🗣️ 0. Introduction

This chapter says what Ukrainian is (§1), where it comes from (§2), what kind of language it is (§3), how its standard is made and how it moves (§4), how its variation is described in this manual (§5), where its vocabulary comes from (§6), the units and the notation of the manual (§§7–8), and what Mova does with Ukrainian today (§9).

1. The language in one paragraph

Ukrainian (українська мова, ukrajins'ka mova) is an East Slavic language, the state language of Ukraine (Constitution of 1996, art. 10; Law No. 2704-VIII of 25 April 2019 "On ensuring the functioning of the Ukrainian language as the state language"). It is written in a Cyrillic alphabet of 33 letters and the apostrophe (05), with a spelling close to the sound. It is a fusional, strongly inflecting language: nouns have three genders, two numbers and seven cases including a living vocative (11); verbs are organised by aspect, with an imperfective and a perfective partner for almost every meaning (13); stress is free and mobile and is not written (03); word order is free and serves information structure. The standard language rests on the speech of the Middle Dnieper (Kyiv and Poltava regions) as written by Kotliarevskyi and Shevchenko in the first half of the 19th century, with a large share of western (Galician) vocabulary and usage added in the second half of the 19th and the early 20th century (08). Its spelling is codified in Український правопис (Ukrajins'kyj pravopys, 'Ukrainian orthography'), whose edition in force was approved by the Cabinet of Ministers on 22 May 2019 (06).

The 2001 census of Ukraine, the last held, recorded Ukrainian as the native language of 67.5% of the population; the share of people who use it day to day has grown since 2014 and especially since 2022, but no census has measured it since, and this manual gives no figure for it.

2. Family and history in brief

The facts of the first three rows are in Simovych 1919 (introduction), Krymskyi 1907 and Vozniak 1911; the 19th-century bans and the history of the orthography are summarised in the preface of the 2019 orthography itself (Pravopys 2019, pp. 2–4). Ukrainian, Belarusian and Rusyn are the closest relatives of each other; Ukrainian shares much of its vocabulary with Polish and Belarusian through the centuries of the Commonwealth (19 §2), and its grammar with all East Slavic.

What sets Ukrainian apart from its neighbours inside East Slavic is mostly old and phonological (09 §2): Old East Slavic ě (ѣ) and o, e in newly closed syllables became і (хліб, кінь : gen. коня, піч : печі); the Old East Slavic i and y merged as и /ɪ/ while a new і arose; the fricative г [ɦ] took the place of g; final voiced consonants are not devoiced; the vocative survived. These sound changes are why one root alternates inside one paradigm (кіт 'cat', gen. кота), the most visible irregularity of the language (02 §5, 11).

3. Typological profile

The written language of the treebanks shows this profile in its distributions. In UD Ukrainian 2.18 (counts of 2026-11-06 over the train, dev and test files of both treebanks; the IU treebank is counted, never quoted):

The genitive is the most frequent case in both: it is the case of possession and of nouns depending on nouns, of quantities, of negation and of many prepositions (11). The vocative is rare in written prose (IU: fiction, news, legal texts, Wikipedia, letters, comments) and frequent in parliamentary speech (ParlaMint), where speakers address each other: шановні колеги 'dear colleagues', пане Голово 'Mr Chairman'. The share of a case is a fact about a genre, not about the language; the manual names the files for every count so that the reader can tell which.

4. The standard and how it moves

4.1 Institutions

Ukrainian has no academy that decides usage in the way the Académie française is sometimes thought to. Its norm is described by dictionaries and grammars and codified in three places:

• the orthography (Український правопис), prepared by the Ukrainian National Commission on Orthography (Українська національна комісія з питань правопису, 2015–2018: linguists of the institutes of the National Academy of Sciences and of universities from different regions), approved by the Presidium of the Academy and the Collegium of the Ministry of Education on 24 October 2018 and by the Cabinet of Ministers on 22 May 2019 (06);
• the academic dictionaries and grammars of the National Academy of Sciences: the Institute of the Ukrainian Language (Інститут української мови НАН України), the Potebnia Institute of Linguistics (Інститут мовознавства ім. О. О. Потебні) and the Ukrainian Language and Information Fund (Український мовно-інформаційний фонд), which publishes the 20-volume explanatory dictionary СУМ-20 and the orthographic dictionaries;
• the law: the 2019 Law on the state language sets where Ukrainian must be used, not how it is spelled or inflected.

The schools, the publishers and the editors of the media apply the norm; the open corpora and the large morphological dictionary VESUM (Rysin & Starko, dict_uk) record what is written (10 §2).

