonant immediately visible in a spelling. Forms belonging to lexical predicates with an established deferential meaning can also resemble inflected forms. An analysis of a productive suffix requires a recognizable base and the appropriate construction.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 받ᄌᆞᆸ다 | *patcʌpta* | `receive-OHON-DECL` | ‘receives [it] from the honored person’ (constructed) |
| 보ᅀᆞᆸ다 | *pozʌpta* | `see-OHON-DECL` | ‘sees the honored person’ (constructed) |
| 받ᄌᆞᆸ시다 | *patcʌpsita* | `receive-OHON-SHON-DECL` | ‘the honored subject receives [it] from the honored participant’ (constructed) |

In a form with both honorifics, the inner marker belongs to the nonsubject relation and the outer marker to the subject. With a deferential final ending added, the analysis must identify the addressee as a third role even if that person is also a participant in the action.

### Final speech levels

The conventional three-way description uses the representative directive forms ᄒᆞ라, ᄒᆞ야쎠 and ᄒᆞ쇼셔. These name ending series. Different sentence moods have different exponents within a series. A level associated with relatively familiar address can also occur in exposition or narration, where no addressee participates in the narrated event.

| Series | Representative directive | Romanization | Gloss | Translation and use |
| --- | --- | --- | --- | --- |
| ᄒᆞ라 series | ᄒᆞ라 | *hʌra* | `do-IMP.PLAIN` | ‘Do it’, with familiar or plain address (constructed) |
| ᄒᆞ야쎠 series | ᄒᆞ야쎠 | *hʌyassyə* | `do-IMP.INTERMEDIATE` | ‘Do it’, in the intermediate address series (constructed) |
| ᄒᆞ쇼셔 series | ᄒᆞ쇼셔 | *hʌsyosyə* | `do-IMP.DEFERENTIAL` | ‘Please do it’, in deferential address (constructed) |

In the high series, -ᅌᅵ- is recognizable in declarative -ᅌᅵ다 *-ŋita*. Interrogative forms have -ᅌᅵᆺ- before the polar or content ending. The directive has -쇼셔 *-syosyə*, a form belonging to that mood’s own high series. A label such as *deferential* is a relation between participants in the utterance. It gives no fixed English translation equivalent to *sir*, *please*, or a title. The English gloss *please* in the directive illustrates one possible translation of the address relation.

## Declarative, interrogative and directive endings

### Declaratives and assertive formations

A declarative can close a statement at the end of an utterance or inside a quotation. -다 *-ta* occurs in the plain series. Forms in -라 *-ra* occur in characteristic complexes, including copular, retrospective and prospective formations. The final spelling 라 thus needs to be interpreted with the immediately preceding material. -니라 *-nira* is a common assertive ending in exposition or narration. It can present a statement as an explanation or as a fact relevant to the argument of the passage.

| Formation | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Plain -다 | 가다 | *kata* | `go-DECL` | ‘[someone] goes’ (constructed) |
| Retrospective -더라 | 가더라 | *katəra* | `go-RETRO-DECL` | ‘I recall that [someone] was going’ (constructed) |
| Prospective -리라 | 가리라 | *karira* | `go-PROSP-DECL` | ‘[someone] will go’ (constructed) |
| Assertive -니라 on a descriptive stem | 하니라 | *hanira* | `be.many-ASSERT` | ‘there are many’ (text) 1459, § preface) |
| Assertive -니라 after the copula | ᄯᆞᄅᆞ미니라 | *stʌrʌminira* | `only.matter-COP-ASSERT` | ‘that is all’ (text) 1459, § preface) |
| High declarative -ᅌᅵ다 | -ᅌᅵ다 | *-ŋita* | `ADDR.DEFERENTIAL-DECL` | Deferential assertion, translated with the predicate and context. |

The plain declarative form can be a useful way of citing a stem. In a text, its interpretation depends on the clause in which it occurs. A form in -니라 can have explanatory force through the way it completes a larger argument, even when the clause itself simply presents a state. In the preface, the predicate ‘be many’ asserts the frequency of the problem just described, and the final copular assertion delimits the narrator’s purpose (text) 1459, § preface).

