🗣️ 30. Pragmatics: an overview

Pragmatics is what a speaker means by saying something in a situation, beyond what the words and the tree say. Ukrainian does much of this work with means English lacks or uses less: a large set of particles (ж, адже, таки, хіба, невже, ну, от, -но), a living vocative case and a two-way choice of ти and Ви, diminutive and affectionate suffixes that soften a request or warm a word, aspect, which turns an imperative from a brusque command into an invitation, and a word order that is free in grammar and fixed in meaning. This chapter maps these devices, says what the open corpora show about them (UD 2.18 Ukrainian treebanks and the Leipzig news corpus, counted on 2026-11-06), introduces the chapters 31–38 that treat each part, and describes exactly what Mova reads of Ukrainian pragmatics today — which is little: its pragmatic labeller and inference modules read English trees only, and its Ukrainian fact extractor keeps particles as members without interpreting them.

Contents

1. The chapters of the pragmatics part
2. Layers of meaning in a Ukrainian utterance
3. Particles: the main pragmatic word class
4. Word order and the information of the clause
5. Address: the vocative, ти and Ви
6. Expressive morphology: diminutives and augmentatives
7. Aspect and mood as pragmatic tools
8. Evaluation, emotion and stance
9. Context: situation, register and the label of a text
10. Ukrainian devices and English counterparts
11. How Mova handles it
12. Sources

Related chapters: word order and information structure in 29. Ellipsis, word order and information structure; the vocative's forms in 11. Nouns; diminutive suffixes in 17. Derivation: suffixes; aspect and the imperative in 13. Verbs; register in 37. Register and variation; particles in dialogue in 38. Dialogue and discourse markers.

Conventions: Ukrainian examples are in italics, with the national romanization (Cabinet of Ministers resolution No. 55 of 2010) in parentheses on first use and an English gloss in quotes: Мені холодно (Meni kholodno) 'I am cold'. Sentences from UD Ukrainian-ParlaMint (CC BY-SA 4.0) are quoted with their sent_id; UD Ukrainian-IU (CC BY-NC-SA 4.0) and Brown-UK (CC BY-NC-SA 4.0) are used for counts only; all other examples are constructed or come from public-domain texts.

1. The chapters of the pragmatics part

2. Layers of meaning in a Ukrainian utterance

The layers are those of every language (Grice's said and implicated, Austin's act, Stalnaker's common ground; see the English 30. Pragmatics: an overview §2), but Ukrainian packs more of them into grammar and particles. Take one constructed exchange between two neighbours:

— Ви ж обіцяли зайти, Ганно Петрівно! (Vy zh obitsialy zaity, Hanno Petrivno!) 'You did promise to drop in, Hanna Petrivna!' — Та я ж хотіла, але дощик пішов. (Ta ya zh khotila, ale doshchyk pishov.) 'Well, I did mean to, but the rain (lit. "the little rain") started.'

None of these layers is visible in a UD tree beyond a discourse or advmod arc for ж, a vocative arc for Ганно, Case=Voc on the names and Aspect=Imp on хотіла. A Ukrainian reader recovers the rest from conventions this part of the manual describes.

3. Particles: the main pragmatic word class

3.1 What a particle is

Ukrainian grammars treat particles (частки) as a part of speech of their own: uninflected words that add a shade of meaning to a word or a clause — emphasis, doubt, insistence, limitation, a question, a wish — without naming a thing, a property or an action. UD has no part of speech "particle" in that sense; Ukrainian treebanks tag most of them PART and attach them as discourse (stance, emphasis) or advmod (тільки, навіть, ще), and the negative не as advmod (ParlaMint writes advmod:neg).

3.2 The inventory

Compound particles are frequent and partly idiomatic: все ж таки 'nevertheless', навряд чи 'hardly', хіба що 'except perhaps', хоч би 'at least', ще б пак 'and how'. Sytar, Shvedova and Kanishcheva (2026, sytar-2026-ukrainian-multiword-expressions-corpus-creation), annotating 5,993 multiword expressions in 12,078 sentences of interviews from the GRAC corpus with the PARSEME 2.0 scheme, found so many such multiword particles that they propose a new category, particle idioms (PartID), for units such as все ж таки, навряд чи, хіба що, хоч би, хоча б, which the existing scheme forced into "other".

3.3 What the treebanks show

Lemmas attached with discourse (or its subtypes), all splits, UD 2.18, counted on 2026-11-06 by an awk pass over the CoNLL-U files (column 8 = discourse*, column 3 lowercased); the top items:

The two treebanks separate written and spoken Ukrainian at once: ну is absent from the top of IU and the most frequent discourse word of the parliament, where it opens turns, softens commands and frames objections (chapter 38). The particle же (23 in ParlaMint) is the full form of ж; many ж tokens are tagged with other relations.