4.2 The character of the standard: conservative and modern at once

Two things are true of standard Ukrainian at the same time, and the manual keeps both in view.

It is very conservative. Forms that other European languages lost centuries ago, or keep only as archaisms, are ordinary everyday Ukrainian, and nobody who uses them feels old-fashioned or bookish:

• the vocative is a living case: every noun that can be addressed has it (мамо, друже, Тарасе, пане професоре), new words take it, and its absence in address is noticed (11, 09 §6.1);
• the synthetic future ходитиму, робитимеш, which goes back to an infinitive plus a form of иму 'I take', is as normal as буду ходити; in the Leipzig news sample of 2024 it is about two thirds of the imperfective futures (13, 09 §6);
• the old dual survives in the stress and form of nouns after два, три, чотири (дві руки́, not ру́ки; два брати́) (12);
• short adjectives (повен, винен, зелен in songs), the pluperfect (я був пішов), the first-person plural imperative ходім(о), and the uncontracted adjective forms of folk song (зеленая, молодая) are understood by everyone and used where they fit (15, 09);
• many words of the old written language are neutral today: гетьман, зрада, воля, пан, господар, оселя.

It is very modern. The standard follows usage, and quickly. The orthography itself says so. Its preface, written by the commission, holds that "the language of the people is made by its history: language changes, and the orthography must reflect above all its present state" (мову народу творить його історія: мова змінюється, і правопис має відображати насамперед її сучасний стан), that the language practice of the second half of the 20th and the early 21st century "has already become part of the Ukrainian orthographic tradition", and that "the co-authors of the Ukrainian orthography are also the wide circle of educated Ukrainians whose written practice makes this or that spelling customary" (співтворцями Українського правопису є також широке коло освічених українців, чия писемна практика узвичаює те чи те написання) (Pravopys 2019, preface, p. 4). The academic institutions work in the same spirit: the norm is described from usage and updated when usage has settled, and the dictionaries catch up within years, not over centuries as in French or Greek.

The measured results of the 2019 code show both halves at work. In the Leipzig news samples (1 million sentences each, CC BY; counts of 2026-11-06, details in 06 §§6, 9, 12):

So a variant that the code allows but writers do not want stays marginal, and a form that writers want is codified. This manual therefore describes the norm descriptively: it says what people say and write, how the 2019 orthography codified that usage, which variants it allows, and where usage is ahead of the codified norm (or behind it). Where it is, the chapter says so with a count.

4.3 What "standard" means here

The manual describes the literary standard (літературна мова) in its written and spoken forms, and alongside it the permitted colloquial usage of educated speakers: forms of Ukrainian origin that are not codified for writing but are not mistakes either (шо for що, нема beside немає, тре' for треба, ото, the dropped є). The two differ less than in many languages, and the colloquial forms often come from the dialects, so that the distance between "standard" and "colloquial" is mostly one of register (19 §7).

5. Variation: the standard and its deltas

The varieties of Ukrainian are described in this manual as deltas over the standard: for each variety, which grammar rules change, which words are added or replaced, and what shifts in pronunciation and spelling. A full parallel grammar is given only for the standard. The varieties are:

Ukraine has one standard and no second national centre: there is no "national variety" of Ukrainian in the sense that there are American and British English. Chapter 8 keeps the English manual's file name, national-varieties, for the place of these deltas.

6. Where the vocabulary comes from

The core of the vocabulary is inherited from Common Slavic and Old East Slavic. On it lie, in order of age: a learned Church Slavonic layer (благо, священний, страждання); Turkic words of the steppe (козак, отаман, чумак, кавун, майдан); a deep Polish layer, often of German, Latin or Czech origin, from the 14th–18th centuries (цікавий, мусити, рахунок, ґанок, шлюб, вада); German loans, many through Polish (дах, шухляда, ярмарок); a large Russian layer of the 18th–20th centuries, much of it in administration, science and technology; and an English layer that has grown since the 1990s and fastest since the 2010s (гаджет, донат, дедлайн, стрім). Ukrainian lexicography and editing have a long purist tradition that judges loans, especially Russian ones; the manual describes loans by measured usage, with counts from the open corpora, and says what the purist judgement is without taking it as the description (19).

7. The units of this description

The manual describes the language in the units of traditional Ukrainian grammar, and names each unit also in the terms of Universal Dependencies (UD) and of VESUM, because those are the two tag sets in the data Mova reads:

10 has the full mapping and the words that belong to several classes; the names of the cases are in 11 §4.

