🗣️ 33. Implicature and presupposition

A Ukrainian speaker who says Деякі гості пішли 'Some guests left' lets the hearer infer that not all did; one who says Я відчиняв вікно 'I was opening / I opened the window (imperfective)' may let the hearer infer that the window is not open now; one who says Він знову запізнився 'He was late again' takes for granted that he was late before. The first two are implicatures — inferences the hearer draws from the choice of a weaker or less specific form, cancellable without contradiction; the third is a presupposition — background the utterance takes for granted and which survives negation and questions. Ukrainian has the scalar and conversational implicatures of every language, and two families of its own: those carried by aspect (completion, result, attempt) and those carried by particles (ж, таки, навіть, ще, вже), together with presuppositions marked by word order in a language without articles. This chapter describes them with examples, counts the triggers in UD 2.18 Ukrainian treebanks (2026-11-06), and says what Mova computes: its implicature and presupposition module reads English trees and English lemmas only; the Ukrainian fact extractor keeps the triggers as adverbials or particles without inference.

Contents

1. Implicature and presupposition in brief
2. Scalar implicatures
3. Numerals, quantifiers and agreement
4. Aspect: completion, attempt and annulled result
5. Object case: partitive genitive and the genitive of negation
6. Presupposition triggers
7. Focus, clefts and саме
8. Givenness without articles: word order and demonstratives
9. Conventional implicatures of particles and connectives
10. Relevance and manner in dialogue
11. Projection, cancellation and accommodation
12. How Mova handles it
13. Sources

Related chapters: aspect in 13. Verbs; negation and the genitive of negation in 27. Negation; word order and information structure in 29. Ellipsis, word order and information structure; particles in 30. Pragmatics: an overview §3; the English chapter 33. Implicature and presupposition for Grice, Horn scales and the classical triggers.

1. Implicature and presupposition in brief

2. Scalar implicatures

A word on a scale of informativeness implicates that the stronger words do not hold. Ukrainian scales (constructed examples; the parallel English scales are listed in SCALES of prag/src/infer.rs):

Conditions under which the implicature is not drawn, as in English: under negation (Не всі гості пішли → some did; не + strong item gives the inference that the weak one holds), when the stronger item is in the sentence (деякі, якщо не всі 'some, if not all'), when the speaker cancels it (а то й усі 'or even all', навіть усі, щонайменше 'at least'). Ukrainian has a lexical cancellation marker the English list lacks: а то й (Чекали годину, а то й дві 'We waited an hour, if not two').

Frequencies of the scale items in the treebanks (lemmas, all splits, 2026-11-06): весь 282 (IU) / 377 (ParlaMint), багато 109 / 67, деякий 40 / 32, більшість 26 / 32. Parliamentary speech uses весь and більшість (votes) more, багато less than written texts.

3. Numerals, quantifiers and agreement

A cardinal (три 'three') is read as "exactly three" by default and as "at least three" in contexts of thresholds (Потрібно три голоси 'Three votes are needed' — four will do). The approximative inversion of numeral and noun (годин зо дві 'about two hours', років п'ять 'five years or so', 22. Phrases) cancels the exact reading by grammar, where English uses about.

The agreement of a quantified subject carries a construal, not an implicature in Grice's sense but a choice with a pragmatic effect. Janda and Palii (2024, janda-2024-bahato-many) studied subjects of the form багато + genitive plural in a Ukrainian corpus: the verb is singular in 69.1% of attestations (Багато людей прийшло) and plural in 30.9% (Багато людей прийшли). Verb-subject order and inanimate nouns favour the singular, subject-verb order and animate nouns the plural; and the plural has been rising since about the 1950s and overtakes the singular in the current decade. They interpret the singular as construing the group as a homogeneous mass and the plural as a multiplicity of individuals. A hearer of Багато людей прийшли is invited to think of the people one by one. Here the codified norm, which allows both, is being reshaped by usage.

4. Aspect: completion, attempt and annulled result

4.1 The perfective asserts the result, the imperfective does not

The perfective past says that the action reached its limit: Я відчинив вікно (Ya vidchynyv vikno) 'I opened the window (it got open)'. The imperfective past names the action without asserting the result (chapter 13. Verbs). In a context where the result is at issue, the choice of the imperfective implicates — by quantity, since the speaker could have used the perfective — that the result was not reached or did not last:

The attempt reading is cancellable: Я відчиняв вікно, і нарешті відчинив 'I was opening the window and finally got it open' is not contradictory. That is the mark of an implicature, not of meaning. The annulled-result reading belongs to verbs of reversible motion and state change (відчиняти/зачиняти, вмикати/вимикати, приходити, брати): До тебе приходив Петро 'Petro came to see you' (and has gone again), against До тебе прийшов Петро 'Petro has come to see you' (he is here).

