On half of norskprøven the computer picks your level. On the other half you do — and it is binding

Norskprøven er delt inn i fire delprøver — a reading test, a listening test, a writing test and an oral test. Reading, listening and writing are taken on a computer, usually on the same day.

They are genuinely independent: Du bestemmer selv hvor mange delprøver du vil ta. Du kan ta alle fire delprøvene, eller du kan ta én, to eller tre. And they are separately repeatable — Du kan ta én eller flere delprøver på nytt senere hvis du vil forbedre resultatet ditt.

That modularity puts norskprøven with the Dutch NT2 and the Goethe modules above B1, and against every exam on this site that reports one verdict. What is unusual is that norskprøven pairs it with no aggregate at all and no pass mark. There is nothing to assemble the parts into.

There is also no syllabus, and HK-dir says so as a design commitment rather than an omission: Det er ikke pensum til norskprøven, and the reason given is that alle som er på samme nivå i norsk, skal ha samme sjanse for å gjøre det godt på prøven, uansett om man har gått på norskkurs eller ikke — everyone at the same level should have the same chance whether or not they took a course. An exam with no reading list is an exam you cannot cram for, which is the intent.

The reading test is adaptive, and HK-dir publishes the mechanism. Everyone starts identically — Alle som melder seg opp til leseprøven starter med den samme prøven, men alle får ikke de samme oppgavene.

It runs in three stages. Two short forprøver of 7–8 tasks each, the first pitched at A2. Det er resultatet ditt fra forprøvene som bestemmer hvilke oppgaver du får i neste del av prøven — your answers there decide the second pre-test and, after it, which of three main papers you receive: A1–A2, A2–B1 or B1–B2. The main paper is about 20 tasks. The whole thing runs up to 75 minutes, and you manage that time yourself.

The result is one of five, and the bottom one is not on the CEFR ladder at all: Under A1, A1, A2, B1, B2. A result below the scale's floor exists and is reported like any other.

HK-dir gives one piece of tactical advice that is worth repeating because it is counter-intuitive: spend no more than 10–15 minutes on a pre-test. Answer even when unsure — Du får 0 poeng på en oppgave hvis du ikke svarer eller svarer feil, so a guess is free. And be aware the routing can misfire in a specific direction: Hvis du bruker veldig lang tid på forprøvene, kan du komme videre til en prøve som egentlig er for vanskelig for deg. Grinding the pre-test can promote you into a paper above your level. HK-dir says you will probably still get the right result, but you will spend the main paper on tasks that cannot show what you can do.

Here the exam stops adapting. Du må velge hvilket nivå du vil ta prøven på når du melder deg opp til prøven — you pick A1–A2, A2–B1 or B1–B2 at registration, before anyone has measured anything.

The consequence is asymmetric, and HK-dir states both halves of it. Upward: På skriveprøven kan du ikke få et høyere resultat enn det nivået du har meldt deg opp til. For eksempel kan du ikke få B1 på A1-A2-prøven. Downward: Men du kan få et lavere nivå enn det nivået prøven måler på. For eksempel kan du få A1 på A2-B1-prøven.

So the band you book is a cap, not a target. Book too low and the exam will not award you the level you actually have — not because you failed to demonstrate it, but because the paper had no way to ask. Book too high and you lose nothing structurally, though HK-dir warns plainly against reaching for B1–B2 unprepared, since det kan være vanskelig å svare på oppgavene på en forståelig måte hvis du ikke er flink nok i norsk.

The papers differ by band. A1–A2: three tasks — a message, a picture description of about 50–80 words, and about 80 words or more on a familiar topic, in 90 minutes. A2–B1: three tasks — a picture description of 80–100 words, 80–200 words on a familiar topic, and an email justifying an opinion, in 90 minutes. B1–B2: two tasks — an email justifying an opinion, and an argued position of about 250–350 words chosen from two prompts, in two hours. In every band, Du må svare på alle oppgavene i prøven for å få et resultat: an unanswered task is not a lost mark, it is no result at all.

This is ours, and it is the reason this page exists. Put the two halves side by side and norskprøven is running two incompatible theories of how to find a candidate's level, in one sitting, on the same morning.

Reading and listening say: we will not ask you what level you are, because you do not reliably know. Two pre-tests, an automatic routing decision, five possible results including one below A1. Nothing you believe about yourself enters the process.

Writing and speaking say: tell us, in advance, and we will hold you to it. And HK-dir knows the weight that puts on the guess, which is why the same page hands you a self-assessment form to take to your teacher, points you at a level-chooser tool, and tells you to look at sample answers first. Those aids exist because the exam has made an untrained candidate's self-estimate load-bearing.

The asymmetry is not arbitrary — it is what the format costs. Multiple-choice comprehension can be routed live by a computer; a 350-word argued text cannot, because a human marks it days later against band criteria fixed before you sat down. So the adaptive half is adaptive because it is machine-scored, and the self-selected half self-selects because it is human-scored. The design is honest about its constraint.

