All cases

2026-07 · Language Learning

An open frequency list for Ukrainian, built as data a learner can check

Six released threads over one spine: the top 12,000 lemmas ranked over 1.5 billion word tokens, a validated CEFR overlay, a derussification pass, paradigms and audio, a vocabulary-size test, and a text profiler. Every tier hashed, every figure re-derived in CI, one license wall enforced by a test.

Data pipelinesVerification and trustAmbiguous requirementsShipping under constraint

31

merges on the surface

12,000

lemmas in the published list

6,403

lemmas with a validated CEFR level

When

#2054#2200#2223#2258#2358#2359#2362#2366#2378#24112026-07-112026-09-22

Part of an arc of 2 merges on ukrainian-frequency, 2026-07-14. The whole arc, derived from the record

before

The join failure that produced the wrong figures left no screenshot: it was a number that looked plausible until the join counts were diffed against the source (#2358). Every tier was rebuilt on the corrected join.

during

The text profiler (thread 06): paste Ukrainian, measure it against a validated level, and nothing pasted leaves the browser (#2359).

2026-09-20 · Headless Chromium, 390 by 844, against production (www.jakelawrence.xyz)

during

The vocabulary-size test (thread 05): pseudowords catch guessing, and the page calls itself a rough proxy, not a certified score.

2026-09-20 · Headless Chromium, 390 by 844, against production (www.jakelawrence.xyz)

after

The hub today: a lemma-based, coverage-graded list at v0.1.0 under CC-BY, with the Ukrainian twin one tap away.

2026-09-20 · Headless Chromium, 390 by 844, against production (www.jakelawrence.xyz)

after

The four figures the page re-derives in the browser and CI re-derives on every build: 12,000 lemmas over 1.5 billion tokens, and the coverage tax Ukrainian's inflection imposes.

2026-09-20 · Headless Chromium, 390 by 844, against production (www.jakelawrence.xyz)

walkthrough

The coverage curve, the list itself, and the profiler set to measure a paste against B1 without sending it anywhere.

2026-10-01 · The studio recorder against production (www.jakelawrence.xyz): headless Chromium, 390 by 844 at 2x, the site's own bumpers and music, nothing staged · on the Case Studies channel

1. Source snapshotcommitted and citedderived aggregate countsUberText 2.0, lang-uk2. Pure builderrelease.mjsno network, no clockcurve re-derived live3. Frozen releaseCSV, JSON, Anki, hashedre-verified live + in CIvalidated levels onlyThe license wallPARSEME 2.0 UK is CC-BY-NC-SA: read one way, as validation of the multiword table,and walled out of the release by a fail-closed test that runs on every build.Predicted CEFR levels never ship as data; the validated overlay does.

Problem

The frequency lists a serious adult learner of Ukrainian needs are thin, often wordform-based, and sometimes quietly calqued from Russian, and the instruments that would measure progress (a vocabulary-size test, a text profiler, a lemmatizer) mostly did not exist as open, cited data.

Constraint

Every source carries its own license and its own headword convention. The only corpus-validated CEFR profile for Ukrainian could be redistributed with its authors' permission; a multiword resource could be read as validation but never shipped. Nothing pasted into the profiler may leave the browser.

Decision

A three-tier release: a committed, cited source snapshot, a pure builder, and frozen hashed files whose coverage curve re-derives live and in CI. Validated levels ship; predicted levels never do. The CC-BY-NC-SA resource is read one way as validation and walled out of the release by a fail-closed test.

Rejected

  • Predict CEFR levels with a model and ship them as data

    A validated level tracks frequency only weakly, so a predicted level would read as a proficiency level it is not. The Python package carries the validated levels and none of the predicted ones.

  • A wordform list

    Ukrainian inflects. The top 1,000 lemmas cover about 61 percent of running tokens, lower than the cross-Slavic figure usually quoted, and a wordform list would have hidden that tax.

What shipped

The list, the CEFR alignment, the derussification pass, audio and paradigms, the vocabulary-size test, the text profiler, a crosswalk of two lexical minimums, a 142-expression multiword table, a hosted lemmatizer with a round-trip as its deploy gate, an installable Python package, and a Zenodo DOI.

What I'd do differently

The join to the validated CEFR profile silently failed on lemmas with apostrophes, which Ukrainian has in ordinary words, because they arrived as HTML entities. Every tier had to be rebuilt on the corrected join (#2358), and the paradigm join had the same bug a day later (#2360). Decode at the boundary, then diff the join counts against the source.

Verify it yourself

If you ask me about this

How do you handle a license you cannot ship under?
Read it one way. PARSEME is CC-BY-NC-SA, so it validates the multiword table and never enters the release, and a fail-closed test checks that on every run.
Why host a lemmatizer instead of shipping one?
The dictionary it needs cannot leave the server. The endpoint normalizes, rate-limits, caches by content hash, retains no text, and a round-trip against the list is the deploy gate.
What did you do when a figure changed?
Rebuilt every tier on it and brought the note to the profile's authors up to the shipped model (#2355). The prose tier is drift-guarded like the data tier (#2411), so a stale number fails the build.

Evidence

  1. Ukrainian Frequency: a lemma-based open frequency list (research field data)

    #2054 · 2026-07-14 · feature · +1,987 −0, 30 files

  2. Ukrainian Frequency thread 02: CEFR alignment (honestly scoped)

    #2200 · 2026-07-22 · feature · +1,507 −27, 14 files

  3. ukrainian-frequency: validated CEFR overlay (thread 02) from PULS

    #2223 · 2026-07-23 · feature · +1,016 −53, 15 files

  4. Ukrainian Frequency: join the author-shared Borodin A1 lexical minimum (the closed seam)

    #2258 · 2026-07-27 · feature · +828 −18, 15 files

  5. Ukrainian frequency: decode the apostrophe entities in the PULS join and rebuild every tier on it

    #2358 · 2026-09-13 · feature · +88 −63, 41 files

  6. Ukrainian Frequency: the text profiler (thread 06)

    #2359 · 2026-09-13 · feature · +2,728 −2, 32 files

  7. Ukrainian Frequency: crosswalk the two lexical minimums against the validated CEFR profile

    #2362 · 2026-09-13 · feature · +3,200 −26, 39 files

  8. Ukrainian frequency: ukvocab package, Zenodo tooling, the hosted lemmatizer with its deploy path, and the upstream worklists (WO-1 to WO-5)

    #2366 · 2026-09-14 · feature · +6,119 −5, 87 files

  9. Ukrainian frequency: the multiword table (B1 extension, stage 3)

    #2378 · 2026-09-17 · feature · +1,495 −16, 22 files

  10. Ukrainian Frequency: make the prose tier drift-guarded like the data tier

    #2411 · 2026-09-19 · feature · +351 −19, 11 files

Receipts

Checked at build against the published corpus.

  • Every cited PR exists in the published corpus10 / 10
  • Pending PRs are named as pending, never passed off as mergedall merged
  • Cited PRs sit in the case's disciplinesLanguage Learning
  • Numbers with no source say so0 unmetered

www.jakelawrence.xyz/research/case-study-library/ukrainian-frequency

Cite: Lawrence, J. (2026). An open frequency list for Ukrainian, built as data a learner can check. jakelawrence.xyz case study library.

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 →