4.2 Aspect and the general-factual question

Ти вже обідав? 'Have you had lunch yet?' (imperfective) asks whether the event has taken place at all; the perfective Ти вже пообідав? asks whether a lunch in progress or expected is done. An imperfective answer Обідав 'I have (had lunch)' is complete; the perfective Пообідав adds that the meal was the expected one.

4.3 The perfective future and the imperfective present

The perfective present form is a future and makes a promise or prediction specific: Зроблю 'I'll do it (and finish)'. The imperfective future Буду робити / робитиму commits to the activity, not the result, and can implicate a lower commitment: Я буду це робити 'I'll be working on it' — not 'I'll get it done'.

5. Object case: partitive genitive and the genitive of negation

The case of a mass-noun object carries a quantity inference. Дівчина купила молока 'The girl bought (some) milk' (genitive) against Дівчина купила молоко 'The girl bought the milk' (accusative): the genitive implicates an indefinite quantity, the accusative a definite or whole one. Chaika, Lehka, Tamm and Vaiss (2024, chaika-2024-partitive-uktenten) tested the traditional claim that this partitive genitive goes with perfective verbs on the ukTenTen 2020 web corpus: for concrete mass nouns (вода, цукор, борщ) the Slavic pattern holds, but imperfective verbs also occur regularly with partitive genitives, and for abstract mass nouns (бадьорість 'vigour') aspect does not influence the object case at all.

Under negation the object may stand in the genitive (Я не бачу книжки 'I don't see a/the book') or the accusative (Я не бачу книжку): the genitive is the older and literary choice, the accusative the more frequent in current speech — objects (obj) of a head negated by не are genitive 264 and accusative 153 times in the written IU treebank, but genitive 144 and accusative 198 times in parliamentary ParlaMint (UD 2.18, all splits, 2026-11-06; the genitives include objects of verbs that govern the genitive anyway) — and the genitive tends to imply non-existence or indefiniteness (Тут немає книжки 'There is no book here'), the accusative a specific object (chapter 27. Negation).

6. Presupposition triggers

6.1 The inventory

6.2 Frequencies

Lemma counts in UD 2.18, all splits, 2026-11-06 (IU / ParlaMint): ще 240 / 220, вже + уже 208 / 194, також 139 / 116, навіть 145 / 37, тільки 136 / 134, лише 109 / 56, саме 64 / 74, теж 35 / 20, знову + знов 48 / 27; verbs знати 114 / 75, думати 57 / 106, вважати 33 / 75, почати + починати 97 / 36, перестати + переставати 17 / 2, продовжувати 8 / 22, забути 15 / 3, шкодувати 0 / 1. Parliament uses the non-factive verbs of opinion (думати, вважати) far more than the factive знати and more than the written treebank does — the language of opinion, not of established fact; written texts use навіть four times as often.

6.3 Factive and non-factive що

The complementiser що is the same after знати and after думати; the presupposition comes from the verb. Знати with an indirect question (знати, хто прийшов) presupposes that the question has an answer. Negation of the matrix leaves the factive presupposition intact: Він не знає, що вона приїхала 'He doesn't know that she has arrived' — she has; Він не думає, що вона приїхала 'He doesn't think she has arrived' — nothing presupposed, and by neg-raising read as 'he thinks she hasn't'. Забути, забувати is factive with що (Я забув, що сьогодні неділя — it is Sunday) and implicative with an infinitive (Я забув вимкнути світло — I didn't switch it off).

6.4 Ще and вже

The phase particles presuppose a prior state and assert its continuation or change:

Вже and ще with numbers carry the speaker's evaluation of the amount as large or small; English needs already / only.

7. Focus, clefts and саме

Ukrainian has no grammaticalised it-cleft; identifying focus is marked by the particle саме 'exactly, precisely', by a fronted це 'it is (this)', by word order and stress, and by the pseudo-cleft:

Саме is attached as discourse or advmod (54 discourse arcs in IU, 64 in ParlaMint; 2026-11-06); in many uses it adds precision rather than contrast (саме тому 'precisely because of that', саме зараз 'right now'). The це-construction is a copular sentence with a clause, analysed in UD with це as expl or nsubj according to the treebank; the presupposition is the same.

8. Givenness without articles: word order and demonstratives

English presupposes the existence and identifiability of a referent with the. Ukrainian, without articles, uses position and demonstratives: a noun phrase at the start of the clause is normally given (identifiable), at the end new.

The presupposition of identifiability, then, is carried by the theme position (chapter 29) and by цей/той; a translation into English must choose the or a from this evidence. The reverse also holds: a hearer who meets an unknown referent in theme position accommodates it as known (Сусід знову грає на трубі 'The neighbour is playing the trumpet again' — a specific neighbour is presupposed).

9. Conventional implicatures of particles and connectives

Some particles add a meaning that is not part of the truth conditions and not cancellable, Grice's conventional implicatures:

The pair аж три години / лише три години shows that the same truth (three hours) is presented as more or less than expected; in English as many as / only.

