Ukrainian Frequency · open data · v1.2.0

The Ukrainian lexical-minimum crosswalk

Every lemma in Ukraine's two published beginner vocabulary sources, joined and compared.

10,366 rows9 columnssnapshot 2026-09-13CC-BY-NC-SA-4.0

Why it exists

Ukrainian now has two independent, open answers to the question every beginner course must answer first, and they were built to answer different questions.

The minimumsThe profile
what“1000 i 1 slovo”, A1 and A2Ukrainian Vocabulary Profile (PULS)
whoBorodin and Turkevych; Borodin and LazarenkoSynchak, Starko, Burak, Svystun
kinda teaching syllabusa descriptive map
sizeabout 2,000 words10,194 lemma strings, A1 to C1
methodexpert selection over corpus frequency, weighted by usefulness at the learning stagecorpus-validated CEFR levels

Nobody had joined them. A syllabus and a descriptive profile are allowed to disagree, and where they do is the interesting part, so it seemed better to measure it than to guess.

What it found

One row per lemma, on a corpus frequency spine (UberText 2.0, top 12,000 lemmas).

ClassRowsMeaning
agree1,200both place it at the same level
earlier583the syllabus teaches it before the profile validates it
later174the syllabus teaches it after
absent776validated A1 (84) or A2 (692), in neither volume
unprofiled172in a volume, outside the profile's scope
beyond7,461validated B1 or higher and outside a 2,000-word syllabus

Two efforts built by different methods place 1,200 words at the same CEFR level. That is a convergence result: the beginner core of Ukrainian is a real object that two independent methods can find.

The sharpest disagreement

Four prepositions and a conjunction

The profile validates all five of these at A1. The A1 volume holds every one of them back to A2.

WordCorpus rankProfileVolume
до9A1A2
для21A1A2
від22A1A2
якщо87A1A2
без130A1A2

The A1 volume does carry по, про and через. So this is a line, not an omission.

A deliberate pedagogical decision, and a good one, now confirmed by the authors: when a word was assigned to A1 or A2, the grammar it needs was weighed with it, and a preposition whose case a beginner has not yet met (the genitive, the instrumental) was held back to A2 on purpose. That reasoning does not survive into a published word list, which is exactly why a crosswalk is worth having: it asked the question.

Two more things the volumes do not say: the selection follows the Official State Standard for Ukrainian as a Foreign Language, aligned with CEFR, under a strict cap of 1,000 words per level. Kseniia Borodin, personal correspondence, September 24, 2026.

What it cannot tell you

Three limits. No reading survives ignoring them.

  1. No part of speech on the minimum side. The published lists carry none, so the join is by lemma string and homographs collapse into one row.
  2. Onset, not sense. A profile level is the level of a lexeme’s earliest sense. A word can be validated A1 for a sense a syllabus does not teach.
  3. The three sources disagree about what one lemma is. те heads the absent list at rank 27 and is not a forgotten word: the frequency list and the profile both call it a lemma, while the minimums call it the neuter of той, which the A2 volume carries.

So the absent column is a list of candidates to check, never a list of omissions. Of the 84 validated-A1 words in neither volume, 40 have no relative anywhere in either volume; those are the ones worth looking at first: ситуація, кожний, момент, музей, ну, мільярд, продукт, герой.

A further 19 absent rows name the phrase that does reach the word, so такий reads against такий самий and жаль against На жаль.

How to use it

One row per lemma. Filter the class column to the question you have.

lemma, rank, pos, volume, entry, validatedLevel, class, relative, relativeVia

The entry column carries the published entry line verbatim, so every row can be traced back to the book it came from.

Download the crosswalk (CSV)How it was built →Profile a text against these minimums

Elsewhere in the instrument

Overview →
The list, the coverage curve and every thread's findings.
Vocabulary test →
120 yes-or-no words, guess-corrected: a rough count of the lemmas you know.
Text profiler →
Paste a text, see what share a learner at a level can read.
Crosswalk method →
How the crosswalk was built, step by step, with every figure derived.
Word lookup →
Any word, any form: stress, level, gloss, audio and definitions in one answer, or GET /api/uk/lookup.
Lemmatizer API →
POST /api/lemmatize: headwords matched to this list's convention.

Back to the overview · All research

Need something like this built?

I build the AI tools and automations a team adopts, and the full-stack apps and data pipelines under them, and ship them to production. Tell me the problem in a sentence and I'll give you an honest read on fit within a day.

Work with me →