Word Counter

Counting words sounds trivial until you ask what a word is. This tool leans on a real Unicode segmenter through Intl.Segmenter when the browser has it, counting words by visible boundaries, characters by grapheme clusters, and sentences by punctuation. The six questions below cover what that means in practice.

How the counting works under the hood

What counts as one word?

Whatever Intl.Segmenter marks as a word: a run of letters or digits. A contraction like don't counts once, while a hyphenated compound like well-known counts as two words, well and known, because the hyphen is a boundary to the segmenter. Numbers count too, which matches how Microsoft Word does it.

How does it handle Chinese, Japanese, or Thai text?

Those languages put no spaces between words, so the segmenter splits on dictionary boundaries and returns a real word count instead of treating every character as a word. A whitespace split would badly undercount them.

Why does my count differ from Microsoft Word?

Usually hyphens and punctuation. This counter splits well-known into two words where Word, which splits on whitespace, counts one, and it breaks a URL into several segments at its periods and slashes where Word keeps the whole thing as one word. On a 1000-word document the two land within about one percent.

Are emoji counted as characters?

A single-code-point emoji counts as one. A compound emoji joined by zero-width joiners, like the family-of-four or a flag, also counts as one when your browser segments it as a single grapheme.

How are sentences detected?

By runs of sentence-ending punctuation: period, exclamation mark, question mark, ellipsis, and the CJK equivalents. Before counting, the tool masks decimals like 3.14, common abbreviations like Dr. and e.g., and initials like D.C., so Dr. Smith is here. counts as one sentence, and an ellipsis counts as a single boundary rather than three.

Is there an input size limit, and does it work offline?

No fixed limit; the tool counts whatever you paste. Everything runs in your browser with nothing sent to a server, so you can pull the network cable and it keeps counting.