{{first_name | みんな}}さん、こんにちは ☕

It's Sunday afternoon. You just sat a full mock. The grammar section ate fifteen minutes more than you budgeted, you hit the reading section already behind, and the last eight questions were guesses you bubbled in while the proctor was packing up.

You walked out and told yourself the same thing everyone tells themselves: my reading isn't ready. So tomorrow you'll buy a new graded reader and read for ninety minutes, untimed, and feel productive.

We've watched that move backfire on every JLPT cycle since we started doing this. The diagnosis is wrong. So is the drill.

If you reread the passages after the test ended — at your own speed, in your own kitchen — you'd understand them cleanly. We've heard this from readers every cycle. "I understand the meaning of the whole paragraph, but I get confused while answering their questions." It's one of the most upvoted threads on the renshuu.org forum, and it gets reposted every July.

That sentence is the whole pathology. Comprehension is fine. The clock is the problem.

This matters because the two failures have different fixes. If your bottleneck is comprehension, you need more input and more grammar. If your bottleneck is pacing, you need a stopwatch and a phone timer.

And here's the part nobody tells you: pacing is the cheapest section of the test to fix in 31 days. It's also the one almost nobody trains, because the dominant cultural script in Japanese self-study is more reading, more immersion, more input. That advice is correct at six months out. At 31 days out, it's the wrong drug at the wrong dose.

We covered the macro plan for these final weeks in the 6-week JLPT final-push plan. This issue is the zoom-in on the section that breaks the most learners: reading.

Imagine the same 300-character Japanese passage on a screen in front of you. Two labels.

The passage didn't change. Your Japanese didn't change. The task changed. Same words, different clock, different skill. That's the whole insight.

Most readers train Version A all week and walk into Version B on test day. They are not the same drill.

Here's a 90-second diagnostic you can run before you do anything else this week.

Pick a 300-character Japanese passage at your level. NHK News Web works for N3–N2; NHK News Web Easy works for N5–N4. Time yourself reading it once, for comprehension. Then do the math: 300 ÷ seconds × 60 = your current characters per minute.

Now compare it to this table. We'll explain its pedigree right after.

Now: what to do with your number.

If you're at or above your target, your bottleneck isn't speed. It's comprehension or test-craft. Drop the speed drill. Go to passage-type budgeting (next section).

If you're below your target, your bottleneck is speed. Welcome — this is the most trainable variable on the entire test in 31 days. Read on.

The 10-minute daily drill is in the Renshuu Time block below. It's not complicated. It's just a stopwatch and a passage and the willingness to stop reading when the timer ends. Most learners can't bring themselves to do that last part — they keep going, "just to finish," and they train the wrong pace. The point of the drill is the cutoff, not the comprehension.

The chars-per-minute number tells you whether you have a speed problem. The passage-type budget tells you where to spend your minutes once you do.

The JLPT reading section isn't one game. It's four games stacked on top of each other, each with a different time-per-point yield. Treating them all the same is how learners burn 8 minutes on a 短文たんぶん short passage worth one point, then run out of clock on the 中文ちゅうぶん mid passages that carry the section.

Two notes. N3 reading does not include 統合理解 — that's an N2 / N1 task only. And 情報検索 is a scan, not a read. If you're reading every line of the info-retrieval block, you're playing the wrong game.

Remember the open loop from a few paragraphs back — the gap between mock pace and real pace? Here's the payoff.

Most mock-test pacing is optimistic by roughly 5–10% versus the real test. That's our estimate, not a published number — but the direction is consistent, and anyone who has sat a paper JLPT will tell you the same thing. Two reasons.

One: bubbling answers on paper takes longer than tapping or clicking. Across 30+ reading-section questions, the bubble-sheet transition tax adds up to a couple of minutes you didn't budget for.

Two: real-test anxiety arousal narrows your reading pace by a small but real margin. Your eyes move slightly slower under stress. So does your inner translation.

Build a 10% buffer into your mock-test budget. If your mock says you finished reading in 60 minutes, your real-test target for the same workload is 54.

One more thing, then we'll let you off the hook with a tiny win. The single biggest reason intermediate readers cap out around 200 chars/min is subvocalization — reading aloud in your head. It caps you at your speaking speed, because that's literally what it is.

At N3 you can survive this. At N2 you can't. You have to learn to chunk by phrase, not pronounce by word.

Drill: take a paragraph you already understand. Read it once normally. Then read it again, deliberately not hearing each word in your head — just grouping by phrase and letting your eye skim. It feels wrong at first. That feeling is the skill arriving. Discourse markers do most of the heavy lifting here — しかし、つまり、一方いっぽうで、とはいえ、たとえば、結論けつろんとして. Train your eye to lock onto those. Skip the rest.

The 10-Minute Stopwatch Drill — do this 5 of the next 7 days.

Pick a 300–500 character article at your level. NHK News Web is ideal for N3/N2 because its kanji density approximates JLPT prose. Open your phone stopwatch. Set the timer to your target — N3: 90 seconds for 300 characters. N2: 70 seconds for 300 characters. N1: 50 seconds.

Read silently, chunk mode (no inner voice). When the timer ends, stop reading. Then re-read untimed. Note one discourse marker or sentence connector you skimmed over that mattered. That's it.

(No literal answer key — this is a measurement drill. Pattern reveal drops in next week's brew.)

You'll know you did it when: you can read a 300-character passage at your level in the target time and answer one inference question about its main point correctly. Do that on 5 different articles and your pace is JLPT-ready.

Last week's answers (The 自動詞じどうし / 他動詞たどうし Decision Rule): No literal key — that was production practice. Reader replies clustered around three patterns. The 〜まる pairs landed first (始はじまる/始はじめる, 閉しまる/閉しめる) — Pattern 2 is the most visible. 付つく/付つける still required a deliberate pause for most readers — it's a lexical pair without a clean morphological shape, exactly what the fallback question is for. And the ている split clicked when sentences were written, not when they were re-read. If you only re-read your own examples, you trained recognition, not selection. Do the writing.

If you only do the chars-per-minute measurement once today and never run the drill, that is still more data than 95% of test-takers walk in with. You now have a number. Numbers fix what feelings can't.

For: N3–N2 readers who want one free, kanji-dense source to drill on.

Use: NHK News Web — open one article in the morning, time yourself once before your coffee is finished, log the seconds. Done.

Why it helps: Kanji density approximates JLPT reading prose, articles run 300–600 characters on average, and updates daily, so you'll never run out of fresh material. For N4/N5: NHK News Web Easy does the same job at a gentler level.

Run the measurement today. Just once. Reply with two numbers: your level and your chars-per-minute.

We read every reply, and if enough of you send numbers, we'll publish the range in next week's brew so you can see where you actually sit in the cohort.

これからも一緒いっしょに頑張がんばりましょうね〜 💪