### Polar and content questions

A polar question asks whether a proposition holds. A content question asks for the value of an expression such as a person, object, place, time, cause or manner. Middle Korean distinguishes these types in several ending series. The polar series characteristically has *a* or *ə* where the corresponding content series has *o*. There are different morphological hosts for the opposition. A nominal question can have a question marker after its nominal predicate, and a verbal question can use a verbal final complex. The special second-person subject series can serve both question types.

| Environment | Polar formation | Content formation | Conditions |
| --- | --- | --- | --- |
| Plain verbal final formations | -녀 *-nyə*, -려 *-ryə* | -뇨 *-nyo*, -료 *-ryo* | The preceding inflection determines which formation is available. The ㄹ forms belong to prospective constructions. |
| Nominal question and forms ending in 가/고 | -가 *-ka* | -고 *-ko* | Identify the nominal or other host and the scope of the question. |
| Deferential interrogatives | -ᅌᅵᆺ가 *-ŋiska* | -ᅌᅵᆺ고 *-ŋisko* | The high addressee marker precedes the interrogative distinction. |
| Special low series with second-person subject | -ㄴ다 *-nta* and related complexes | -ㄴ다 *-nta* and related complexes | The grammatical subject is second person. The interrogative expression or context distinguishes the readings. |

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가ᄂᆞ녀 | *kanʌnyə* | `go-NPST.VOL-Q.POLAR` | ‘Is [someone] going?’ (constructed) |
| 어듸 가ᄂᆞ뇨 | *ətɨy kanʌnyo* | `where go-NPST.VOL-Q.CONTENT` | ‘Where is [someone] going?’ (constructed) |
| 가려 | *karyə* | `go-PROSP.Q.POLAR` | ‘Will [someone] go?’ (constructed) |
| 어듸 가료 | *ətɨy karyo* | `where go-PROSP.Q.CONTENT` | ‘Where will [someone] go?’ (constructed) |
| 사ᄅᆞᆷ가 | *sarʌmka* | `person-Q.POLAR` | ‘Is it a person?’ (constructed) |
| 므스 것고 | *mɨsɨ kəsko* | `what thing-Q.CONTENT` | ‘What thing is it?’ (constructed) |

A questioned word does not always have the interpretation of an interrogative in isolation. Its interpretation can depend on an ending or a higher predicate that establishes the scope of a question. Conversely, an interrogative ending can carry the force of a question without punctuation in the witness. Direct quotation can preserve the force and the speech level of the quoted question even when the containing clause is a statement.

### Imperatives and hortatives

An imperative normally has an addressee understood as the one expected to act. An overt expression for that person can be a subject or a form of address. A hortative typically invites the addressee to join an action in which the speaker or the speaker’s group is included. A benefactive reading, an appeal to a superior and a strong command can all arise from the relationship of the participants and the interaction of the ending with the lexical predicate.

| Construction | Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- | --- |
| Plain imperative | 가라 | *kara* | `go-IMP.PLAIN` | ‘Go.’ (constructed) |
| Intermediate imperative | 가야쎠 | *kayassyə* | `go-IMP.INTERMEDIATE` | ‘Go’, with intermediate address (constructed) |
| High imperative | 가쇼셔 | *kasyosyə* | `go-IMP.DEFERENTIAL` | ‘Please go.’ (constructed) |
| Hortative | 가져 | *kacyə* | `go-HORT` | ‘Let us go.’ (constructed) |

Directive force is most straightforward with an action someone can control or undertake. With a property or circumstance outside the addressee’s direct control, directive morphology can support a wish or appeal, where the discourse permits it. A negative command uses its own prohibition construction. The person who is supposed to refrain is determined by the directive even when the lexical predicate is in a dependent form.

The appearance of 라 *ra* by itself does not identify a plain imperative. The whole ending complex distinguishes 가라 *kara* `go-IMP.PLAIN` ‘Go’ (constructed) from 가리라 *karira* `go-PROSP-DECL` ‘[someone] will go’ (constructed). A phrase naming a possible future action may be used to urge it in conversation while retaining prospective declarative morphology.