In the Leipzig news corpus of 2024 (20,000 documents, about 14.2 million words, CC BY; counted as lowercase whole words with grep -ow on 2026-11-06) the particles of reasoning dominate those of conversation: адже 5,066, таки 1,774, хіба 235, ну 225, невже 22. In Brown-UK (1.04 million words, mixed genres including fiction; counts only) ну (60), хіба (116) and невже (19) are per million words about four, seven and twelve times as frequent as in news — fiction quotes conversation.

3.4 What a particle does not change

A particle does not change the truth conditions of the proposition in most cases: Він таки прийшов and Він прийшов are true in the same situations; таки adds that someone expected otherwise. Exceptions are the focus particles (тільки, навіть, саме), whose contribution is partly truth-conditional (тільки excludes alternatives) and partly presupposed (навіть marks the least expected alternative; chapter 33), and хіба in rhetorical questions, which reverses the polarity of what is asserted (chapter 26).

4. Word order and the information of the clause

Case endings mark the roles, so order is free to mark information: the given goes first (theme), the new last (rheme), and the end of the clause carries the main stress in neutral speech. Книжку читає брат 'The book is being read by (my) brother' / 'It is my brother who is reading the book' and Брат читає книжку 'My brother is reading a book' have the same tree and different answers to an implicit question (Хто читає книжку? against Що робить брат?). English uses articles, the passive or a cleft for the same work. The full treatment is in 29. Ellipsis, word order and information structure. Pragmatically, three uses matter:

• Contrast by fronting: Каву я не п'ю, а чай — залюбки 'Coffee I don't drink, but tea — gladly'.
• Emotional inversion: the verb or the adjective first in exclamations and in speech, Гарна в тебе хата! 'What a fine house you have!'.
• Focus by particles instead of a cleft: Саме Петро розбив вікно 'It was Petro who broke the window' (chapter 33 §6).

5. Address: the vocative, ти and Ви

5.1 The vocative

Ukrainian keeps a distinct vocative for addressing someone: Іване! (Ivane!), мамо! (mamo!), друже! (druzhe!), пане професоре! (pane profesore!), Ганно Петрівно!. The 2019 orthography (ukrainskyi-pravopys-2019, §§ 74, 87) gives its endings and prescribes it in every word of a compound address: both name and patronymic (Оксано Іванівно, Ярославе Андрійовичу), both common noun and name (друже Грицю, пане Віталію), both title and noun (пане лейтенанте). Its forms are in 11. Nouns; its social uses in 32. Politeness and hedging.

The norm and usage differ, and the treebanks measure it. Heads of the vocative relation, all splits, UD 2.18, counted 2026-11-06:

In parliament the common nouns of address take the vocative almost always (колеги 'colleagues' in Шановні колеги is the most frequent addressee, 421 times; пане, друже), but a first name, usually with a patronymic, is in the nominative more often than in the vocative: Дякую, Тарас Іванович (ParlaMint-UA_2022-01-25-m0.u93.p1.lang1.s1) 'Thank you, Taras Ivanovych' against the normative Вибачте, Олександре Валентиновичу (NSDC_UA_28_Feb2014-383) 'Excuse me, Oleksandr Valentynovych'. Shvedova, Lukashevskyi and Rysin (2025, shvedova-2025-developing-universal-dependencies-treebank-ukrainian), who built the ParlaMint treebank, parsed the Rada's transcripts of 1990–2024 (88 million tokens) and extracted more than 128 thousand vocative contexts: the nominative-vocative variation is large and depends strongly on the lemma, and only the transcripts of the early 1990s and 1997–2001, apparently edited, show the vocative alone. Here usage is ahead of — or beside — the codified norm: the nominative of address with name and patronymic is a spoken feature shared with Russian, and the written norm does not admit it.

5.2 ти and Ви

The second person plural ви is also the polite singular. In letters and official texts addressed to one person it is written with a capital (Ви, Ваш; ukrainskyi-pravopys-2019 § 60.2: Повідомляємо Вам…, Вітаємо Вас…). In the news corpus mid-sentence capitalised Ви/Вам/Вас/Ваш… occur 1,700 times against 45,481 lowercase forms (Leipzig news 2024, a form preceded by a letter or comma and a space; 2026-11-06); in Brown-UK 62 against 968. The capital is a genre marker — of letters, interviews, addresses to readers — not of every polite ви. Chapter 32 gives the social rules, including the traditional ви to parents.

6. Expressive morphology: diminutives and augmentatives

Simovych (1919, simovych-1919-hramatyka-ukrainskoi-movy-dlia, § 112) calls diminutive and affectionate words "a very characteristic phenomenon of the Ukrainian language": almost every noun has them (голова → голівка, голівонька, голівочка; серце → серденько, сердечко), almost every adjective (малий → маленький, малесенький), and even pronouns, numerals, adverbs and verbs (сам → саменький, самісінький; тут → тутечки; спати → спатоньки in talk to children). The people use them, he writes, "very gladly and on every occasion": not only to children, lovers or the grieving, but in ordinary talk. Other languages, he notes, have diminutives too, but use them more rarely, in fewer parts of speech and with fewer suffixes.

