How many words do you know?

Free vocabulary size tests in nine languages · two minutes each · no sign-up

Each test takes about two minutes. It shows you around sixty words, one at a time, and asks one question: do you know this word? Some of the words are real and some are invented to look real. At the end you get an estimate of how many words of the language you know, worked out from the course's own frequency list, and — for every language with a large enough corpus — how much of ordinary film and television dialogue those words account for.

There is no sign-up, nothing is stored and nothing is sent anywhere: the test runs in your browser. You can share your result as a link with its own picture, and take the test again as often as you like — every attempt draws different words.

How a two-minute test can count thousands of words

By sampling. Nobody could sit through every word on a list of eighteen thousand, but you do not need to: a frequency list can be cut into bands — the first thousand words, the next thousand, and so on — and a handful of words chosen at random from each band tells you, roughly, what share of that band you know. Multiply each share by the size of its band, add the bands up, and you have an estimate of the whole. This is how vocabulary size has been measured in research for decades, and it works for the same reason an opinion poll does.

The weakness of a yes/no test is that it takes your word for it, and people are optimists. That is what the invented words are for. A word you have never seen that looks like the language is a fair test of how readily you say yes; if you say yes to two of ten of them, the test assumes you would have said yes to about two in ten of the real words you do not know as well, and takes that off every band. Say no to all of them and you are taken at your word.

What the number is, and what it is not

It is a count of words you recognise, in the course's own list, by frequency. It is not a count of the words you can use, which is always smaller, and it is not a level: two people with the same number can have very different command of a language. It stops where the list stops, so on a short list a strong speaker simply reaches the top. And it is not a ranking — each page says plainly that we have no data on how other learners score, rather than inventing a percentile.

What it is good for is a quick, honest bearing: roughly where you are on the frequency list, and so which words to learn next. The word lists are the same lists, in the same order, and each course teaches them from the top.