Word counter

Words, characters, reading time and platform limits, counted correctly in any script.

54words316characters14 secread

Kept in this browser only, so a reload does not lose it. Nothing is uploaded. Clear removes it.

Reading and speaking speed

Defaults: 238 wpm silent reading of English non-fiction and 183 wpm reading aloud, from Brysbaert’s 2019 meta-analysis of 190 studies.

Longest sentence

20 words, against an average of 13.5

Reading time assumes 238 words a minute, the average for adults reading English non-fiction, and you can change it below.

Most-used words

  1. reading2 · 3.7%
  2. type2 · 3.7%
  3. adults1 · 1.9%
  4. anything1 · 1.9%
  5. assumes1 · 1.9%
  6. average1 · 1.9%
  7. browser1 · 1.9%
  8. change1 · 1.9%
  9. count1 · 1.9%
  10. english1 · 1.9%

The common-word filter is English only. Numbers and links are not ranked.

Character limits

Each one counted the way that platform says it counts, so these numbers differ from the character count above.

Social

Messaging

Video and ads

Meta description: no bar, on purpose. Google sets no limit on a meta description. It cuts the snippet to fit the screen width, so the 155–160 characters often quoted is a rule of thumb, not a rule. Google’s snippet guidance

In short

Words are counted with the Unicode word-boundary rules your browser already has, so text in any script counts properly. Hyphenated words, contractions, numbers and links count once each; emoji and punctuation count as characters, not words.

What a word count is

The number of words in a text under a stated rule for what a word is. Counters disagree because the rule differs: splitting on spaces finds one word in a line of Chinese, and splitting on punctuation makes “don’t” two. This one uses the Unicode word-boundary rules through the browser’s Intl.Segmenter, then rejoins hyphenated words and treats a link as one word.

What this tool does that others do not

  • Words in scripts written without spaces

    Chinese, Japanese and Thai are segmented with the browser’s own dictionary, so a sentence of Chinese is several words rather than one. Hyphenated words, contractions, decimals and links each count once, which matches how a reader counts them.

    我喜欢学习中文 counts as 4 words (我 · 喜欢 · 学习 · 中文); a space-splitting counter reports 1.

  • Limits counted the way each platform counts

    X weighs Chinese, Japanese, Korean and emoji double and every link as 23; Bluesky counts graphemes; an SMS drops from 160 to 70 characters the moment one character falls outside the GSM alphabet. Each bar uses that platform’s published rule and names the page it comes from.

    A curly apostrophe (’) is outside the GSM 7-bit alphabet, so a 100-character text containing one needs 2 messages instead of 1.

  • Reading and speaking rates with a source

    Reading time uses 238 words a minute and speaking time 183, the averages for adults silently reading English non-fiction and reading aloud. Both are inputs, because your audience and your pace are not the average.

    Brysbaert (2019): 190 studies and 18,573 participants for silent reading; 77 studies and 5,965 participants for reading aloud.

How to use it

  1. Paste or type into the box

    Counts update as you type. On a phone the word count, character count and reading time stay pinned above the box while you scroll.

  2. Set the reading and speaking rates

    Leave 238 and 183 for a general audience, or time yourself reading a page aloud and enter your own rate before planning a talk.

  3. Check the limit you are writing for

    Find the platform in the limits list. The bar says how many characters are left or over, in that platform’s own unit.

  4. Look at the longest sentence

    It is shown in full beside the average sentence length, which is usually the quickest edit to make long prose easier to read.

Worked examples

Real input, real figures.

A five-minute talk

You give it

A script of 915 words, speaking rate left at 183 wpm.

You get

Speaking time 5 min: 915 ÷ 183 = 5.0 minutes. At a slower 150 wpm the same script runs 6 min 6 sec, so cut about 165 words to stay at five minutes.

A text message with an apostrophe

You give it

A 120-character reminder typed on a phone, with “we’ll” in it.

You get

The curly apostrophe forces UCS-2, where one message holds 70 characters and each part of a longer one 67, so it goes as 2 messages. Swap ’ for a straight ' and the same text fits in 1 of 160.

Against a typical online word counter

AspectThis toolA typical counter
Chinese, Japanese, ThaiSegmented into real wordsOften 1 word per line
“well-known”, “don’t”, a URL1 word eachVaries: 2, 2 and 5 are common
Emoji like 👍🏽1 characterOften 4 (UTF-16 code units)
Platform limitsEach platform’s own rule, with its sourcePlain character count
Reading rate238 wpm, cited and adjustableUnstated, or 200–300 without a source
Where your text goesNowhere; kept in this browserOften sent to a server

Guide

What it will not do

  • No readability score. The formulas estimate syllables with English-only rules and were fitted on school texts, so a grade level here would be a guess dressed as a measurement. Average and longest sentence length are shown instead, because they are counts, not estimates.
  • Japanese and Chinese are counted in words as the browser’s dictionary splits them, which can differ slightly between browsers. Where a character count is the convention, as it usually is for these languages, use the character figure.
  • Platforms that do not publish how they count emoji are counted in visible characters. Leave a small margin near those limits.

