How-to · Writing
How to count words in Chinese and Japanese correctly
The short answer
Count Chinese and Japanese by character, not by spaces. CJK text has no word-delimiting spaces, so ordinary counters see a whole paragraph as one "word". Word Counter counts Han, Hiragana and Katakana per character, applies word rules to the Latin parts of mixed text, and runs entirely locally.
- 1
Select the Chinese or Japanese text on any page — or copy it.
- 2
Click the Word Counter icon; paste into the popup if you didn't select.
- 3
Read the headline count: CJK characters counted one each, Latin words by word rules.
- 4
For mixed text, check the breakdown of Latin words vs CJK characters.
- 5
Use the character totals (with and without spaces) against platform or manuscript limits.
Paste a Chinese paragraph into a typical word counter and you get a number like 1 — or 3, if the text happens to contain some punctuation. The tool isn't broken; it is splitting on spaces, and Chinese and Japanese don't put spaces between words. The result is useless, because everything that measures CJK text — publishers' manuscript limits, translation pricing, platform character caps — measures it in characters. Here is how to get counts that match that convention, in Chrome, without uploading your text.
Count per character, because that is the convention
Word Counter counts every Han, Hiragana and Katakana character as one unit — including iteration marks like 々 — instead of applying Latin word rules to text that has no spaces. That matches how Chinese and Japanese are actually measured: a 字数 (character count) requirement, a translation quote priced per character, a 3,000-character essay limit. Select the text on any page and open the popup, or paste the text in; the totals appear instantly.
Mixed text gets both rules — and shows its work
Real documents mix scripts: a Japanese article quoting an English product name, a Chinese report full of Latin acronyms. Word Counter applies word rules to the Latin runs and per-character counting to the CJK runs, sums them into the headline number, and — when a selection genuinely mixes scripts — shows the breakdown, so you can see how many Latin words and how many CJK characters went into the total instead of trusting a single opaque figure.
Korean is deliberately different
One honest boundary: Korean is not counted per character. Hangul is written with spaces between words, and Korean text is conventionally measured in words — so it goes through the word rules, where it belongs. A counter that lumped Korean in with Chinese and Japanese "because CJK" would inflate every Korean count several-fold.
Sentences and characters work for CJK too
Sentence counting recognizes the full-width terminators 。！？ alongside . ! ?, so Japanese and Chinese prose gets real sentence totals. Characters are counted as Unicode code points, so an emoji or a rare ideograph counts as one character, not two. And the character totals come with and without spaces — the number platform limits actually check.
All local, on any page
The extension reads a page selection only when you click its icon on that tab — no background access, no upload, no account, no paid tier. That matters precisely for the texts people count most carefully: unpublished manuscripts, client translations, exam essays. For the general workflow — selecting text and reading the full set of counts — see counting words on any web page.