KOTOBA BREW

第051号 · Issue 051

The Meaning Is the Part You Keep

{{first_name | みんな}}さん、こんにちは ☕

In one experiment, people studied a set of items and were then asked how much of it they would still have later. They predicted 75.7 percent. 📌

They recalled 60.3 percent.

That fifteen point gap is not laziness and it is not you. It is what happens when you judge how well you know something while the answer is still sitting in front of you. 👀

Now think about where your thumb was last night. A popup dictionary puts the answer on screen, under your cursor, for exactly as long as you want it. Then it takes it away.

Know someone who looked up forty words last night? Send them this. Share this issue →

Issue 051 illustration 1

The evening where nothing stuck

You know the one. 🎴

You read for an hour, looked up everything, followed the chapter, closed the laptop feeling productive. Next morning, you could not have said what any of those words meant.

Not a discipline problem. Close to universal around N3.

Half of it is the gap up there: you rate how well you know a word while the answer is on screen. In Japanese that tap also hands you a reading and a guess at where the word started, so more arrives at once. Our reading, that. 🚪

The other half is harder. 💡

A study in 2000 counted how often learners looked each word up: the count had almost no relation to whether the word stayed. Someone else made glosses easier to reach and watched with eye-tracking. Learners looked more, and longer, and recognised no more. ⚡

And the largest meta-analysis here, 42 studies and 4,628 participants, pulled the standard model of task demand apart. Deciding about the word contributed most, needing it came second, and searching for the meaning did not contribute at all.

一
言

☝️ Chotto ちょっと

That is about parts of a model, not about dictionaries. Looking a word up beats not looking it up: three studies found dictionary users retained more than readers with none, and in one an active lookup beat a gloss right there in the margin. The meaning is necessary. The null sits on searching, and on searching again. 🤝

Issue 051 illustration 2

Three moves, and two we will not give you

Both obvious repairs have been run. Both came out flat.

Do not guess instead of looking up. Guess-then-memorise was set against being-told-then-memorise. Two weeks later: 47 percent against 50, and guessing took longer. 😅

Do not make the lookup harder on purpose. Effort is not the mechanism. A paper dictionary is not a virtue.

And 辞書(じしょ)は引(ひ)かない is not wrong, just aimed elsewhere. That first rule of Japanese extensive reading traces to a 2003 book grounded in its author's teaching, not in evidence, and it is about reading fluency, not about whether words stay. The best known voice on vocabulary says the opposite: guess first, then confirm. 🙌

Three things, then, inside thirty seconds.

One. Guess before you tap. Three seconds. The one procedure with a positive result behind it is inference plus consultation, which beat either alone. The guess does not replace the lookup, it is something you do to it, and it needs something under it. No purchase at all? Say "no idea" and tap. 🎯

Two. When the meaning appears, reread the sentence and ask whether it fits. Deciding about a word is the component with the largest measured contribution, and this is its cheapest version inside real reading. The rereading is our extrapolation, not a tested finding. Sometimes the meaning will not fit, which is the most useful thing all evening. ⚡

Three. Take one word out of the session. One, not forty. When someone sat and watched learners look words up, a decent share were on words nobody had reason to keep. Take the one that turned up twice. 🍵

危
険

⚠️ Abunai 危(あぶ)ない

We are not adding "look it up again next time." It feels productive and has the clearest null behind it. And none of this is the big lever: producing a word from memory later ran near 80 percent after a week against near 35 without. Thirty seconds is triage. 💪

This Issue's Japanese

Word

確(たし)かめるOrdinary conversation. 確認(かくにん)する is the office-register sibling

Means: to check for yourself, to make sure. Commonly listed as N3, though the JLPT publishes no official list. 📌

Use it: 答(こた)えを確かめる。/ 意味(いみ)を確かめてから次(つぎ)へ行(い)く。/ 戸締(とじ)まりを確かめる。

Okurigana trap: the okurigana starts at か. 確かめる, never ×確める. And 確か alone is an adverb, "if I remember rightly": 確か、あの単語(たんご)は… One character for I think I remember and for let me find out. ⚡

Kanji 漢字

引

Meaning: to pull. In Japanese you pull a dictionary. 辞書を見(み)る is natural but means running your eye over it.

Grade: elementary, second year.

Readings: ひ(く)· イン

Common words: 引(ひ)っ越(こ)す to move house · 取引(とりひき)a transaction · 索引(さくいん)an index

The good bit: you have owned ひく since N5 and probably never met イン, rare for a grade two character. 👀 And 引っ越す is ひっこす because 引き tightens before the second half: gemination, not rendaku.

This issue's connection: 引く is what your finger does, and the finger is not the part that learns. 確かめる is the part that does. Same thirty seconds, two verbs. 💡

Issue 051 illustration 3

Renshuu Time 練習(れんしゅう)📝

「三秒(さんびょう)、ひと手間(てま)、一語(いちご)」. Three seconds, one extra step, one word. About fifteen minutes.

→ Open whatever Japanese you were going to read tonight. Nothing special. Set a timer.

→ Three seconds before every lookup. Guess, out loud if you are alone. If nothing comes, say "no idea" and tap. That counts. 🎯

→ After the meaning appears, reread the whole sentence with it in place. Does it fit?

→ Keep two tallies: guesses in the neighbourhood, guesses nowhere near.

→ At the end, pick exactly one word. Put it wherever you keep words.

You will know you did it when: you can name your one word without looking, and say which tally was bigger. More near misses means you have been outsourcing a judgement you could already make. The other way round is about the text, not you.

✨ Yatta やった

If you managed the three seconds and forgot the tallies, keep the one word anyway. That is the part that feeds what actually works. 🙌

The tiny win

この三十秒(さんじゅうびょう)に、ひと手間(てま)かける。 One small extra step. Not a longer lookup, not a harder one. You are not counting lookups any more. You are choosing what leaves the room with you. 🍵

One reply back

Hit reply with the one word you kept and which tally was bigger. We are collecting both, and will say so if they come back lopsided. 👀

Stay connected 🙌

We are around during the week if you want to talk shop, share a win, or just lurk. Both are completely fine.

Follow us: X / Twitter · Reddit

KOTOBA BREW

これからも一緒(いっしょ)に頑張(がんば)りましょうね〜 🍵

Kotoba Brew Editorial Desk