Ukraine's state language exam: you do not pick the level, the percentage does

For the people Ukrainian law obliges to know and use the state language at work, the Порядок sets out the bands as percentages of the maximum possible result. There is no level to register for. There is one exam, and this is what the number turns into:

0–29% — no certificate at all, and the Commission adopts a formal decision refusing to issue one. 30–39% — A1 (початковий рівень першого ступеня). 40–49% — A2. 50–59% — B1. 60–69% — B2. 70–89% — C1 (рівень вільного володіння першого ступеня). 90–100% — C2.

Read the widths and the exam's priorities are visible without anyone explaining them. Five of the six bands are exactly ten points wide. One is twenty: C1, from 70 to 89. C1 is also the level the law actually requires for appointment to the posts it covers — the Commission states that officials must hold C1 or C2 — so the band that matters most is the one the scale is least precise about. A candidate on 71% and a candidate on 88% receive the same document.

This comparison is ours. Almost every laddered exam documented on this site asks you to choose first and then measures you against that choice, and several of them punish the choice. Norway's norskprøven caps the writing and speaking result at the band you registered for — du kan få et lavere nivå, but never a higher one. Cambridge English fixes your ceiling when you book the paper, so B2 First cannot report above 190 on the Cambridge scale no matter how well you do. Ukraine's officials' exam removes the decision entirely. You cannot register for the wrong level, because there is no level to register for — and, correspondingly, you cannot aim.

One consequence is worth stating plainly for anyone preparing: on this paper there is no pass mark to hit, only a band to land in. The only threshold that behaves like a pass is 30%, below which nothing is issued.

The officials' exam is 60 minutes and three tasks: 30 test items in 30 minutes, one open written task (a text on a set topic) in 20 minutes, and one spoken monologue in 10 minutes. The Commission publishes the point totals: 47 points for the tests, 37 for the written text, 18 for the monologue — 102 in total.

It then publishes something that does not follow from those totals. The result is composed у такому співвідношенні: tests 40%, open written task 30%, oral part 30%.

The arithmetic is ours; the two sets of numbers are theirs. The monologue carries 18 of 102 points — under 18% of the marks — and 30% of the result. Divide the weight by the points available and one monologue point moves your final percentage by 1.67, one test point by 0.85, one written-task point by 0.81. A mark in the monologue is worth about two marks anywhere else on the paper.

Per minute the gap is wider still. The oral part is worth 3% of the final result for each of its ten minutes; the test section is worth 1.33% per minute. Ten minutes of speaking outweigh thirty minutes of multiple choice.

This is not hidden — both halves are published — but it is only visible if you divide, and the Commission does not do the division for you. It changes what preparation is rational: on a paper where the spoken task is a fifth of the marks and a third of the outcome, spending most of the preparation on test items is spending it in the cheapest currency on the page.

The Порядок states who marks what, and the pattern is consistent with the weighting rather than with convenience.

On the officials' exam, the test section is scored automatically. The open written text is marked by one examiner, selected at random by the exam system from the Commission's approved list. The monologue is marked independently by two examiners, also drawn at random, and the electronic protocol records the average of their two results.

So the only task on the paper that gets a second reader is the same task that carries double weight per mark. That comparison is ours, and it reads as deliberate design: the body spends its scarce second rater on the performance it has decided matters most, and on the one where a single marker's judgement would be least checkable.

The foreigners' exam does the opposite, and it is the more surprising half. It has five marked tasks — two written, three spoken — and five examiners, each randomly assigned exactly one of them. No task is marked twice. A candidate whose citizenship application depends on the result gets no second reader on any single performance; the redundancy is spread across tasks instead of stacked on one.

This comparison is ours too. Cambridge English rates its Speaking tests independently by two examiners and does not record them — two readers, no artefact. Russia's TORFL records the speaking subtest continuously and forbids editing the audio, but offers a resit rather than a re-reading — an artefact, no second reader. Ukraine gives the civil servant two readers on the spoken task and the prospective citizen none, on the same language, in the same building, in the same year.

The Commission also publishes its marking rate, which very few bodies do: each examiner assesses six oral tasks, or four written texts, per hour of work. Fifteen minutes per written text; ten minutes per spoken task. Russia's Pushkin Institute publishes the same underlying fact from the other end, as academic hours spent per candidate. Finland's OPH publishes it as money, pricing the YKI speaking and writing subtests roughly double the receptive ones. Three bodies, three units, one admission: marking a person speaking is a different job from marking a multiple-choice paper.