## Connective, adnominal and nominalizing forms

### Connective forms

A connective attaches to a predicate and relates its clause to another predication. The relation can be temporal, causal, concessive, sequential or one of comparison, purpose or manner. A given ending may permit more than one relation. The interpretation comes from the ending together with the two events and the order in which they are presented. The subject need not be expressed in either clause. Some combinations favor a shared subject, and an auxiliary construction can impose a tighter relation between the predicates than ordinary clause linkage.

| Ending | Romanization | Typical uses | Relevant restriction or ambiguity |
| --- | --- | --- | --- |
| -고 | *-ko* | Coordination or sequence | The predicates can share a subject or have distinct subjects in ordinary clause linkage. Interpret the connective separately from nominal or final interrogative -고. |
| -아/-어 | *-a/-ə* | Sequence, circumstance, manner and connection to certain auxiliaries | Stem vocalism and contractions govern the shape. The following predicate helps distinguish an auxiliary sequence from linkage of two events. |
| -니 | *-ni* | A circumstance or reason leading to the following clause | The propositions determine whether the relation is causal, temporal or explanatory. The complete ending -니라 belongs to the declarative series. |
| -며 | *-myə* | Concomitance or additive linkage | The relation may hold between simultaneous events or between descriptions made together. |
| -면 | *-myən* | Condition | Context determines whether the premise is hypothetical, habitual or taken as given. |
| -아도/-어도 | *-ato/-əto* | Concession | The following proposition holds despite the situation introduced by the first. |
| -(ᄋᆞ/으)ㄹᄊᆡ | *-(ʌ/ɨ)lssʌy* | Reason or grounds | The written ㄹ belongs to the connective formation. The matrix clause supplies the consequence or decision. |
| -고져 | *-kocyə* | Desire or purpose | A following predicate can carry the higher inflection. The relationship between the desirer and the subject of the desired action is constructional. |
| -디 | *-ti* | Complement of a following negative predicate in long negation | The negative predicate bears the clause inflection. The complement ending itself carries no negative force. |

Vowel quality in -아/-어 participates in vowel harmony. The pattern is most visible after stems whose final vowel and consonant allow the ending to remain separately identifiable. When a vowel stem and a vowel-initial ending meet, contraction or a stem alternation can obscure the ending. The written result has to be related to the rest of the paradigm.

The preface offers several independent illustrations of clause linkage. “달아” *tara* `differ-CVB` ‘being different’ links a descriptive predication to the following account of difficulty in communication. “이셔도” *isyəto* `exist-CONCESSIVE` ‘although there is’ introduces something people have to say despite their inability to express it. “아니ᄒᆞᆯᄊᆡ” *anihʌlssʌy* `NEG.do-REASON` ‘because they do not [accord]’ presents the grounds for the situation that follows (text) 1459, § preface).

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가고 오다 | *kako ota* | `go-CONN come-DECL` | ‘goes and comes [back]’ (constructed) |
| 가면 보다 | *kamyən pota* | `go-COND see-DECL` | ‘if [someone] goes, [someone] sees [it]’ (constructed) |
| 달아 | *tara* | `differ-CVB` | ‘being different’ (text) 1459, § preface) |
| 이셔도 | *isyəto* | `exist-CONCESSIVE` | ‘although there is’ (text) 1459, § preface) |
| 아니ᄒᆞᆯᄊᆡ | *anihʌlssʌy* | `NEG.do-REASON` | ‘because they do not [accord]’ (text) 1459, § preface) |

### Adnominal forms

An adnominal predicate modifies a following noun, including a dependent noun such as 바 *pa* ‘matter, content’. The modifying clause can identify a participant, characterize a thing or specify the content or occasion that the noun denotes. An adnominal ending marks how the predication is presented in relation to the head. In a participant relative, that head is construed as filling a role in the predication. In an expression of content, the relation can be supplied by the meaning of the dependent noun.

| Adnominal formation | With action or process verbs | With descriptive predicates or copular predication |
| --- | --- | --- |
| -(ᄋᆞ/으)ㄴ *-(ʌ/ɨ)n* | Typically anterior or completed with respect to the reference point. A resulting state can remain in force. | Typically a current or contextually given property or classification. |
| -ᄂᆞᆫ *-nʌn* | Ongoing, habitual or otherwise nonpast verbal predication. | The regular property construction takes the ㄴ series. Existential or process-like meanings have to be interpreted according to the predicate’s own distribution. |
| -(ᄋᆞ/으)ㄹ *-(ʌ/ɨ)l*, including spellings with ᇙ in relevant contexts | Prospective, intended, suitable or otherwise modal action. | A projected property or classification, where the construction allows a prospective or modal reading. |
| Retrospective formations with -더- *-tə-* | An action or process presented from an earlier viewpoint. | A property or state as remembered or observed at an earlier time. |

A vowel-final stem can take the written ㄴ or ㄹ directly. Consonant-final stems regularly have an intervening vowel in the corresponding series. Stem-final ㄹ has its own boundary behavior, so a single final ㄹ in an adnominal form does not establish how many successive consonants the morphological analysis posits. Prospective forms written with final ᇙ have to be read together with the environment and the following expression.