8. Notation and conventions

• Ukrainian is written in Cyrillic. On first use a word is followed by its transliteration in italics and a gloss in single quotes: хата (xata) 'house'. The transliteration is the scholarly one of Slavic linguistics (05 §9.3): h for г, g for ґ, y for и, i for і, ji for ї, j for й, je ju ja for є ю я, ž š č šč x c for ж ш ч щ х ц, and ' for the soft sign and the apostrophe. Names of authors and works in the source lists follow the same scheme or the spelling the work itself uses (Smal-Stockyj, as in his German grammar).
• Stress is marked with an acute accent where it matters (ру́ки : руки́). Ukrainian does not write it, and Mova strips it when it reads a word (norm in kg/src/morph.rs).
• Sounds are in the IPA, phonemes between slashes (/xɑtɑ/) and pronunciations in square brackets ([ˈxɑtɑ]). 01 and 02 give the inventory.
• Examples are constructed for this manual unless they carry a source. Examples from UD Ukrainian-ParlaMint (CC BY-SA 4.0) carry their sent_id (ParlaMint-UA_2003-10-14-m0.u130.p1.lang1.s2) and are quoted sparingly. UD Ukrainian-IU, Brown-UK and VESUM are under CC BY-NC-SA 4.0: the manual counts them and quotes none of their sentences. The Leipzig Corpora Collection samples (ukr_news_2024, ukr_newscrawl_2018, ukr_wikipedia_2021; CC BY, Goldhahn, Eckart & Quasthoff 2012) are counted; their sentences are not quoted.
• Counts were made on 2026-11-06 over the files named next to them. "IU" and "ParlaMint" mean the train, dev and test files of UD 2.18 together unless a chapter says otherwise.
• Sources are cited briefly in the text (Simovych 1919 §112, Pravopys 2019 §74, Hentschel 2023) and listed with licences and library paths in Sources: chapters 00–19; each chapter also ends with its own short list.
• "How Mova handles it" is the last section of every chapter and names real files. Where Mova does nothing, it says so.

9. How Mova handles Ukrainian today

Mova's Ukrainian is small, and each chapter says where it stops. The state on 2026-11-06:

1. The English parser on Ukrainian treebanks. Mova's dependency parser (en) can be trained on a Ukrainian treebank after en xpos-fill has mapped the UPOS and FEATS columns to the Penn tag set it knows (xpos_fill and xpos_of in en/src/main.rs). The Penn tags keep number and a few verb forms; case, gender, animacy and aspect are lost in the tag. Measured with the en binary of 2026-10-22 on UD 2.18, training on train and testing on dev, two runs with identical scores:

For comparison, the same parser gives LAS 87.00 on French GSD (What every language gets). The gap is what one expects of a tag set without case in a language whose syntax rests on case: the relations obj, iobj, obl and nmod are told apart in Ukrainian mostly by the case ending, and the tagger never sees it. On IU, 23.6% of the dev tokens are unknown to the tagger, against 12.9% on ParlaMint: an inflecting language multiplies word forms (11). The open neural parsers for Ukrainian (Stanza, UDPipe 2, COMBO, Trankit) are being installed and measured as witnesses under uk; their numbers are not yet in Open parsers: witnesses and tests.

2. The tokenizer. en tok-eval, learning its dictionary from the train file and evaluating on dev (2026-11-06, two identical runs): IU 673 sentences, 92.6% of them tokenized exactly as in the treebank, token F1 99.05; ParlaMint 739 sentences, 99.6% exact, F1 99.95. The apostrophe inside a word (в'їзд, м'ясо) is kept as part of the word, as UD Ukrainian requires; the misses are hyphenated words that the dictionary learned to split, initials and numbers such as 1-1 (07 §11).

3. The kg crate: Mova's own Ukrainian. kg holds the only Ukrainian-specific code in Mova (kg/src/uk.rs, kg/src/morph.rs, kg/src/gram.rs, kg/src/textgen.rs; kg — the Mova knowledge graph):

• a morphological analyser and synthesizer over VESUM (dict_corp_vis.txt.bz2): a word form gives its lemmas and grammars, a lemma and a grammar give the form back;
• a 4-byte grammar code (Gram) with part of speech, the seven cases (v_naz … v_kly), number, three genders, person, tense or form, aspect, animacy and degree, and flags for pronoun, proper name, negation and an implicit member;
• a tokenizer that keeps hyphens and the apostrophes ', ’, ʼ inside words, and a normaliser that lowercases, unifies the apostrophes and removes stress marks and invisible characters (tokenize, norm, invisible);
• a conservative parser from a simple sentence to a fact (predicate, subject, objects by case, adverbials with preposition and case, attributes): sentences of up to eight words without commas, conjunctions or digits; an ambiguity it cannot resolve is a rejection with a reason, never a guess; verb and preposition government comes from the VESUM rv_* tags;
• a deterministic generator from fact back to sentence (chunk order from a Markov chain over member classes, forms from VESUM synthesis) with a round-trip gate.

