SANG CAO LABS · blog

How many English words do you need? Vocabulary size by CEFR level

CategoriesLearning

Ask how many words you need to be fluent in English and you will find answers from 500 to 20,000. They are not all wrong. They answer different questions, and they count different things as a “word”. Once those two things are clear, the research gives numbers you can plan around.

First: what counts as a word

Researchers count vocabulary in three main ways, and the choice changes the number a lot:

  • Word family: a base word and its common relatives. Decide, decides, decided, decision, decisive, undecided are one family.
  • Lemma: a base word and its grammatical forms, one part of speech. Decide, decides, decided, deciding are one lemma; decision is another.
  • Word form: every spelling counted separately.

The same vocabulary can be 3,000 word families or well over 6,000 lemmas. When two sources disagree, check this first. Most of the classic figures below are in word families.

How much of a text you need to know

The most useful way to think about it is coverage: the share of words in a text you already know. Research on reading found two thresholds:

  • At 95% coverage, one unknown word in every twenty, most readers get the gist but miss details.
  • At 98% coverage, one unknown word in fifty, comprehension becomes comfortable and you can read without a dictionary.

Paul Nation worked out how much vocabulary gives 98% coverage for different kinds of English (Nation, 2006):

ForWord families for 98% coverage
Spoken English (conversation, films)6,000–7,000
Written English (novels, newspapers)8,000–9,000

A later study of learners set a minimal threshold of 4,000–5,000 word families (95% coverage) and an optimal one of 8,000 word families (98% coverage) for reading (Laufer and Ravenhorst-Kalovski, 2010).

Those numbers are larger than most “you only need 1,000 words” claims, and there is a reason for the gap. The most frequent 1,000 or 2,000 words do cover a large share of everyday speech, often quoted as around 80%. But 80% coverage means one word in five is unknown. That is enough to get by, not enough to understand.

Vocabulary size at each CEFR level

The CEFR levels (A1 to C2) describe what you can do in a language, not how many words you know. Researchers have measured both in the same learners, though, and the results line up well. For English, Milton and Alexiou (2009) found:

CEFR levelWords known (approx.)
A1under 1,500
A21,500–2,500
B12,750–3,250
B23,250–3,750
C13,750–4,500
C24,500–5,000

Two cautions. These were measured with a test that only checks the 5,000 most frequent words, so C2 learners probably know more than 5,000; the test simply cannot see beyond that. And the ranges are averages of real learners, not requirements: a B2 speaker with 5,000 words and weak grammar is still B2.

What the table does show is that most of the climb from B1 to C1 is not grammar. It is a thousand or two more words.

How many words native speakers know

For comparison, a large study of native speakers estimated that an average 20-year-old American knows about 42,000 lemmas, from about 11,100 word families. The range ran from about 27,000 lemmas for the lowest 5% to 52,000 for the highest 5%, and adults kept learning roughly one new lemma every two days until 60 (Brysbaert and colleagues, 2016).

You do not need that to be fluent. It explains why advanced learners keep meeting unknown words in novels long after they are comfortable in conversation.

What this means for learning

  • At A1–A2, the most frequent 2,000–3,000 words matter more than anything else. Every one of them appears constantly.
  • At B1–B2, the jump to 4,000–5,000 word families is what makes unsimplified text readable. This is where most learners stall, because these words appear less often and are not met enough by chance.
  • At C1 and above, the gains are in less frequent words and in depth: collocations, register, the difference between meticulous and fussy.

Frequency is the thread through all of it. A word’s usefulness falls steeply with its rank, so learning words roughly in frequency order, within your level, is the most efficient route.

Finding your own number

Free vocabulary size tests show you words from different frequency bands and extrapolate from how many you know. They take a few minutes and are accurate enough to tell you which band to study next. Be honest when checking words: “I have seen it” is not “I know what it means”.

What I am building for this

I am building an iPhone app, VocabPeek, that puts one English word at a time on your Lock Screen and Home Screen. Its words are graded by CEFR level from A1 to C2, plus a C2+ band above that, and within each level the more common words come first, following the order this research suggests. A short check when you first open it sets your starting level, with sample words at each level so you can judge for yourself.

It is not on the App Store yet, and it has no release date.

Short answers

How many English words do you need to be fluent? About 3,000 word families covers most everyday conversation. Comfortable reading of unsimplified text takes around 8,000 word families.

How many words do you need for B2? Learners at B2 know roughly 3,250–3,750 words, according to Milton and Alexiou (2009).

How many words do you need for C1 and C2? Roughly 3,750–4,500 for C1 and 4,500–5,000 or more for C2, measured on the 5,000 most frequent words.

How many words does a native English speaker know? An average 20-year-old American knows about 42,000 lemmas, from around 11,000 word families.