{{first_name | みんな}}さん、こんにちは ☕

It's mid-May. You opened Bunpro this morning and there were 240 grammar points still grey. You opened your SRS and the review count had a 4 in front of it. You opened the practice test you printed last weekend and reading passage 3 is still circled in red.

And your brain did the thing brains do six weeks out from a deadline: I'll just do double sessions this week.

We've watched that move backfire on every July cycle for as long as we've been doing this. So before you double anything, read this.

Here's the rule most JLPT guides won't tell you: the last 6 weeks before the test are not for learning Japanese. They're for finishing the test.

That's not a typo. Japanese-learning continues forever. The test is a specific, timed performance on July 5. Those are two different jobs, and the second one runs on different fuel.

The cram instinct says: I haven't covered everything. I need to add more. The research says the opposite. Decades of memory research — going back to Tulving's retrieval work in the 1960s and locked in by Roediger and Karpicke's 2006 testing-effect studies — keep landing on the same finding: retrieving what you already know strengthens memory more than reading new material does.

In JLPT terms? A timed mock section beats another grammar chapter. Drilling the 300 vocab words you sort-of-know beats adding 200 new ones you'll half-learn.

Most July test-takers are 30–50% through their target-level syllabus right now. That's the median, not the failure case. The plan doesn't require you to be finished. It requires you to be test-fit in the retrieval-and-pacing sense.

Those are different things. And the next six weeks should be aimed at the second one.

One rule per gate. Three gates, based on how far out you are.

6+ weeks out (right now, just barely): You can still add new high-frequency grammar and vocabulary — but at half the rate you were three months ago. The window is closing, not closed.

3–6 weeks out: Freeze new grammar. Vocabulary in only from mock-test errors and practice passages — nothing from a fresh textbook chapter. Aim for 70% retrieval / 30% input.

Under 3 weeks out: Zero new grammar. Zero new kanji. Zero new chapters. The only "new" content allowed is what you missed on this week's mock. 90% retrieval / 10% comfort input.

The weekly mock rule: one full, timed mock per week. Then a 90-minute post-mortem the next day. The post-mortem matters more than the score — we'll get to that.

The weak-section rule: if any mock section comes in more than 8 points below your target, you reallocate 50% of next week's time to that section. Maintenance on the rest.

That's the whole tree. Pin it. Screenshot it. We'll explain why it works.

Every major JLPT prep resource publishes a "what to study" list. Tofugu has one. Migaku has one. Coto, JP YoKoSo, So-Matome, 新完全（しんかんぜん）マスター. They're useful resources. We'd point friends at any of them six months out.

Six weeks out, though, they share a quiet flaw: they keep adding new material into the final fortnight. That's directionally what learners want to hear (more = better). It's directionally opposite to what cognitive psychology has documented for sixty years (consolidation > new exposure under time pressure).

Most advice says cram harder. We think that's wrong, and the science backs us up.

What the research actually shows

Testing > restudy. Karpicke and Blunt's 2011 Science study put retrieval practice head-to-head with elaborative concept-mapping — the most respected restudy method in the literature. Retrieval won, and the gap widened on delayed tests. Translation: a timed mock section produces more durable learning than re-reading the same grammar chapter for the same minutes.

Spaced > massed. Cepeda and colleagues' 2006 meta-analysis of 300+ experiments confirmed what most of us feel but ignore: six 25-minute sessions across a week beat one 2.5-hour Sunday block. Every time. Especially for verbal material.

Consolidation is real. Continuing to practice material past the point it feels easy ("I know this already") protects performance under stress. It's the cognitive equivalent of overlearning a piano piece before a recital — easy in your bedroom, easy under the lights. Skip this step and the easy-in-your-bedroom stuff disappears when the proctor starts the timer.

The Mock-Test Trap

If you don't have 90 uninterrupted minutes to review your errors by category, you've spent 2.5 hours running a test that taught you nothing. One mock. One post-mortem. Repeat for six weeks. That's the cadence.

Most learners score the mock and stop. The score is the cheapest information in the test. The expensive information — the part that changes next week — is in the categories of error.

Mark every wrong answer with one of three tags:

The fixes are different. Knowledge gaps go into your SRS as a tight cluster. Pacing errors get drilled with a timer, not more flashcards. Register misses get fixed through input, not more grammar.

If you've been reading us a while, you've seen this argument before in the SRS Triage edition — treat error patterns as categories, not as individual cards. Same logic applies here. Don't drill the 14 items you missed. Drill the 3 patterns underneath them.

This Sunday (May 24), do one full timed mock test at your target level. Monday morning, do the 90-minute post-mortem. Three artifacts on one page by Monday lunch.

(Reply with what stumped you most — we read every one.)

Starter: Sit one timed mock Sunday morning. Score it. Note section scores against the pass mark (for the three-section levels, that's 19 out of 60 in each section, plus your total against the level cutoff). That's it. Mock + score = done.

Standard: Add the post-mortem Monday. Categorize every wrong answer (knowledge / pacing / register). For each category with 3+ misses, write one sentence: "Next week, I will [specific drill]." Then — this is the hard part — pick one thing to drop from this week's schedule to make room. New post-mortem time has to come from somewhere.

Challenge: Standard + run the weak-section rule. If any section came in 8+ points below your target, reallocate 50% of next week's study time to that section. Write the new weekly schedule before Tuesday morning. Stick it on the fridge.

You'll know you did it when: by Monday lunch you have one written list of error categories, one drill plan per category, and one thing dropped to make room. Three things, one page.

Last week's answers (The Counter Decision Tree): The walk was self-diagnostic — no single key — but the most-used counters in reader replies were 本, 枚, 台, 冊, and 〜つ / 個 as the fallback. Chargers, throw pillows, and amorphous plastic objects triggered the most fallbacks. If half your 15 ended in 〜つ, you used the system correctly.

If you read this far, you already know more about how to spend the next six weeks than most July test-takers do. The plan isn't finish the book. The plan is finish the test. You have time.

For: Anyone sitting July who wants to feel the real test format before test day.

Use: Spend 20 minutes on the official JLPT sample tests at jlpt.jp. One section, timed, at your level. Free, official, exact format.

Why it helps: Mock tests from third-party books are close. The official samples are the actual format your July paper will use — answer-sheet layout, instruction wording, timing structure. One pass through these in the next week eliminates the "wait, what does this question want?" moment on test day.

If you're going to overlearn one thing in the next six weeks, make it the core vocabulary you've already met. Jitsuryoku: Kihon is our free Anki deck of high-frequency core words, sequenced for spaced retrieval. Built for the N5–N3 audience who needs consolidation, not new exposure. Stop adding new cards. Drill what's already familiar.

Then this Sunday: one mock. Monday: post-mortem. Reply with one error category that surprised you — we read every reply, and the patterns we see across readers shape the next issue.

これからも一緒に頑張りましょうね〜 💪