Online Word Counter

Paste or type text below and the word, character, CJK character and Latin word counts update as you go. Words are counted the way Microsoft Word counts them: each CJK character is one word and each Latin word is one word, so mixed text comes out right. Drop in a .txt / .md file. Everything is counted locally in your browser, nothing is uploaded.

0
Words (Word's rule)
0
Characters (with spaces)
0
Characters (no spaces)
0
CJK characters
0
Latin words
0
Paragraphs
Characters (with spaces)0All characters, including spaces, line breaks and punctuation
Characters (no spaces)0Spaces, tabs and line breaks removed
CJK characters0Each Han character, kana or hangul syllable counts as 1
Latin words0Runs of Latin letters split on spaces and punctuation
Digits0Each run of digits counts as 1, e.g. 2026 or 3.14
Punctuation0Both Western and CJK punctuation, including hyphens and apostrophes inside words
Sentences0Split on full stops, question marks and exclamation marks
Paragraphs0Non-empty blocks separated by blank lines or line breaks
Lines0Including blank lines
Reading time0 minEstimated at 400 CJK characters / min and 220 Latin words / min
Speaking time0 minEstimated at 200 CJK characters / min and 150 Latin words / min
Most frequent words

Latin text is counted by word, CJK text by single character (no segmentation). Top 20 only, excluding punctuation and single letters.

Notes:
1. “Words” follow Microsoft Word's rule: each CJK character is one word, each Latin word is one word, and each CJK punctuation mark and each run of digits also counts as one by default, matching the “Words” figure in Word's status bar; a comma or full stop after a Latin word is not counted separately. Untick the two boxes above if you do not want them included.
2. “Characters (no spaces)” matches the item of the same name in Word; the “word count” required by journals and theses usually means this, or “CJK characters + Latin words” — when in doubt, report both.
3. Latin words are split on spaces and punctuation, so don't, e-mail each count as one word; URLs and email addresses count as one word.
4. Chinese is not segmented into words, so “Most frequent words” lists single characters for Chinese text; use a dedicated segmentation tool to count phrases.
5. Your text stays in your browser — close the page and it is gone, nothing is saved or uploaded.

About word counting

What counts as a “word” differs from tool to tool

The same passage of mixed Chinese and English can differ by ten or twenty percent between Word, WPS, the WeChat editor and Zhihu, because they count differently. Word counts each CJK character, each Latin word, each free-standing punctuation mark and each run of digits as one “word”; WPS matches Word by default but its “characters” column is a different thing; the WeChat editor is closer to a character count. This page follows Word's rule by default, because “at least 3,000 words” in a thesis brief, a submission guideline or a contract usually means Word's figure — and it lists characters, CJK characters and Latin words separately so you can report whichever one the other side asks for.

Mixed Chinese and English is where counts go wrong

Pure Chinese and pure English are easy; the hard case is English inside a Chinese sentence. In Word, 「我们用 React 写了一个 demo」 is 9 words: 7 CJK characters plus 2 Latin words. Many simple online counters either count the English by letter or split the whole thing on spaces — the first turns React into 5, the second turns 「我们用」 into 1. This page first picks out every Han character, kana and hangul syllable by Unicode range, then splits the remaining Latin runs on spaces and punctuation, counts the two sides separately and adds them, so mixed text comes out right.

When to use characters, CJK characters or words

SMS, Weibo, Twitter and all kinds of forms limit characters, with or without spaces depending on the platform, and both figures are here. Translation quotes, official documents and Chinese theses usually go by CJK characters, English abstracts by word count; do not mix the two. Reading time is estimated at 400 CJK characters or 220 Latin words a minute, which predicts how long an article takes far better than a raw character count; speaking time is about half that speed, for podcasts and speeches.

This page and the other common ways to count

Word has Ctrl + Shift + G, WPS has it on the Review tab and Excel has the LEN function, but each gives one rule only, and you have to move the text in first. Here you just paste and every rule is shown at once, and when you are done you can glance at which words appear most to check a draft for overused terms. If what you have is a PDF, use Extract Text from PDF to pull the text out first; if the counted text still needs tidying, hand it to the Markdown Formatter Online.

Before an English draft goes out, the capitalisation usually needs tidying too: Title Case for the headings, sentence case for body text that was typed in all caps. The Case Converter handles that; paste the result back here to count it and the numbers will match.

FAQ

Is my content uploaded to a server?

No. Counting is done entirely by JavaScript in your browser; the page has no backend, the pasted text never leaves your computer, and a refresh clears it. Contracts, unpublished drafts and internal documents are all safe. Disconnect from the network after the page has loaded and the tool keeps working — the most direct way to check.

Why does my count differ from what Word shows?

First check that the “Count CJK punctuation” and “Count numbers” boxes match what you expect — Word's “Words” includes both. Next, Word treats a half-width space between two CJK characters as a separator while a full-width space is not; this page does the same. Finally, Word's rules for splitting URLs, email addresses and hyphenated words vary between versions, so text with a lot of them can differ by a few words, which is normal.

My thesis says “at least 5,000 words” — which figure do I use?

For a Chinese thesis it is usually “Words”, i.e. CJK characters plus Latin words; some universities explicitly say “characters (no spaces)”, which is the third card. The English abstract is reported in Latin words. The safest approach is to match the wording of the requirement to the labels here, and if it does not match, write both figures down.

Can Chinese be counted by word?

No — Chinese on this page is counted by character, and “Most frequent words” lists single characters for Chinese. Word segmentation needs a dictionary and an algorithm and the result varies by segmenter, so building it in would only make the main word count uncertain. If you need phrase frequencies, use a dedicated segmentation tool.

How are Latin words split?

On spaces, line breaks and punctuation; a run of letters, digits, apostrophes and hyphens is one word, so don't, state-of-the-art, 2026 each count as one. The dots and slashes inside URLs and email addresses are not split; the whole thing is one word. This is essentially what Word does.

Can I count the words in a file?

Yes — drag a .txt, .md or .csv text file into the input box, or click “Open file” to choose one, and its contents are read in and counted. Word and PDF files cannot be read directly: open the Word file, select all and copy; for a PDF, use this site's “PDF to Text” tool first.

How is reading time calculated?

400 CJK characters a minute for Chinese and 220 words a minute for English — the median silent-reading speed of an adult on general prose; for mixed text the two parts are calculated separately and added. Speaking time is 200 CJK characters or 150 words a minute. These are estimates, good for planning the length of a talk or a podcast; technical documentation and fiction will deviate from them.

Is there a length limit?

No hard limit — a book manuscript of several hundred thousand characters can be counted, it just makes pasting and each recount a little slower. Counting runs on your device, so the speed depends on your computer or phone.