Should you look up every word when reading?
Ask this in any language-learning forum and you get two confident, opposite answers within the hour. "Look up everything, that's how you learn." "Never look anything up, you'll kill your reading."
Both camps are right, and the thread never ends because they're answering about different texts. The question can't be settled in general — only for the thing currently open in front of you.
What a lookup actually costs
The reason "just look it up, it takes three seconds" is wrong isn't the three seconds.
You lose the sentence. You come back from the dictionary and reread from the start, because the first half has gone. On a hard page that's not a three-second tax, it's more like fifteen, times forty words, and now the chapter takes ninety minutes instead of twenty.
Then you lose the thread — who's speaking, what was just implied, why this paragraph matters. Comprehension is built across sentences, and dictionary trips keep resetting the build.
And eventually you lose the habit. Reading that feels like a lookup queue stops being something you do in the evening. This is the failure mode that ends most people's reading in a foreign language, and it never announces itself as "too many lookups" — it just quietly stops being fun.
None of which means never look anything up. It means a lookup has a price, so it should buy something.
The deciding number
One number sorts almost every case: how many words on a page do you not know?
- Under about 2 in 100 — you're in comfortable territory. Look up almost nothing; context will carry you, and the few words that matter will repeat.
- Around 5 in 100 — the useful working zone for learning. Look up selectively, a handful per session.
- Over about 10 in 100 — you can't read this text yet in any useful sense. Either look up everything and treat it as a study exercise, or put it down and find something easier.
That ~98% figure comes out of vocabulary research on unassisted reading comprehension, and it's the most practical thing in the whole field: it tells you the answer to "should I look up every word" is mostly decided by the text, not by your discipline.
If you're constantly agonising over this question, that's information. It usually means you're reading one rung too high. Here's how to pick the rung.
When the answer is genuinely yes
Look up everything when the text is short and you've chosen it deliberately: one news paragraph, the first page of a chapter, a song, a scene of dialogue. Small, dense, worked over completely.
This is a legitimate and useful mode. The point isn't reading volume — it's taking one piece of real language apart. Twenty minutes on a paragraph, every unknown word checked, is a good session, and forums that shout "never look things up!" are wrong about it.
The mistake is doing this to a novel. That's a study exercise applied at the wrong scale, and it's where the ninety-minute chapter comes from.
When the answer is no
Look up almost nothing when you're reading for volume: a full article, a chapter, anything where finishing is the point.
Not because looking up is bad, but because at that scale the arithmetic doesn't work. You need the pages more than you need those particular words — and words met repeatedly across a lot of text stick better than words looked up once and forgotten by Thursday.
The three words that earn a stop
Even in a no-lookups session, some words are worth it. In practice, three kinds:
- It's blocking the sentence — not "I'd like to know this," but you genuinely cannot follow what's happening without it.
- It's come up three times — the text is telling you it matters in this context. That's the strongest signal you'll get.
- It's in your life — a word from the domain you actually read or talk about. Those pay rent.
Everything else can go unlooked-up with no guilt whatsoever. A rare word you'll never meet again isn't worth the sentence it costs.
What about the ones you skipped?
This is the part the "never look anything up" camp never answers well, and it's a fair objection: if you skip a word, you've learned nothing from meeting it.
The fix isn't looking it up mid-sentence, it's deferring it — mark it, finish the passage, then deal with the two or three that mattered. You keep the sentence and the word.
That's the whole reason tap-to-look-up readers exist: they collapse the price of a lookup close to zero, so the deferral gets easy. Mossglen (full disclosure: ours) is a free Chrome extension that turns any page into a clean reading view in the original language — tap a word, get the meaning in place, and it's saved with the sentence it appeared in. Nothing to import, no account, 8 languages.
What to do with those saved words afterwards is a separate question, and it's here.
The short version
- The argument is unsettleable in general and easy per text
- Count unknown words per 100: under 2, look up nothing; around 5, be picky; over 10, the text is too hard
- Short deliberate passage → look up everything, that's a real method
- Full article or chapter → look up almost nothing, finish it
- Always worth a stop: blocks the sentence, appeared three times, from your world
- When in doubt, mark and keep going — deal with it at the end of the page
If this keeps coming up because reading feels slow, that's a different problem with a different answer: how to read faster in a foreign language. And the case for reading as the main engine is in learning a language by reading.
Mossglen turns the pages you browse into reading practice.
Add to Chrome — freeKeep reading
FluentU alternatives: what you're actually paying for
Looking for a FluentU alternative? Break the subscription into its three layers — video library, interactive captions, review — and replace each one.
Migaku vs Anki: do you still need both?
Migaku vs Anki in 2026 — Migaku now has its own SRS, so the old "card creator plus Anki" setup is optional. What each costs, and when one is enough.
Spaced repetition without Anki: four routes that work
Want spaced repetition without Anki? Four routes compared — built-in SRS tools, readers that collect the words for you, paper, and other card apps.