How a word is counted here

Every counter answers this differently, so the rules are written down. Each row is tested.

TextWords hereWhy
don’t, can't2An apostrophe inside a word does not split it (Unicode UAX #29).
well-known e-mail2A single hyphen with no spaces joins its two halves.
3.14, 1,000,000, 10:303A point or comma inside a number, or a colon between digits, keeps it whole.
https://dsyn.studio/word-counter1A link or email address is one word.
我喜欢学习中文4The browser’s dictionary splits it into 我 · 喜欢 · 学习 · 中文.
I ❤️ it 👍🏽2Emoji are characters, not words.
word—word2An em dash separates, with or without spaces.

Characters are what you see as one character, so 👍🏽 and 🇬🇧 count 1 each. JavaScript’s ownlength says 4 for both, which is why some counters overshoot on emoji. Line breaks are not characters. A sentence ends where the Unicode sentence rules say it does, with titles such as “Mr.” and “Dr.” rejoined, and a paragraph is any line with something on it.

Where every limit comes from

Only limits a platform publishes itself are shown. A number found only on third-party blogs is left out.

LimitCounted asSource
X postWeighted: Chinese, Japanese, Korean and emoji count 2; a link counts 23.twitter-text config v3 (X’s own counting library)
X post, PremiumThe same weighting, on the longer-post limit.X Help Center: X Premium, longer posts
Bluesky postGraphemes, so an emoji counts 1 (and at most 3,000 bytes).AT Protocol lexicon app.bsky.feed.post: maxGraphemes 300
Threads postCharacters, with each emoji counted as its UTF-8 bytes (usually 4).Threads API: text posts are limited to 500 characters
Instagram captionCharacters as you see them. The platform does not publish how it counts emoji, so leave a margin near the limit.Instagram Graph API: caption maximum 2200 characters
LinkedIn postCharacters as you see them. The platform does not publish how it counts emoji, so leave a margin near the limit.LinkedIn Help: the character limit for a post is 3,000
SMS160 per message in the GSM alphabet, 70 when any character is outside it; long messages split into parts of 153 or 67.3GPP TS 23.038 (alphabet) and TS 23.040 (140-octet user data)
YouTube titleCharacters as you see them. The platform does not publish how it counts emoji, so leave a margin near the limit.YouTube Help: video titles have a limit of 100 characters
YouTube descriptionCharacters as you see them. The platform does not publish how it counts emoji, so leave a margin near the limit.YouTube Help: descriptions have a limit of 5,000 characters
Google Ads headlineChinese, Japanese and Korean characters count 2.Google Ads Help: responsive search ad headlines support up to 30 characters
Google Ads descriptionChinese, Japanese and Korean characters count 2.Google Ads Help: descriptions support up to 90 characters each

Two common figures are missing on purpose. Google sets no limit on a meta description. It cuts the snippet to fit the screen width, so the 155–160 characters often quoted is a rule of thumb, not a rule. And the 220-character LinkedIn headline quoted across the web does not appear on any LinkedIn Help page we could find, so it is not stated here as fact.

Why there is no readability score

Readability formulas such as Flesch–Kincaid need a syllable count, and English has no rule that gets syllables right without a dictionary: “created” has three and “cared” one. The formulas were also fitted on school texts, and they mean nothing for the other scripts this counter handles. The two inputs that matter, sentence length and the longest sentence, are counts rather than estimates, so those are shown instead. The link text auditor refuses a score for the same reason, on even shorter text.

Questions people ask

Straight answers. Each one links directly: copy the address after opening it.

How many words is a 5 minute speech?

About 915 words at 183 words a minute, the average rate for adults reading aloud in Brysbaert’s 2019 meta-analysis. Many speakers on stage are slower, near 130–150 words a minute, which puts five minutes at 650–750 words. Time yourself reading one page aloud and enter that rate: it is the only figure that fits your talk.

Does word count include numbers?

Yes. A number is a word here, so “2026” counts as 1, and so do “3.14” and “1,000,000”, because the comma and point sit inside the number. A time such as “10:30” is also 1. Numbers are left out of the most-used words list, because they are not the words anyone is overusing.

Are hyphenated words counted as one word?

Yes. “Well-known” and “e-mail” count once, as a reader counts them. The Unicode rules the browser uses split them at the hyphen, so the tool joins them back. A dash with spaces round it (“word – word”) is not a join, so that is 2 words.

Why does my word count differ from Word or Google Docs?

Because each counter has its own rule. Word processors count runs of characters between spaces, so “word—word” with an em dash and no spaces is 1 word there and 2 here, and a line of Chinese is 1 word there and several here. Links, emoji and symbols on their own also differ. The rules used here are listed on this page.

How many characters is a tweet?

280 on a standard account and 25,000 with X Premium, but X counts a weighted length: Chinese, Japanese, Korean and emoji count 2 each and every link counts 23, whatever its length. The X bar in the limits list uses that rule, from X’s own twitter-text library.