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 간 사ᄅᆞᆷ | *kan sarʌm* | `go-ANT.ADN person` | ‘a person who went’ (constructed) |
| 가ᄂᆞᆫ 사ᄅᆞᆷ | *kanʌn sarʌm* | `go-NPST.VOL.ADN person` | ‘a person who goes’ or ‘a person who is going’ (constructed) |
| 갈 사ᄅᆞᆷ | *kal sarʌm* | `go-PROSP.ADN person` | ‘a person who is to go’ (constructed) |
| 큰 집 | *kɨn cip* | `be.big-ADN house` | ‘a large house’ (constructed) |
| 먹은 것 | *məkɨn kəs* | `eat-ANT.ADN thing` | ‘what [someone] ate’ (constructed) |
| 먹을 것 | *məkɨl kəs* | `eat-PROSP.ADN thing` | ‘something to eat’ (constructed) |

The head noun ordinarily follows its relative clause. Case marking outside the relative clause applies to the whole nominal phrase. A subject suffix on a dependent noun does not make that noun the subject of its adnominal modifier. In the preface, “배” *pay* `matter-NOM` ‘the matter’ carries the subject marking for the existential predicate. The preceding “호ᇙ” *holq* `do-O-PROSP.ADN` ‘[which someone] is minded to do’ specifies the content of that matter. The lower and higher clauses assign different roles to the nominal expression (text) 1459, § preface).

### Nominalizing forms

A nominalizer allows a predication to occupy a nominal position while retaining verbal or adjectival structure inside. The nominalized expression can take case or another nominal particle. A subject, object, modifier or honorific inside the nominalization continues to be interpreted with the embedded predicate. This is the structure illustrated by the preface’s “ᄡᅮ메” *psume* `use-NMLZ-LOC` ‘in using [them]’, where locative -에 applies to the nominalized act of using (text) 1459, § preface).

| Nominalizing strategy | Form | Use | How to recognize it |
| --- | --- | --- | --- |
| ㅁ nominalization | -옴/-움 *-om/-um*, with contractions conditioned by the stem | An act, process, state, proposition or fact construed as a nominal expression | A nominal particle can follow the ending. The predicate can retain its arguments and modifiers. |
| 기 nominalization | -기 *-ki* | An act, undertaking or manner where the lexical construction accepts this nominal form | The surrounding construction licenses a nominal expression. The form can also be lexicalized. |
| Adnominal predication plus dependent noun | An adnominal ending followed by nouns such as 바 *pa* or 것 *kəs* | Content, a thing or another referent specified by the adnominal clause | The noun is present as a head and bears any particle assigned to the resulting phrase. |

| Form | Romanization | Gloss | Translation |
| --- | --- | --- | --- |
| 가기 | *kaki* | `go-NMLZ` | ‘going’ (constructed) |
| 먹움을 | *məkumɨl* | `eat-NMLZ-ACC` | ‘the eating’, as an object (constructed) |
| ᄡᅮ메 | *psume* | `use-NMLZ-LOC` | ‘in using [them]’ (text) 1459, § preface) |
| 먹을 것 | *məkɨl kəs* | `eat-PROSP.ADN thing` | ‘something to eat’ (constructed) |

The final consonant of the ㅁ nominalizer can become the written onset of the next syllable when a vowel-initial particle follows. To recover the nominalizer, take the particle and the stem into account. A lexical noun in ㅁ may have a settled meaning and distribution apart from its verbal origin. An occurrence is productively nominalized if the clause structure still licenses the interpretation of the predicate with its dependents. An occurrence can admit both a lexical and a compositional reading if the resulting expression has become a usual name for an action or result.

