About this site
vocabularybuilding.org is written by the team behind Readlang Mobile, an iPhone and iPad app for reading books in a foreign language with tap-to-translate. We publish what the research says about learning vocabulary and our own counts of words in books, with every figure linked to its source.
Who we are
We are a small team that builds Readlang Mobile, an app for iPhone and iPad for reading in a foreign language. You tap a word or phrase and see its translation in the line you are reading; the app has a library of public-domain books graded from A1 to C1, opens web articles, and on its paid plan reads your own EPUB files. Readlang Mobile is an independent app and is not affiliated with Readlang (readlang.com), the web reader made by Steve Ridout.
Building a reading app means spending a lot of time with the questions this site covers: how hard a book is, how many words a learner needs, which words are worth reviewing. We started the site to put what we have learned, and the data we have, in one place that does not depend on using our app.
How we write
- Figures come from the original papers. Every number from research is checked against the paper itself, or its abstract when the full text is not open, and cited with a link. In the rare case where we could only confirm a figure through later studies that report it, the page says so. Where studies disagree, or a result comes from one small study, we say so.
- Tools are compared on the same terms. When we compare reading and flashcard tools, our own app appears in the same table as the others, with its limits, and it is marked as ours. Prices change; we give the date we checked them.
- No invented examples. Numbers about books come from our own counts, described below, not from estimates.
- Few pages, kept up to date. We add pages slowly and update the date at the top of a page when its content changes.
Where our book data comes from
The word counts for 117 books were made from the public-domain EPUB files in the Readlang Mobile library, originally from Project Gutenberg and Standard Ebooks. A script strips licence text, tables of contents and colophons, splits the text into words with the same rules for every book, and counts running words, different word forms, forms used once, how many forms cover 95% and 98% of the text, and how many different forms appear per 5,000 words. The full method and its limits are on the data page. The CEFR level shown for each book was assigned by our editors when the book was added to the app; it is a judgement, not a measurement.
The data is free to reuse under CC BY 4.0: download the CSV.
About this domain
The domain vocabularybuilding.org has had several owners since the late 2000s. The current site was started in September 2026 and has no connection to the sites that used the name before.
Contact and corrections
If you find a mistake, a broken link or a study we should include, write to [email protected]. We fix errors and note significant corrections on the page.