It is evaluated by hand: on the development sample kg/examples/gold-v1.tsv (30 Tatoeba sentences, verdicts of 2026-09-24) 19 parses are right, 2 half right and 9 wrong; kg/examples/heldout-v2.tsv holds a second sample graded before any fix based on it. VESUM is CC BY-NC-SA, so kg is a research and test path: what is built from it cannot go into a release.

4. Wiktionary. The raw Ukrainian Wiktionary dump of 2026-09-01 and the Ukrainian sections of the English Wiktionary are held (wiktionary); tools/wikt-count reads them in its uk mode (Data, libraries and licences). Wiktionary is CC BY-SA: for learning and tests only. Chapter 3 uses its stress marks for counts.

5. What Mova does not have. No Ukrainian in the en lexicon, lemmatiser or inflector; no phonology, stress model or grapheme-to-phoneme module; no transliteration in the public code; no spelling normaliser between the 1993 and 2019 norms; no labels for dialect, register or surzhyk (VESUM has tags arch, slang, bad, alt, up19, up92, and kg drops them when it reads the dictionary); no model of a variety as a delta over the standard; no Ukrainian pragmatics, frames or world. The chapters say, point by point, what the nearest piece of Mova does and what would be needed.

Sources for this chapter

• Український правопис (2019), approved by Cabinet of Ministers resolution No. 437 of 22 May 2019; preface pp. 2–5. Official document, not protected by copyright. Library books/u/uk/ukrainskyi-pravopys-2019 (uk-gram-pravopys-2019); archive.org/details/6032727-Ukr-Pravopys-2019
• Simovych, Vasyl'. 1919. Hramatyka ukrajins'koji movy dlja samonavčannja ta v dopomohu škil'nij nauci. Kyiv–Leipzig: Ukrajins'ka nakladnja. Public domain. Library books/s/si/simovych-1919-hramatyka-ukrainskoi-movy-dlia (uk-gram-simovych-1919).
• Smal-Stockyj, Stephan. 1913. Ruthenische Grammatik. Berlin–Leipzig: Göschen. Public domain. Library books/s/st/stockyj-1913-ruthenische-grammatik (uk-gram-smal-stockyj-1913); archive.org/details/ruthenischegramm00smaluoft
• Krymskyi, Ahatanhel. 1907–1908. Ukrainskaja grammatika dlja učenikov vysšich klassov gimnazij i seminarij Pridneprov'ja. Moscow. Public domain. Library books/k/kr/krymskyi-1907-ukraynskaia-hrammatyka-dlia-uchenykov (uk-gram-krymskyi-1907).
• Vozniak, Mykhailo. 1911. Hramatyka Lavrentija Zyzanija z 1596 r. Public domain. Library books/v/vo/vozniak-1911-hramatyka-lavrentiia-zyzaniia-z (uk-gram-vozniak-1911-zyzanii).
• Shvedova, Maria, Arsenii Lukashevskyi & Andriy Rysin. 2025. Developing a Universal Dependencies treebank for Ukrainian parliamentary speech. CC BY 4.0 (ACL Anthology) (shvedova-2025-developing-universal-dependencies-treebank-ukrainian).
• Goldhahn, Dirk, Thomas Eckart & Uwe Quasthoff. 2012. Building large monolingual dictionaries at the Leipzig Corpora Collection. LREC 2012. (The ukr_news_2018/2024 and ukr_wikipedia_2021 samples, CC BY.)
• de Marneffe, Marie-Catherine, Christopher Manning, Joakim Nivre & Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics 47(2). CC BY 4.0 (demarneffe-2021).
• Data: UD Ukrainian-IU 2.18 (CC BY-NC-SA 4.0, counts only) and UD Ukrainian-ParlaMint 2.18 (CC BY-SA 4.0), ud-treebanks-v2.18; VESUM / dict_uk 6.8.6 (CC BY-NC-SA 4.0, counts only); counts of 2026-11-06.
• Mova: en/src/main.rs (xpos_fill, xpos_of, tok_eval, parse_eval), kg/src/uk.rs, kg/src/morph.rs, kg/src/gram.rs, kg/src/textgen.rs, kg — the Mova knowledge graph, kg/examples/; What every language gets.

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/uk/00-introduction/