Vocabulary Size Estimate Calculator

How vocabulary tests work

Asking “do you know all 100,000 words in Russian?” is impossible. Vocabulary tests sample instead. They draw words evenly from across a defined frequency range, ask which ones you know, and scale your score up to that range.

The formula:

estimated_size = (words_known ÷ words_tested) × the frequency range the test covers

The multiplier is the test’s range, not the size of the dictionary. This is the mistake that makes most homemade estimates useless, and it is worth being blunt about. Paul Nation’s Vocabulary Size Test draws ten words from each of the first fourteen 1,000-word frequency bands, so its range is 14,000 word families. Score 35 out of 50 on it and you know roughly 9,800 families. Multiply that same 70% by the 170,000 entries in an unabridged English dictionary instead and you get 119,000, which is more words than any human being has ever known. The dictionary is full of archaic, technical, regional and proper-noun entries that no test samples and no speaker uses.

Common ranges to enter below:

Estimates get noisier with small samples. Fifty words gives a rough figure, two hundred gives a solid one. Whatever you use, the sample has to be frequency-graded: words picked at random from a dictionary land almost entirely in the rare tail and will tell you that you know nothing.

How big is a language, really?

The right-hand column is the one that matters. Educated natives use well under a quarter of what the dictionary holds, and a second-language learner at C2 typically sits at 12,000 to 15,000 word families. If a calculator hands you a number above 35,000, it has multiplied by the wrong thing.

The 80/20 reality of vocabulary

Research by Paul Nation and others measures how much of an ordinary text a given vocabulary covers:

The first hundred words carry half of any text, because they are the connective tissue rather than the content. The first two thousand carry most of the rest, which is why beginners can survive a simple conversation surprisingly early. Then it flattens hard: going from 95% to 98% costs another three thousand families for three percentage points. Reading without a dictionary needs that 98%, and below it you are stopping on every page.

Those are the same figures the vocabulary coverage calculator uses, so the two pages will always agree about what a given word count buys you.

What this means for study

For practical fluency, target 5,000 to 8,000 word families. That covers casual reading, all conversation, and most television with subtitles. Going past 15,000 is a slow grind that pays back mainly in literary reading and academic writing.

Spaced repetition (Anki, Memrise) is the most efficient vocabulary-acquisition method anyone has measured. Ten to twenty new cards a day for a year gets most learners to the 5,000 level in a Category I language. Doubling that pace tends to break people within a few weeks, so consistency wins here.