Their pragmatic uses:

In ParlaMint the chair manages turns with diminutive time words: одну секундочку / одну хвилинку / одну хвилиночку 17 times against одну секунду / одну хвилину twice (all splits, 2026-11-06), e.g. Одну секундочку, колеги, будь ласка, не розходьтесь, бо ми зараз будемо голосувати (ParlaMint-UA_2022-01-25-m0.u197.p4.lang1.s1) 'One second (dim.), colleagues, please don't leave, because we are about to vote'. The diminutive here belongs to an official, public register and signals courtesy, not intimacy. Chapter 32 §6 treats it as politeness.

7. Aspect and mood as pragmatic tools

Aspect (perfective against imperfective, 13. Verbs) is grammatical, but in the imperative and the past it carries pragmatic meanings:

The negative imperative is overwhelmingly imperfective in the treebanks: of second-person imperatives with не (excluding будь and давай(те)), 35 are imperfective and 5 perfective (IU 13/3, ParlaMint 22/2; UD 2.18, 2026-11-06). The conditional with би/б softens requests and wishes (Я б випив кави 'I'd like a coffee'; Чи не могли б Ви… 'Could you perhaps…'). The first-person plural imperative (ходімо, зробімо) and давай(те) + future make proposals (chapter 31).

8. Evaluation, emotion and stance

Ukrainian evaluates by vocabulary (чудовий 'wonderful', жахливий 'awful'), by suffixes (diminutive warmth, augmentative -ище: вовчище 'a huge wolf', дурнище), by intensifiers (дуже, страшенно, аж), by interjections (ой, ох, ех, ого, фу), and by particles of stance (таки, хіба, нібито). Distance from the content is marked by нібито, начебто and мовляв (reported, doubted), closeness by адже and ж (shared). The particles of distance are frequent in news: нібито 3,221 and начебто 740 in the Leipzig news corpus (2026-11-06), used to report claims not verified by the paper.

Emotion research on Ukrainian is recent. Dementieva, Babakov and Fraser (2025, dementieva-2025-emobench-ua-benchmark-dataset-emotion) built EmoBench-UA, 4,949 Ukrainian tweets labelled by crowd workers for joy, anger, fear, disgust, surprise, sadness or none (multi-label; Krippendorff's α 0.85 after filtering). Encoders trained on Ukrainian beat multilingual ones of the same size, the best encoder was a large multilingual sentence-embedding model, and anger and disgust were the hardest classes; models trained on the native Ukrainian data consistently beat the same models trained on emotion data translated from English (the translated data did better only on anger in some cases). Emotion in these texts often rides on the devices above: their example of anger is a sarcastic question with з мене … іздіваєшся?! — a surzhyk verb (standard знущаєшся) and the doubled punctuation of speech.

Stance in dialogue has two sides, which Korotenko and Kyslyi (2026, korotenko-2026-two-axis-framework-analyzing-ukrainian) separate for Ukrainian online discussions (Telegram channels and the Ukrainska Pravda forum): relational conduct (respect, engagement, handling of conflict) and substantive contribution (relevance, evidence, reasoning). Collapsing their rubric into these two axes raised inter-annotator agreement, and an ordinal model of each axis fitted expert scores better than a binary "constructive" label; a comment can be rude and substantive, or polite and empty. Politeness devices (chapter 32) belong to the first axis.

9. Context: situation, register and the label of a text

The same words do different things in a village, in parliament and in a song. This manual labels a text by three slots (chapter 37): language (uk), a dialect modifier (standard, a regional variety, surzhyk, diaspora), and a style of speech (official, neutral literary, colloquial, slang, song). Pragmatic conventions vary along all three. The deltas over the literary standard that matter for this part:

Each line is a delta: the grammar of the standard holds, and the line lists only what is added, replaced or used more.

10. Ukrainian devices and English counterparts

11. How Mova handles it

11.1 The pragmatic labeller and the inference modules: English only

Mova's sentence-level pragmatics is the prag crate. Its schema (prag/src/schema.rs) is a set of closed enums:

The values are language-neutral codes and would fit Ukrainian: Сідайте! is form=imp act=request, Хіба це важко? is form=q indirect=1. But every component that produces or reads these labels works on English:

• the features (prag/src/feats.rs) read trees of the English parser (en::annotate::Word) and a cue list, prag/data/cues.tsv, of 206 lines, all English (34 hedges such as maybe, perhaps, i think; evaluation words; please, thank, sorry);
• the frame reader (prag/src/frames.rs) maps English verbs of speech to acts (act_of: ask, beg, promise, thank …);
• implicatures and presuppositions (prag/src/infer.rs), irony (prag/src/irony.rs) and the expression lens (prag/src/lens.rs) use English word lists and the English Wiktionary's expression base;
• the politeness strategies of the dialogue layer (politeness in world/src/talk.rs) match lowercase English strings (thank, please, could you);
• the register lexicons in global/data/ (register-legal.tsv, register-tale.tsv, register-children.tsv) are measured on English books.

