How the vocabulary test works
Press Start the vocabulary test and you will see 80 words, one at a time. For each one, answer I know it or I don't know it (keys 1 or ← and 2 or →). There is no time limit and the test takes about 3 minutes. Only say you know a word if you know what it means; recognising that it looks familiar is not enough.
56 of the words are real English words, 8 from each of 7 frequency bands, picked at random from a larger list each time you play. The other 24 are made-up words built to sound English, such as plimble or drosk. They are there to catch guessing: every made-up word you accept lowers your estimate. Unlike the verbal memory test, which checks whether you remember words you just saw, this test asks what you already know. It pairs well with the reading speed test.
How your estimate is calculated
The real words come from seven bands of the open wordfreq frequency list (Speer, 2022), which ranks words by how often they appear in books, subtitles, news and the web:
- words 1–1,000 and 1,001–2,000 (everyday words such as house or weather)
- words 2,001–4,000 and 4,001–8,000
- words 8,001–16,000 and 16,001–32,000
- words 32,001–64,000 (rarer words such as taciturn or moiety)
For each band, the page takes the share of real words you said you know and subtracts the share of made-up words you said you know. This correction for guessing is the same idea used in large online vocabulary studies by Keuleers, Brysbaert and colleagues. A negative value counts as zero. Sampling a few words from each frequency band and scaling up is the logic of the Vocabulary Size Test by Nation and Beglar (2007), which uses 1,000-word levels. The corrected share is multiplied by the size of the band, the seven results are added up and rounded to the nearest 100. The maximum is 64,000.
An example: if you accept 2 of the 24 made-up words (8 %) and 6 of the 8 words in the 8,001–16,000 band (75 %), that band counts as 67 % of 8,000, about 5,300 words. Your result card lists the corrected share for every band.
If you accept 12 or more of the 24 made-up words, the page shows no number, because it would mostly measure guessing. The card also shows an accuracy score in the style of LexTALE (Lemhöfer & Broersma, 2012): the average of the percentage of real words you said yes to and the percentage of made-up words you correctly rejected.
What your number means
In the largest study so far, Brysbaert, Stevens, Mandera and Keuleers (2016) estimated that an average 20-year-old native speaker of American English knows about 42,000 lemmas (base words, not counting names), with the lowest 5 % around 27,000 and the highest 5 % around 52,000. Between ages 20 and 60 people learn about 6,000 more.
Your estimate is not directly comparable to those figures, which is why the result shows no "typical" marker. Our bands follow a word-form list that also contains inflected forms and names, and each run samples only 8 words per band. In the rarest band, 8 words stand for 32,000, so one answer there moves the estimate by 4,000 words. Treat the number as a rough guide, and compare your own results over time or with friends taking the same test. Your best estimate and recent results are saved in this browser and shown as a small line chart.
What affects your score
- Reading. Most rare words are met in books and articles, not in conversation. Regular reading is the main driver of vocabulary growth.
- Age and education. In the Dutch crowdsourcing study by Keuleers and colleagues (2015), age, education and knowing several languages were the strongest predictors of vocabulary size.
- Native language. English learners usually score lower in the rare bands. LexTALE was built to give advanced learners a quick measure of their English vocabulary.
- Honesty. Saying yes to half-known words inflates the bands but also raises the made-up-word penalty.
Limits
This is a self-check for curiosity, not a language certificate or an IQ test. Word frequencies differ between British and American English and between fields, so a scientist or a nurse may know rare words that the list ranks low. The sample per band is small, so expect a few thousand words of difference between runs.
More tests like this are collected on the memory and brain tests page.
Sources
- Brysbaert, M., Stevens, M., Mandera, P., & Keuleers, E. (2016). How many words do we know? Practical estimates of vocabulary size dependent on word definition, the degree of language input and the participant's age. Frontiers in Psychology, 7, 1116.
- Lemhöfer, K., & Broersma, M. (2012). Introducing LexTALE: A quick and valid Lexical Test for Advanced Learners of English. Behavior Research Methods, 44(2), 325–343.
- Keuleers, E., Stevens, M., Mandera, P., & Brysbaert, M. (2015). Word knowledge in the crowd: Measuring vocabulary size and word prevalence in a massive online experiment. Quarterly Journal of Experimental Psychology, 68(8), 1665–1692.
- Nation, I. S. P., & Beglar, D. (2007). A vocabulary size test. The Language Teacher, 31(7), 9–13.
- Speer, R. (2022). wordfreq (version 3.0) [Software]. Zenodo.
Frequently asked questions
How many words does the average person know?
Brysbaert and colleagues (2016) estimated that an average 20-year-old native speaker of American English knows about 42,000 base words, with most people between about 27,000 and 52,000. This test uses a different word list and a small sample, so its number is a rough guide and not directly comparable.
Why are there fake words in the test?
Made-up words show how often you say yes without really knowing a word. The share of made-up words you accept is subtracted from every frequency band, so guessing lowers your estimate instead of raising it.
Why did I get no number?
If you accept 12 or more of the 24 made-up words, the result would mostly reflect guessing, so the page shows tips instead of a number. Only answer I know it when you know the meaning.
Why does my score change between attempts?
Each run picks 8 words per band at random. In the rarest band 8 words stand for 32,000, so a single answer there moves the estimate by 4,000 words. Take the test a few times and look at the trend.
Is this test good for English learners?
Yes, as a quick self-check. The yes/no format with made-up words is the same idea as LexTALE, a short test designed for advanced learners of English. Learners usually know fewer words in the rare bands.
How can I improve my vocabulary?
Read widely and regularly, especially books and long articles, and look up words you meet more than once. Rare words are learned mostly from reading, not from conversation.