10. Relevance and manner in dialogue

Indirect answers are understood by relevance, as in English:

— Підеш з нами в кіно? 'Will you come to the cinema with us?' — У мене завтра іспит. 'I have an exam tomorrow.' → no.

— Як новий начальник? 'How's the new boss?' — Ну, костюм у нього гарний. 'Well, he has a nice suit.' → nothing good to say about him; ну signals the reluctance.

Manner (orderliness): coordinated perfective verbs are read in the order told: Він прийшов і сів 'He came and sat down'. Ukrainian imperfective coordination does not imply order (Він читав і писав 'He was reading and writing'); the "be orderly" inference attaches to perfectives, as in English it attaches to event verbs.

Quality: a question implicates that the speaker does not know; a rhetorical question with хіба or невже asserts the opposite of its form (chapter 26). Exclamations implicate a strong feeling.

11. Projection, cancellation and accommodation

Presuppositions project through negation, questions, modals and conditionals: Якщо він знову запізниться, я піду 'If he's late again, I'll leave' — he has been late before. They are filtered when the context entails them: Якщо він колись запізнювався, то він знову запізниться does not presuppose anything. They are cancelled (with effort) in metalinguistic negation: Він не «знову» запізнився — він раніше ніколи не запізнювався!. They are accommodated when new: Вибачте, я запізнився — моя машина зламалася 'Sorry I'm late, my car broke down' — the hearer accepts that there is a car.

The particle ж turns accommodation into a reproach: Ти ж обіцяв! 'You did promise!' presents the promise as known common ground and blames the hearer for ignoring it.

12. How Mova handles it

12.1 The inference module: English trees, English lemmas

prag infer computes scalar implicatures, presuppositions and two maxim inferences (prag/src/infer.rs). Everything it matches is English:

• SCALES: some < many < most < all, or < and, possible < likely < certain, may < must, sometimes < often < always, warm < hot, cool < cold, good < excellent, and others — English lemmas only;
• presupposition triggers: again, still, also, too, stop, quit, continue, keep, know, realize, regret, forget (with a clausal complement); a presupposition is "licensed" when the earlier 50 tokens contain the same predicate, otherwise reported as accommodated;
• Quality and Manner: questions with English wh-words and inverted English auxiliaries; and between event verbs, blocked by English stative verbs and reordering words (after, before, while …).

The functions take en::annotate::Word trees from the English parser. No Ukrainian sentence reaches them: there is no Ukrainian model in tools/mova-parse, and langseg segments only Latin-script languages. Aspect plays no part in the English module, which has no counterpart of §4.

12.2 The Ukrainian fact extractor

kg (kg/src/uk.rs) records aspect and tense on the predicate (from VESUM), and keeps the trigger words as members — знову, теж, ще, уже as adverbial, це as particle — without drawing any inference. Measured 2026-11-06 (CPU, kg parse from the worktree at commit ac9a23149, VESUM, empty name list):

12.3 Not there yet

• No Ukrainian scales, triggers or maxim rules; prag infer does not run on Ukrainian.
• Aspect is in kg's grammar but not used for completion, attempt or annulled-result inferences.
• Factive complements (знати, що; шкодувати, що) are rejected by the kg gates (comma, що).
• Саме is misread as the adjective сам; focus and clefts are not represented.
• Word-order givenness is not computed: a kg fact keeps the word positions but no theme/rheme or definiteness.
• Conventional implicatures of particles (таки, ж, навіть, аж, лише) are not computed; the particles are kept as members without meaning.

13. Sources

• Janda, L. A., Palii, Y. 2024. Understanding 'many' through the lens of Ukrainian багато. Russian Linguistics 48(1). janda-2024-bahato-many (CC BY 4.0)
• Chaika, L., Lehka, N., Tamm, A., Vaiss, N. 2024. Ukrainian Aspect and Object Case in ukTenTen: The Partitive Genitive of Perfective Verbs and Mass Nouns. In Partitive Constructions and Partitive Elements Within and Across Language Borders in Europe, Edizioni Ca' Foscari. chaika-2024-partitive-uktenten (CC BY 4.0)
• UD Ukrainian-ParlaMint 2.18 (CC BY-SA 4.0) and UD Ukrainian-IU 2.18 (CC BY-NC-SA 4.0; counts only): lemma counts with awk, all splits, 2026-11-06.
• Grice 1975, Horn 1972/1989, Karttunen 1973/1974, Stalnaker 1974 and the English treatment: as described in the English chapter, not re-read for this one.
• Mova: prag/src/infer.rs (SCALES, implicatures, presuppositions, maxims, STATIVE, REORDER), kg/src/uk.rs, kg — the Mova knowledge graph, mova-parse.
• The English counterpart: 33. Implicature and presupposition.

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/uk/33-implicature-and-presupposition/