There is no Ukrainian parser model in the multilingual dispatcher (mova-parse) and the language segmenter langseg has a Latin-script table only (English and French), so a Cyrillic Ukrainian sentence reaches none of these modules. None of the measurements of prag (frame bench, labels in prag.tsv) concerns Ukrainian.

11.2 The Ukrainian fact extractor

The one component that reads Ukrainian is kg (kg — the Mova knowledge graph, kg/src/uk.rs): a conservative rule parser of simple sentences (up to 8 words, no commas, no conjunctions, no question mark) into a fact — predicate, subject, objects by case, adverbials, and particle members — using the VESUM dictionary (CC BY-NC-SA) for morphology. It records particles but gives them no meaning: ж, хіба, таки, хай become particle members; це before the predicate is taken as a particle; не sets a negation flag on the predicate. Its gates reject many pragmatically marked sentences outright: any comma (so every vocative and every parenthetical hedge), every question, every sentence containing чи, ніби, як, що or another word of its conjunction list.

Measured on 2026-11-06 (CPU, kg parse built from the worktree at commit ac9a23149, VESUM dict_corp_vis.txt.bz2, an empty name list):

11.3 Not there yet

• No Ukrainian pragmatic labels: no form, act, hedge or polarity for any Ukrainian sentence.
• No Ukrainian cue list: the particles of §3, the hedges of chapter 32 and the politeness formulas of chapter 31 are not in any Mova lexicon.
• No reading of the vocative, of ти/Ви, of diminutives as politeness or irony, or of aspect as a pragmatic signal.
• The kg gates exclude vocatives, questions, parenthetical hedges and complex sentences — most of the material of chapters 31–38.
• A language-independent meta-pragmatics, into which Ukrainian particles and English cues would map, is a design direction of the French and English work; nothing implements it for Ukrainian.

12. Sources

• Simovych, Vasyl. 1919. Граматика української мови для самонавчання та в допомогу шкільній науці. Kyiv–Leipzig. § 112 (diminutive and affectionate words), §§ 252–253 (imperative). simovych-1919-hramatyka-ukrainskoi-movy-dlia (public domain)
• Український правопис. 2019. § 60.2 (capital Ви), § 74 note 1 and § 87 notes 1–4 (vocative in compound addresses). ukrainskyi-pravopys-2019 (official text, not under copyright)
• Shvedova, M., Lukashevskyi, A., Rysin, A. 2025. Developing a Universal Dependencies Treebank for Ukrainian Parliamentary Speech. UNLP. shvedova-2025-developing-universal-dependencies-treebank-ukrainian (CC BY 4.0)
• Sytar, H., Shvedova, M., Kanishcheva, O. 2026. Ukrainian Multiword Expressions Corpus: Creation, Annotation, and Linguistic Analysis. MWE 2026. sytar-2026-ukrainian-multiword-expressions-corpus-creation (CC BY 4.0)
• Dementieva, D., Babakov, N., Fraser, A. 2025. EmoBench-UA: A Benchmark Dataset for Emotion Detection in Ukrainian. Findings of EMNLP. dementieva-2025-emobench-ua-benchmark-dataset-emotion (CC BY 4.0)
• Korotenko, A., Kyslyi, R. 2026. A Two-Axis Framework for Analyzing Ukrainian Dialogues. UNLP. korotenko-2026-two-axis-framework-analyzing-ukrainian (CC BY 4.0)
• UD Ukrainian-ParlaMint 2.18 (CC BY-SA 4.0): quoted by sent_id, counted. UD Ukrainian-IU 2.18 (CC BY-NC-SA 4.0): counts only. Counted 2026-11-06 with awk over all splits in ud-treebanks-v2.18.
• Leipzig Corpora Collection, Ukrainian news 2024 (Goldhahn, Eckart, Quasthoff 2012; CC BY): word counts, 2026-11-06. Brown-UK (Starko & Rysin; CC BY-NC-SA 4.0): counts only.
• Mova: prag/src/schema.rs, prag/src/feats.rs, prag/data/cues.tsv, prag/src/frames.rs, prag/src/infer.rs, prag/src/irony.rs, prag/src/lens.rs, world/src/talk.rs, kg/src/uk.rs, kg — the Mova knowledge graph, mova-parse.
• The English counterpart: 30. Pragmatics and discourse: overview.

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/uk/30-pragmatics-overview/