Our methodology is very straightforward: we have a list comprised of thousands of English words sorted by the frequency of use and only dictionary words are left in the database.
Respondent is shown 96 random words and asked to mark the words they know at least one definition for. First 32 words are taken from the list of TOP-1000 most used English words. Next 32 are from the TOP-10000 and the final 32 are from the rest of the list.
Then we simply extrapolate from there. We assume that the person who knows only 50% of words out of the TOP-10000 has a vocabularity consisting of approximately 5000 English words. Of course, it is not scientific, but presumably the margin of error is acceptable.
We also toss up in some fake words, and if respondent marks them as words they know, it does affect the total estimated word count.
If you have a suggestion on how we can improve the methodology, please let us know