But the cost lands entirely on the candidate, and it lands unevenly. The person most likely to under-book is the person with the least access to a teacher who could tell them not to — the self-taught learner, the one HK-dir explicitly designed the no-syllabus rule to protect. On the reading paper that learner is fine, because the machine finds their level regardless. On the writing paper they can be capped at A2 while writing like a B1, and nothing in the result will say so. That is the single most consequential decision on norskprøven, it is made weeks before the exam, and it is made in a registration form.

The split is stated. On reading and listening, Prøvene blir vurdert automatisk av datasystemet — no human marker. On writing and speaking, det er sensorer som vurderer prøven din og setter resultatet ditt, and those sensorer are experienced teachers of Norwegian as a second language. Your written text goes to minst to sensorer through a computer system. The oral test has a sensor listening in the room plus an eksaminator who asks the questions.

All four results are released together — Resultatene for alle delprøvene sendes ut samtidig — after some weeks; HK-dir's own estimate elsewhere is about three. Results arrive by email and the prøvebevis follows in the post.

Now the part worth planning around: Du har rett til å klage på resultatet på skriveprøven, and Det er ikke mulig å klage på resultatene fra de andre delprøvene. One of four sub-tests is appealable, and it is the writing test. The deadline is alltid 4 uker after the result is ready, and the outcome is final: Du kan ikke klage på nytt etter at du har fått resultatet fra klagebehandlingen.

An appeal is not a one-way bet. Når du klager, kan du få et dårligere resultat, det samme resultatet eller et bedre resultat — you can come out worse. Very few bodies state this as bluntly, and it changes the decision: appealing a B1 you are content with, hoping for B2, risks the B1.

This comparison is ours. Two of the four papers are marked by humans, and only one of those two can be appealed. So appealability here does not track was a person's judgement involved — it tracks whether the performance left a durable artefact a second reader can re-read. A written text can be sent to new sensorer. A conversation that happened once in a room, in front of one sensor and one eksaminator, cannot be re-heard, so there is nothing for an appeal to examine. Against the rest of this site that is a third distinct posture: France grants no appeal at all on the DELF, Finland grants one reassessment on the YKI and then declares it final, and Norway grants one appeal on exactly one skill out of four.

Norway publishes no national price, because it does not set one: Det er kommunen som bestemmer prisen — the municipality decides. HK-dir publishes only what is typical: En delprøve koster ofte mellom 400 og 1200 kroner, mens en hel prøve (4 delprøver) ofte koster mellom 1500 og 2500 kroner.

Some candidates pay nothing. If you have rett og plikt — the right and obligation to Norwegian training — you may take all the sub-tests free once. Beyond that first attempt, and for anyone without that right, you register yourself and you pay.

This comparison is ours. Two certificates documented on this site are free to a defined class of candidate by entitlement rather than by hardship discretion, and they draw the class completely differently. Norway exempts the people the exam was built for — those with rett og plikt — once. Ukraine exempts an entire nationality: its state language exam is free for citizens of Ukraine, and charged only to foreigners and stateless applicants, at three times the statutory subsistence minimum. One waives the fee for the candidates the policy is aimed at; the other charges only the candidates it is aimed at. The rest range from a fixed national fee (the NT2 at €50 a part) through legal ceilings (Poland, Celpe-Bras) to no published figure at all. Norway sits at the far end of that range: not a price, not even a ceiling, but a devolved decision plus a right that removes the price entirely for the people the exam was built for. It follows from what the exam is for — immigration and citizenship policy — and it means the fee question here is genuinely local. Ask your prøvested, and ask whether you qualify for the free attempt before you spend it.

HK-dir lists the prohibited conduct explicitly, including some that will surprise: no dictionary, no notes, no smart glasses or smart watches, no other program or website than the exam system, no memorised text passed off as your own — and pasting text into the writing test is automatically registered as cheating.

The consequences are proportionate but real: you may be removed from the room, du får ikke prøvebevis, du må vente ett år før du kan ta prøven på nytt, and you must pay for the next attempt. That last one matters most for the free-attempt group — it is how the entitlement is lost.

Detection is not limited to the exam room: Vi kan finne ut om du fusker under prøven, etter prøven eller når vi retter prøven. A decision is issued, and it can itself be appealed.

Norskprøven A1–B2 is one of three tests HK-dir runs for this population; the others are the samfunnskunnskapsprøve (social studies) and the statsborgerprøve (citizenship). They are separate exams with separate registrations, and passing one says nothing about the others.

Norskprøven's ceiling is B2 — A1 det laveste språknivået og B2 er det høyeste. If you need C1 Norwegian certified, this is not the exam and no amount of preparation makes it the right one.

The practical planning point is the one from the writing section, and it is worth ending on. Decide your band with evidence, not optimism and not modesty: use the self-assessment form, ask a teacher if you have one, and remember the papers are independent. You may book reading and listening with no decision at all, and sit only the writing test in the band you are confident of — then repeat that one sub-test later in a higher band if the result surprises you.