The written open task on the officials' exam is scored out of 37 across twelve criteria, and two of them are gates rather than components. Повнота розуміння аудіотексту і відтворення почутого на письмі — completeness of understanding the audio text and reproducing what you heard in writing — is worth 0 to 8. Вияв свого ставлення до теми висловлювання і наведення прикладів із власного досвіду — showing your own attitude to the topic and giving examples from your own experience — is worth 0 to 4.

Then: якщо за вказаними вище критеріями виставлено 0 балів, подальша перевірка роботи й виставлення балів не відбувається. Score zero on either and the marking stops. The remaining ten criteria — vocabulary, syntax, spelling, punctuation, structure, length — are never reached.

One of those two gates is not a language criterion at all. A text that is linguistically clean but expresses no opinion and offers no example from the writer's own life can be closed unmarked. The task is only 20 minutes long, and the instruction that protects you is not about grammar.

The spoken task carries an equally concrete rule, published as a step-by-step instruction for the candidate. Your face must be fully visible, you must not read from a text, and your eyes must be directed at the recording device for more than 90% of your total speaking time. The optimal length is given as two to four minutes within the ten. The system uses combined proctoring, and signs of a gross breach of the oral rules are grounds for examiners to mark the answer negatively — zero on every criterion.

This is ours: a published, numeric gaze threshold for a speaking test is not something we have found on any other examining body's pages. It is the kind of rule that a candidate who has prepared the content and not the format can fail without ever being told which one cost them.

The citizenship exam is written (Слухання listening, Читання reading, Письмо writing) plus oral (Говоріння speaking). The Порядок states that all four parts are рівноважні — equally weighted — each worth a maximum of 25% of the total.

The clock is not equal at all. Of the 150 minutes: listening 15, reading 60, writing 40 (15 minutes for a short message to a given prompt, 25 for a detailed text on a chosen topic), speaking 35 (10 minutes describing what you see, 10 answering questions, 15 for a monologue).

The arithmetic is ours. Four equal quarters over unequal minutes means a minute of listening is worth exactly four times a minute of reading — 25% earned in 15 minutes against 25% earned in 60. Listening is the shortest section on the paper and the densest place on it to lose a quarter of your result. Candidates who ration their preparation by how long a section lasts are rationing it backwards.

The pass is a single number: 70% to 100% and you receive the B1 certificate (series УМІ); 0% to 69% and the Commission adopts a decision refusing to issue one. There is no lower consolation band on this paper — unlike the officials' exam, where 30% still produces an A1 document.

Two practical constraints that catch people out. The exam cannot be sat outside Ukraine. The Commission's list of rooms for the citizenship exam names five, in three regions — Kyiv, Lviv oblast and Kharkiv oblast — down to the room number, each flagged (ДЛЯ НАБУТТЯ ГРОМАДЯНСТВА), and the page states in bold that sitting the exam online is not possible. It is a computer-based online exam that you must travel to a specific room in Ukraine to take. And sitting it does not commit you to anything: the Commission states that passing this exam does not oblige a person to file a citizenship application.

A B1-or-nothing paper is unusual, and how unusual is visible only side by side: Norway's citizenship test is oral-only B1, Spain's is A2, France's is B2. Ukraine's row and seventeen others are in the citizenship-level comparison.

For citizens of Ukraine the exam is free. The Commission states it plainly: для громадян України іспит на визначення рівня володіння державною мовою є безкоштовним.

For foreigners and stateless people the fee is not a price at all — it is a formula. Розмір оплати за складення іспиту становить три розміри прожиткового мінімуму для працездатних осіб, встановленого на 1 січня календарного року: three times the subsistence minimum for able-bodied persons as set on 1 January of that calendar year. The Commission then gives the figure that formula currently produces: from 1 January 2026, 9,984 ₴.

The observation is ours. Of the exams documented on this site, this is the only fee defined by reference to a social-policy index rather than set as a number, a band or a ceiling. It re-prices itself every January without anybody at the Commission deciding to re-price it, and it moves with a benchmark chosen for an entirely unrelated purpose. (The arithmetic implied by the published figure is a subsistence minimum of 3,328 ₴; the Commission publishes the multiplier and the product, not the base.)

The payment mechanics are specific and worth reading before you book: you apply first, then pay after the application and before the exam date, at a bank or through a banking app, then send a scan of the receipt to the Commission's dedicated address for foreign candidates — and keep the receipt until the certificate is in your hands.

