Word Counter
Count words, characters, sentences and paragraphs, with reading time.
Everything runs in your browser. Nothing you type is uploaded or stored.
In-depth guide
Word counter: how each figure is counted
The exact rules behind words, characters, sentences, paragraphs, lines and reading time, including surrogate pairs, emoji, combining marks, Chinese and Japanese word segmentation, abbreviations at sentence ends and where other counters will disagree.
Key figures at a glance
- Words
- Runs of non-whitespace, Chinese and Japanese segmented
- Text is split on whitespace; unspaced Chinese or Japanese runs are then split into words with Intl.Segmenter, or one word per character if the browser lacks it.
- Characters
- Unicode code points
- An emoji such as 🙂 counts as 1 (not 2 UTF-16 units), but 👨👩👧 counts as 5 and 🇸🇬 as 2.
- Reading time
- 200 words per minute
- Speaking time uses 130 words per minute.
- Line breaks
- CRLF counts as one
- A trailing newline adds one more (empty) line.
What is counted, and how
All counting runs in your browser as you type. Windows line endings (\r\n) and lone \r are first turned into a single \n.
| Figure | Rule |
|---|---|
| Words | Split on any run of whitespace; every non-empty piece is a word. well-known, don't and 3.14 are one word each; a lone — or - between spaces also counts as a word. A piece that contains Chinese or Japanese is split into words with Intl.Segmenter('word'); without it, each Han, hiragana or katakana character counts as one word. |
| Characters | Number of Unicode code points, line breaks included. |
| No spaces | Same count after deleting every whitespace character (JavaScript \s). |
| Sentences | A run of . ! ? followed by whitespace or the end of the text ends a sentence, and so does 。 ! ? (no space needed). Closing quotes and brackets stay with their sentence. A single full stop after Dr, Mr, Mrs, Ms, Prof, Sr, Jr, vs, e.g or i.e does not end one. Text after the last terminator counts as one more sentence. |
| Paragraphs | Blocks separated by two or more consecutive newlines; empty blocks ignored. |
| Lines | Number of newline-separated lines; empty text is 0. |
| Reading / speaking time | Words ÷ 200 and words ÷ 130, rounded to the second. Under 60 seconds it shows seconds, otherwise whole minutes. |
| Longest word | The longest word, measured in UTF-16 units, punctuation included (Hello,). Chinese and Japanese are measured after segmentation. |
Characters: code points, not letters you see
JavaScript strings are UTF-16, so "🙂".length is 2. The tool deliberately counts code points, which treats such an emoji as one. It does not count grapheme clusters (what readers perceive as one character, defined in Unicode UAX #29). Computed with the page's code and Intl.Segmenter:
| Text | UTF-16 units | Tool's characters | Graphemes |
|---|---|---|---|
🙂 |
2 | 1 | 1 |
é (one code point) |
1 | 1 | 1 |
é (e + combining accent) |
2 | 2 | 1 |
👨👩👧 (three people joined) |
8 | 5 | 1 |
🇸🇬 (flag) |
4 | 2 | 1 |
தமிழ் |
5 | 5 | 3 |
新加坡 |
3 | 3 | 3 |
So the figure can be higher than what you see for composed emoji, flags, decomposed accents and Tamil conjuncts. A platform that limits by UTF-16 units (some social networks and SMS segments count differently again) will give another number; check the platform's own rule.
Whitespace for "no spaces" is JavaScript \s: it removes space, tab, line breaks, non-breaking space U+00A0, U+FEFF and Unicode space separators, but not the zero-width space U+200B. In a test, a, non-breaking space, b, zero-width space, c, space, d gave 7 characters, 5 without spaces and 3 words, because the zero-width space is invisible but is neither whitespace nor removed.
Words in Chinese, Japanese, Thai and Tamil
Words are found by whitespace, and then Chinese or Japanese runs are split further. With Intl.Segmenter('word'), which current browsers provide, 我喜欢新加坡。你呢? counts as 5 words (我 / 喜欢 / 新加坡 / 你 / 呢) in 10 characters, and I like Singapore 新加坡 a lot counts as 6. Without the segmenter each Chinese character counts as a word, which gives 8 and 8 for those two texts. The exact split is the browser's dictionary decision, so another browser may count a few words differently. Thai is still counted only by spaces. Tamil is space-separated like English, so தமிழ் மொழி is 2 words.
Sentences, paragraphs and lines
- Sentence terminators are ASCII
. ! ?followed by whitespace or the end of the text, and the Chinese and Japanese。!?, which need no space:你好。我是小明。你呢?is 3 sentences. - A terminator must be followed by whitespace or the end, so
3.14andexample.comdo not end a sentence. A single full stop after Dr, Mr, Mrs, Ms, Prof, Sr, Jr, vs, e.g or i.e does not either:Dr. Smith arrived. He sat.is 2 sentences, andMr. and Mrs. Lee met Ms. Tan, e.g. at noon, i.e. late. Done.is 2. Other abbreviations still end a sentence.Wait... what?! Really.is 3; a run like...or?!counts once. A closing quote after the terminator stays with its sentence:"Hello." She waved.is 2. - Text after the last terminator counts as one more sentence:
Hi. thereis 2. With no terminator at all, any non-empty text is 1. - A paragraph break needs two consecutive newlines. A "blank" line with spaces in it does not count:
para one, a line with one space,para twois 1 paragraph on 3 lines. trailingfollowed by a newline is 2 lines.
Reading and speaking time
Time is simply words ÷ 200 or ÷ 130. Examples: 65 words gives 20 s reading and 30 s speaking; 200 words gives 1 min reading and 2 min speaking (92 s rounds to 2 min); 500 words gives 3 min reading (150 s rounds up) and 4 min speaking. Chinese and Japanese are timed from the segmented word count, which is only a rough guide.
Tip: When a form has a strict limit, test the exact string in the form. Different systems count line breaks (1 or 2), emoji and accents differently.
Common mistakes
- Treating the word figure for Chinese or Japanese as exact; it depends on the browser's word segmentation.
- Comparing with another counter and assuming one is wrong; the difference is usually line breaks, hyphens, or emoji.
- Assuming a visible character equals one counted character for flags, family emoji and Tamil.
- Forgetting that invisible characters such as U+200B add to the character count.
Related tools
Hidden character cleaner finds zero-width characters that inflate counts. Unicode explorer looks up the code point of a single character. Text sorter, Case converter and Remove line breaks tidy text before counting. Text diff compares two versions.
Frequently asked questions
Is my text sent to a server?
No. Everything runs in your browser using JavaScript. Nothing you type or paste is uploaded, logged or stored anywhere.
Is there a length limit?
There is no fixed limit — the practical ceiling is your browser's memory. Documents of tens of thousands of words are handled comfortably.
Does it handle non-English text?
Yes. Words are counted by whitespace, which suits English, Malay and Tamil. Chinese and Japanese are written without spaces, so those runs are split into words with the browser's built-in word segmenter (Intl.Segmenter); where it is missing, each Chinese or Japanese character counts as one word. Sentences end at . ! ? and also at 。 ! ?.
Does it change my original text?
No. There is no output text at all: the tool only reads what you type and shows the figures.
Sources
- Unicode UAX #29 — Text Segmentation — Defines grapheme clusters, the 'user-perceived characters' this tool does not count.
- MDN — String iterator (code points)
- MDN — Intl.Segmenter
- MDN — Character class escape \s
This guide explains published rules in plain language. It is not financial, legal or official advice — rules change, so confirm with the agency linked above before you commit.