Word counter

How the words are counted

Counting by splitting on spaces is the obvious approach and it is wrong in ways people notice. Chinese and Japanese are written without spaces, so a whole page of them counts as one word. Hyphenated compounds and contractions get miscounted everywhere else.

This tool uses your browser's Unicode segmentation instead, the same rules that decide what a double-click selects. That gets Chinese and Japanese right, keeps don't as one word, and treats punctuation as punctuation rather than as part of the word beside it.

And the sentences?

Splitting on full stops breaks Mr. Smith into two sentences, and does the same to 3.14 and J. R. R. Tolkien. Unicode sentence segmentation handles the decimals and most lowercase shorthand on its own, but it has no way to know that Mr is an abbreviation rather than a very short sentence — that normally comes from a per-language exception list the browser does not provide.

So a short list of English abbreviations and initials is applied on top of it. Titles like Dr. and Prof., shorthand like etc. and vs., and single initials no longer end a sentence, while genuine full stops still do.

Where the reading time comes from

238 words per minute for silent reading of English prose, which is the figure from Brysbaert's 2019 review of some 190 studies — a more defensible number than the 200 or 250 usually quoted without a source. Speaking time uses 130 words per minute, around the pace of an unhurried presentation.

Why ignore common words?

Because without it, every document's most-used words are the, of and and — accurate, and no use to anyone. Switching it off shows the raw counts. The list of ignored words is deliberately short and English-only: a longer one starts removing words that carry meaning.

Is my text sent anywhere?

No. Everything is counted in your browser as you type. Nothing is uploaded, nothing is stored, and closing the tab discards it.