JLPT writing practice: the exam has no writing section, so why do it anyway?

The JLPT has no writing section and no speaking section. Every question on every level from N5 to N1 is multiple choice. So "JLPT writing practice" is not exam practice — it is diagnostic practice. Writing a sentence forces you to choose a grammar point with nothing to choose from, and that is the fastest way to find out which of the points you can recognise you cannot actually use.

I build BetterLingo from Japan, and this is the search that brings people to the wrong page more than any other. Most results for it are kanji stroke-order drills, which are handwriting practice, not writing practice. A few are past-paper multiple-choice tests, which are the opposite of writing. This page is about the third thing: producing your own Japanese sentences on purpose, and what that buys you on an exam that never asks you to.

What the JLPT actually tests

Check the official format on jlpt.jp before believing anyone, including me. Every level has three scored sections and all of them are selection tasks:

  • Language Knowledge — vocabulary and grammar, multiple choice.
  • Reading — multiple choice.
  • Listening — multiple choice.

There is no essay, no short answer, no interview. The 作文 (composition) sections you may remember belong to other exams — the EJU, the BJT, or a university entrance test — not the JLPT.

So the honest version of the answer is: if your only goal is the score on the certificate, writing practice is not the shortest path. Drilling grammar-point recognition and reading speed is. What follows is the case for doing it anyway, and it is a real case, not a sales one.

Why write anyway: recognition is not the same skill as production

A four-option question hands you the answer and asks you to notice it. Producing the same sentence from nothing asks you to retrieve it. These are measurably different tasks, and the gap between them is where most JLPT plateaus live: you feel like you "know" 200 N3 grammar points because you recognise all of them in a list, and then you sit a mock test and score in the 60s because recognition under time pressure is much weaker than recognition at rest.

Writing exposes that gap immediately. Three specific things it finds that multiple choice hides:

  • The point you always skip. Given a free sentence, most learners route around the grammar they are shaky on and use something safer. Notice what you never reach for — that is your real weak list, and no test will hand it to you.
  • Pairs you have never actually separated. ~たところ vs ~たばかり, に対して vs について, ように vs ために. In a multiple-choice question the distractors do half the work. Write a sentence and you have to know which one your meaning needs.
  • Collocation, which the JLPT tests indirectly. Grammatical but unnatural Japanese still costs you on the reading section, because reading fluently depends on the same instinct for what phrasing goes together.

How to practise writing for the JLPT, concretely

1. Write from the grammar list, not from your day

"Write a diary in Japanese" is the most common advice and the weakest, because you will write about the same day using the same twelve structures. Instead, take your level's grammar list in order and write one sentence per point, about anything. The constraint is the point, not the topic. Ten sentences is a session.

2. Make each sentence long enough to be wrong

「雨が降ったばかりです」 is too short to test anything. Add a clause and a time reference and you force yourself into tense, connective and particle decisions all at once — which is where the errors you need to see actually live.

3. Get every sentence marked

This is the step that decides whether the whole exercise is worth doing. Unmarked output practice reinforces your errors as confidently as it reinforces the correct sentences, because you cannot feel the difference from the inside. You need one of three things: a native speaker, a correction community like HiNative or a language exchange partner, or a tool that marks written answers. If none of the three is available today, do recognition drills instead — they are less valuable per hour and much better than practising a mistake.

4. Re-write the corrected sentence yourself

Reading a correction is not learning it. Write the fixed version out again from memory the next day. This is the same spacing logic your SRS deck uses, applied to production instead of recall.

What BetterLingo does here, and what it does not

I should disclose that I build BetterLingo, so treat this section as an interested party talking. It is a JLPT practice app whose open-answer questions mark Japanese you have written yourself, rather than only scoring multiple choice — which is the fourth step above, automated. You can see the marking without an account on the grammar check page.

What it is not: it is small and new, it has nothing like the mature SRS deck libraries that Anki or WaniKani have accumulated, and its marking is a model reading your sentence, not a Japanese teacher. On the specific error classes where meaning decides the answer — は vs が in a longer passage, register — a human reader is still better and it is not close. Use it for volume and same-minute feedback; use a person for the sentences that matter.

A three-week routine that fits around exam prep

  • Weekdays, 10 minutes: five sentences from the next five grammar points on your level's list, marked, corrections noted.
  • Weekends, 20 minutes: rewrite the week's corrected sentences from memory, then one longer paragraph using five points at once.
  • Every two weeks: a timed multiple-choice mock section, because that is the actual exam format and it needs its own practice.

The ratio matters. Writing is the diagnostic; the mock test is the exam. If you are inside a month of your test date, invert it — mocks first, writing only on the points the mocks show you failing.

Related reading

Next steps

Pick the next five grammar points on your level's list, write one sentence for each tonight, and get them marked before you write any more. Five marked sentences beat fifty unmarked ones, and the list you build of what you always avoid is the most useful study document you will own.