## Disputed analyses and contextual readings

### Segmentation and the status of ending complexes

An inflected word can often be represented at two useful levels. One records an ending as a surface unit with its distribution. The other identifies the internal sequence of markers when the paradigm supports it. The choice matters most when the surface vowel is fused or the written boundary runs through a syllable. Treating every written syllable as a single suffix would miss the structure of forms such as “ᄡᅮ메” *psume* `use-NMLZ-LOC` ‘in using [them]’. Treating every recurring vowel or consonant as an independently chosen suffix would assign extra categories to contracted endings (text) 1459, § preface).

| Form | Surface description | Finer analysis when supported | Practical consequence |
| --- | --- | --- | --- |
| -노니 *-noni* | A connective formation with verbal nonpast and speaker association | -ᄂᆞ- plus -오-, followed by -니 | The person contribution is interpretable even though -오- has no separate written syllable. |
| -니라 *-nira* | An assertive final ending | An internal relation to 니 and 라 can be considered in comparisons of the ending system | The occurrence is clause-final and assertive. Segmenting it does not give it the syntax of an independent connective clause. |
| -리라 *-rira* | A prospective declarative formation | -리- and declarative -라 | The prospective reading and the clause type can be represented separately. |
| -ᅌᅵᆺ가 / -ᅌᅵᆺ고 *-ŋiska / -ŋisko* | Deferential polar and content interrogatives | A common deferential interrogative base with a final question-type distinction | The social level and the kind of answer sought are separate properties. |
| -옴/-움 *-om/-um* | A nominalizing ending | The vowel and ㅁ can be considered separately in analyses that relate these forms to -오/-우 | Nominal syntax establishes nominalization independently of a claim about person. |
| -앳-/-엣- *-ays-/-əys-* | A contracted aspectual formation where attested | A connective vowel plus existential material in its historical source | A compositional result-state reading and a more grammaticalized reading require different contextual support. |

The preface’s “ᄆᆡᇰᄀᆞ노니” *mʌyŋkʌnoni* `make-NPST.VOL.PERS-CONN` ‘since I make [them]’ gives contextual support for a speaker-related analysis of -노-. The same witness’s nominalized expression in “ᄡᅮ메” belongs to an account of what other people are to do with the letters. These contexts establish distinct distributions for forms with a similar vowel. An analysis that connects the forms historically still needs to explain those distributions (text) 1459, § preface).

### The interpretation of -오/-우

Accounts of -오/-우 can give greater weight to subject person, speaker perspective, the semantic relation of the adnominal head or the construction in which the form appears. A finite first-person statement gives evidence for the person association. It provides little evidence for how an adnominal construction selects its form. An object-oriented relative or content expression gives evidence for an expanded adnominal form and can occur with a separate understood subject. A nominalizer with similar vocalism establishes neither person marking nor an object relation by itself.

The two preface forms “호ᇙ” *holq* `do-O-PROSP.ADN` ‘[which someone] is minded to do’ and “몯ᄒᆞᇙ” *mothʌlq* `be.unable-PROSP.ADN` ‘who cannot [do so]’ show why the head must remain part of the analysis. The first supplies the content of a wish and modifies a dependent noun. The second modifies the people in whom the inability is predicated. A finite first-person rule by itself would leave the contrast unexplained. The forms support distinguishing the closing function of the adnominal ending from the particular head relation in each phrase (text) 1459, § preface).

The gloss `O` preserves the presence of the vowel where the adnominal form has it. `PERS` singles out a person-sensitive finite or connective analysis where the subject and form both support it. These glosses state different observations about distribution. An explanation that derives them from a single grammatical element must still specify the contexts that produce each effect.

### Tense, aspect and viewpoint

The temporal interpretation of -ᄂᆞ-, -더- or -리- should be established relative to the clause’s temporal reference point. The reading of a quotation is relative to the situation of the quoted speech. The reading of a modifier is often relative to the containing clause or to the noun’s contextual reference. An earlier narrative time can contain a future projection. A retrospective description can concern a property that still holds. Those readings are available even when an English translator chooses a simple past or present tense.

