Our data · 117 books in five languages

How many different words are in a book?

It depends on length and language. In our set of 117 public-domain books, the median English book runs to 84,315 words and uses 7,226 different word forms; Alice's Adventures in Wonderland uses 2,634, Don Quijote 22,943. Compared over the same length of text, German, Spanish and Italian books use roughly a quarter more different forms than English ones, mostly because their words change form more.

By the Readlang Mobile team · Updated 28 September 2026

What we counted

We counted every book in the public-domain part of the Readlang Mobile library: 117 books taken from Project Gutenberg and Standard Ebooks. For each one the table gives five numbers.

Words

Running words in the text of the book, from the first line after the title page to the last line of the story, notes and all. Publisher boilerplate, the Project Gutenberg licence, tables of contents and colophons are removed.

Different forms

How many distinct spellings occur, ignoring capitals. Walk, walks and walked are three forms; names such as Holmes or Bennet count too.

Used once

Forms that occur exactly once in the whole book.

Forms for 95% and 98%

Sort the book's forms from most to least frequent and add up their occurrences: this is how many of the top forms it takes to account for 95% (or 98%) of the running words.

Per 5,000 words

The average number of different forms in each consecutive stretch of 5,000 running words. Because every book is measured over the same length, this is the fairest way to compare a novella with a long novel. Books shorter than 5,000 words have no value here.

What the numbers show

Longer books use more words, but not in proportion

Pride and Prejudice is two and a half times as long as The Great Gatsby (122,164 against 48,659 words) yet uses only a few hundred more different forms (6,407 against 5,940). New words keep appearing as a book goes on, but more and more slowly, so total counts mostly tell you how long a book is. Over equal stretches of 5,000 words, Fitzgerald uses 1,445 different forms and Austen 1,222.

Inflected languages show more forms

Over the same 5,000 running words, the median book uses 1,311 different forms in English, 1,558 in French, 1,614 in German, 1,663 in Spanish and 1,720 in Italian. A Spanish verb can appear as hablo, hablas, habló, hablaban and dozens more spellings, each counted separately; English has far fewer endings. This does not mean Spanish novels have richer vocabularies. It means a learner of Spanish meets many forms of the same word and gets a lot of mileage from learning how endings work.

Half of the vocabulary appears only once

In the median English book, 46% of the different forms occur exactly once; in German it is 62%, in Spanish 58%, in Italian 58% and in French 54%. A word list made from a whole novel is therefore mostly words you will not see again in that book. It is usually better to look such words up while reading, or skip them, and to spend review time on the words that recur. The guide to learning words from reading explains why recurring words are the ones that stick.

The level label is not a vocabulary measure

The level shown for each book is the CEFR level our editors assigned when adding it to the app. It reflects sentence length, style and subject as well as vocabulary, and it is a judgement, not a measurement. In our English books, the median variety per 5,000 words is 1,311 at B1 and 1,222 at B2: several B2 books, such as The Wealth of Nations, repeat a limited technical vocabulary in long, demanding sentences. Treat the level as a starting point and check a page yourself, as described in the first-week plan.

Summary by language

All 117 books

Download the full table as CSV. Columns are explained in What we counted.

English (37 books)

German (20 books)

Spanish (20 books)

French (20 books)

Italian (20 books)

Method and limits

• Source files. The EPUB files in the Readlang Mobile library, as downloaded from Project Gutenberg and Standard Ebooks. Other editions of the same book can differ by a few percent because of prefaces, notes and translations.
• What is removed. Everything before the Project Gutenberg start marker and after its end marker, each file's page title, the table of contents, and Standard Ebooks title, imprint, colophon and copyright pages. Translators' introductions, footnotes and endnotes stay in, because a reader of that edition meets them.
• What counts as a word. A run of letters in any alphabet, lower-cased; an apostrophe inside a word keeps it one word (don't, Sherlock's, l'homme). For the vocabulary counts, French and Italian elided articles and prepositions are cut off, so l'homme, d'homme and homme are one form, homme; otherwise every noun would be counted again with each article in front of it. Numbers are not words. Hyphenated words count as separate words. The same rules are used for the book data on our sister site readingcompanion.org, so the figures match.
• Names are included. Character and place names are part of every count, which raises the numbers for books with large casts.
• Two readers are special. First Italian Readings and Lecturas fáciles con ejercicios are textbooks with English notes and glossaries. Their counts mix two languages, so they are left out of the medians.
• Checking. Alice's Adventures in Wonderland comes out at 26,775 words, including the chapter titles and the list of contents printed in that edition. The three Sherlock Holmes collections, by the same author and in the same style, come out almost identical: 1,311, 1,283 and 1,332 different forms per 5,000 words.

The data is free to reuse under CC BY 4.0; please link to this page. Found an error? Write to [email protected].