How many words do you need to read a book in another language?

The research answer is about 8,000 word families for comfortable independent reading, and around 4,000 to get through a novel with help. Both numbers sound discouraging, and both are less binding than they look.

What "coverage" means

Vocabulary research talks about lexical coverage: the share of the words on a page that you already know. The thresholds that keep coming back are 95% and 98%.

They sound close. They are not. At 98% coverage you meet an unknown word roughly every fifty — about one per short paragraph, which you can guess past without stopping. At 95% it is one word in twenty, which on a normal page is every other line.

That is the difference between reading a book and decoding one, and it is why three percentage points decide whether people finish.

Why the number is so high

Word frequency is brutally uneven. The most common 1,000 word families cover about 80% of ordinary text, and the next thousand adds only about five points. After that each additional thousand buys less and less.

1,000 families — roughly 80% coverage. One unknown word in five.
2,000 families — around 85%. Enough for graded readers, not for novels.
4,000 families — around 95%. A novel becomes possible with steady help.
8,000 families — around 98%. Comfortable, mostly unaided reading.

So the last stretch from 4,000 to 8,000 doubles your vocabulary to buy three percentage points. Those three points are the ones that matter, and they are the slowest to earn — which is exactly why they are earned by reading rather than by studying.

Why the tail never quite ends

Even at 98% there are words left, and they are not the ones a course teaches: proper nouns, dialect, period vocabulary, an author's particular favourites, and the idioms whose meaning you cannot get from the parts. A native reader meets these too and mostly does not notice, because guessing from context is what fluent reading actually is.

Chasing 100% is not a goal anybody reaches in their own language either.

The number is not a gate

Coverage is a property of a text and a reader together, not of the reader alone. You do not have one number — you have a different one for a children's book, a thriller and a nineteenth-century novel.

Which means the practical move is not to wait until you have 8,000 families. It is to pick a book where your existing vocabulary already lands near 98%, finish it, and let the number rise. Volume raises coverage faster than lists, because frequent words repeat until they stick and rare ones do not need to.

Making 95% feel like 98%

There is a second way to close the gap: reduce what an unknown word costs you. At one unknown word in twenty, a lookup that takes a minute ends the session — you lose the sentence, then the paragraph, then interest. A lookup that takes a second does not.

That is the whole argument for in-context explanations over a dictionary. A dictionary returns every sense of a word and leaves the choosing to you, which is the hard part when you are still learning. The question is never "what can this word mean?" — it is "what does it mean here?"

Get the cost of a lookup low enough and a book at 95% becomes readable long before your vocabulary says it should be.

Read above your level

ClickBook explains the word you tapped in the sentence you tapped it in, translated into your mother tongue and generated on your device in about a second. See how it works, or read how to read a book in a language you are still learning.