Three tests help identify the contribution of a form in context.

| Question to ask of the passage | Distinction it helps establish |
| --- | --- |
| Does the form present an event itself or a state resulting from it? | Current event presentation, anterior action and resultative state are different contributions. |
| Is the speaker recalling the situation from an earlier vantage point? | A retrospective reading may involve evidence or viewpoint in addition to anterior time. |
| Is the situation asserted as actual, projected, intended or available for realization? | Prospective and modal readings can coexist with an earlier or later time reference. |

These tests also apply to auxiliary constructions. An existential in a resultative sequence can retain its state meaning. With greater grammaticalization the same historical material may convey anteriority over a wider range of predicates. The scope of that use has to be established for the dated text. A corresponding modern form alone does not settle the reading of a fifteenth-century token.

The preface provides an example where the English idiom could obscure the modal relation. “몯ᄒᆞᇙ” *mothʌlq* `be.unable-PROSP.ADN` ‘who cannot [do so]’ describes a group whose inability motivates the provision of the letters. Its prospective adnominal form characterizes the relevant possibility of expressing oneself. It can apply to the people currently under discussion (text) 1459, § preface).

### Dependence, quotation and address

An ending’s syntactic function and its pragmatic reading can point in different directions. A connective can carry a speaker-related prefinal marker. A quoted declarative or imperative can stand inside a larger dependent expression. A plain final ending in an account addressed to a general readership can be ordinary exposition. Honorifics inside that account can still mark a respected subject or nonsubject. These are properties of the individual predicates and their immediate discourse relations.

| Evidence in the passage | Grammatical decision it supports | Context needed |
| --- | --- | --- |
| Honorific before the tense or aspect complex | Identify subject or object honorification from the suffix’s position and shape. | Determine which participant is the respected subject or nonsubject. |
| Honorific material near a declarative or interrogative ending | Identify the final address level. | Determine who hears the utterance and whether the clause is quoted. |
| A form in -고 | Consider a connective or an interrogative according to the host and the clause’s role. | Check whether a proposition follows as its partner, or whether the clause asks for information. |
| A form in -라 | Identify the imperative or the declarative complex containing it. | Check the preceding markers and whether the utterance defines a directed action or asserts a proposition. |
| A nominal particle after an inflected predicate | Test for a nominalizer or a following nominal head. | Identify where the phrase closes and which higher predicate governs it. |

Where both segmentations or readings remain compatible with a passage, the attested form and the grammatical possibilities are more informative than a single translation. In the source preface, clause linkage, nominal syntax and the alternation between first-person intention and other people’s use of the letters constrain many such choices. The explanatory assertive forms close the propositions to which those dependent expressions contribute (text) 1459, § preface).

## How Mova handles predicate inflection

Mova has no level-2 parser model for Middle Korean (`okm`) and holds no annotated Middle Korean treebank. Consequently, there is no measured treebank baseline for the language. When a model is missing, `tools/mova-parse` reports that in the `# parse` comment. The tokens receive `_` in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL and `Parse=None` in MISC. Mova does not tokenize, tag, lemmatize or parse Middle Korean.

At level 1, `tools/langseg` assigns a language to each sentence and span within its supported distinctions. Hangul falls under `other`. At level 2, `tools/mova-parse` uses a model file for the language of the span when such a model exists. The `en` crate provides `en xpos-fill` and `en train` for a baseline trained on a language’s UD training split, and `en parse-eval` measures UAS and LAS on the development split. Models exist for English (`en.bin`) and French (`fr.bin`). The language package described in [What every language gets](https://md.mova.elamur.ai/free/en/docs/languages/PACKAGE/) asks for the same baseline for every language.

Squid reads each grammatical point in this chapter as a claim. Formalizable claims, such as suffix-order rules and closed inventories, wait in its queue for a place in Mova’s code. None of the chapter’s predicate-inflection rules is implemented. The first points to take up would be those needed for tokenization, tagging and a UD treebank baseline. The English tree readers `prag` and `world` have no parsed Middle Korean trees on which to apply frame, pragmatic or world analysis.

## Sources for this chapter

- Hunminjeongeum eonhae (Vernacular Translation of the Correct Sounds for the Instruction of the People). Prefatory text in Worin seokbo, vol. 1, Joseon royal edition, 1459. A vernacular rendering of the alphabet’s statement and letter descriptions, also a dated witness to Middle Korean prose. — public domain; not held in the library, to verify
- [← previous 14. Particles and other short grammatical forms](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/14-particles/)
- [next → 16. Stem alternations and conjugation paradigms](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/16-stem-alternations-and-paradigms/)
- [↑ back to Middle Korean (okm) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/ko/okm/)

> 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)