Державний сертифікат діє безстроково. The state certificate is valid indefinitely — no expiry, no revalidation, in line with Poland's ważny bezterminowo, Spain's DELE, and Germany's TestDaF.

And then a sentence the others do not have. Державний сертифікат може бути анульований за рішенням суду, ухваленим за позовом Комісії. The certificate can be annulled — by a court, on a suit brought by the Commission, where breaches committed during the exam were substantial in character and led to a decision to issue that was wrong on the merits.

This is ours, and it is a distinction worth keeping. Every other certificate documented here either states a term, states that there is none, or declines to say. None of them publishes a route by which an issued certificate can be withdrawn. Ukraine states that there is no term and publishes the route. Indefinite validity turns out to describe the clock, not the grant — and the Commission cannot revoke on its own authority. It has to go and ask a judge.

There is a second document that proves the same thing without an exam at all. For several of the categories listed in Article 9 of the language law, a school-leaving certificate showing that Ukrainian was studied as a subject is accepted as proof of command of the state language — and the Commission adds that no requirement is set for the mark: Вимоги до оціночної шкали за підсумками вивчення української мови не встановлені. The grade you were given is not part of the test.

And a deferral rather than a waiver: people with distinguished service to Ukraine, including foreigners serving in the Armed Forces who hold a state decoration, may acquire citizenship without certifying their language level — but are then obliged to reach the level set by law within one year of acquiring it.

You may retake an unlimited number of times, but not more often than once every four months — to obtain a certificate for the first time or to raise a confirmed level.

The Порядок then enumerates who may re-register before the four months are up, and the list is unusually concrete: a breach of exam procedure that affected the objectivity of the result; illness on the day; a natural disaster, fire, flood, earthquake, military action or civil unrest, or a technological catastrophe on the day or the day before; theft of one of the documents needed to enter the exam room, occurring not earlier than five calendar days before; the death of a relative by blood or marriage, occurring not earlier than seven calendar days before; a sharp deterioration in health that stopped you finishing; and having registered for the wrong exam.

This is ours. A published, legislated grief window — seven days for a death in the family, five for a stolen passport — is not a thing we have found in any other examining body's rules. Most bodies write "documented exceptional circumstances" and leave the boundary to an administrator. Ukraine draws the boundary in the regulation, which makes it both fairer and blunter: on day eight it is not an exception any more.

Refusals work by fault, and the Commission sorts them into two annexes of the same published decision. Annex 1 — candidates refused because of technical problems during the exam: they get the right to retake immediately. Annex 2 — candidates refused for breaching the exam rules: not earlier than four months. The remedy in this system is not a re-mark and not a discount. It is time, and the amount of it depends on whose fault the failure was.

This comparison is ours, and it is a fourth position beside three already documented here. Norway allows an appeal on the writing test only, and warns that an appeal can lower your result. Cambridge cannot re-check a Speaking test because it does not record it. Russia records the speaking subtest, forbids editing the tape, and still answers a complaint with a resit. Ukraine answers it by waiving the wait.

You can also stop mid-exam: the candidate's cabinet carries a menu item, Відмовитися від іспитування, and a candidate may decline any part at any time. It is worth knowing what that button costs — a candidate who refuses is recorded as not having completed the procedure, and the tasks already done are not marked.

The exam itself is anonymous: each candidate works through the Commission's exam system as an анонімний комп'ютерний онлайн-іспит, and the script is encrypted. One instructor supervises no fewer than one and no more than five candidates in a session.

The result, though, arrives as a public act. Official results are fixed in a decision of the Commission and published on its own website, in the Documents section, no later than the day after the decision is adopted. Its first annex carries the electronic protocol data — the score for each part separately, and the overall result as a percentage. Further annexes list the people who hold each level. The refusal decision names those refused, split into the two annexes above.

This is ours: among the bodies documented on this site, publishing results as a named public decision rather than delivering them privately to the candidate is the exception, not the rule. It is consistent — this is a certificate that governs eligibility for public office, and the register of certificates is public too — but it is a materially different privacy posture from an exam board emailing a score report, and it is worth knowing before you sit.

One thing the pages we read do not settle. The Commission publishes the exact percentage bands for the officials' certificate and the exact 70% threshold for the citizenship one. On the pages read here it does not publish candidate volumes, pass rates, or the distribution of levels awarded — the kind of statistics Poland's certification system and Portugal's CAPLE both put on their own sites. That is a statement about what we found on 9 September 2026, not a claim that the figures do not exist.