How it works : Count words
“How many characters?” has more than one right answer, and that is the first thing this tool shows you: one emoji is two code units, one code point, and a single visible sign. All three counts are shown separately, with what each of them is for.
A word counter ought to be the simplest tool on the web. In practice, most of them give a character figure and stay quiet about the question that matters: which of the three possible figures they gave. One emoji takes two code units in a computer's memory, one code point in the Unicode catalogue, and one visible sign on screen. A family emoji takes seven and is still one sign. Depending on the count, your text either fits or does not fit the field you have to paste it into.
So this tool shows all three, and says what each is for: forms and databases nearly always count in code units, a human being counts visible signs. It also gives words, sentences, paragraphs, distinct words, average sentence length, and the longest sentence in your text, which is the single most useful thing to know when you rewrite.
Finally it computes a readability index, and it shows the formula it used with its original publication AND its limit. These indices measure two things only: how long your sentences are and how many syllables your words have. They completely ignore the coherence of your text, its structure, and what your reader already knows. That criticism has been published and recognised since 1981, and a tool that hands you a score without mentioning it lets you believe it measured quality.
How it works
- Paste your text. It does not leave your device, which matters for a text that is not published yet.
- Pick the language of the text: the readability formula is not the same in English and in French, and applying one to the other means nothing.
- Read the three character counts, then the readability and its limit, then the longest sentence. Fixing one over long sentence often improves a text more than ten minutes of tinkering.
Formats accepted
The tool works on plain text. Formatting pasted from a word processor is not counted: only the words are. To strip a text of what copy and paste dragged in with it, the “Clean up pasted formatting” tool in the same corner is built for that.
Text, typed or pasted
Up to 200,000 characters, roughly a 350 page book. Beyond that, counting syllables takes a visible moment in the tab.
Its limits
Syllable counting is APPROXIMATE, and for French no reference method exists
Counting syllables exactly takes a pronunciation dictionary, and pronunciation depends on context. We use a vowel group heuristic, dropping the silent final “e” and endings in “es” for French. For English a published heuristic has existed since 1986; for French our research found NONE that serves as a reference. The consequence: the readability index is an order of magnitude, never a measurement to one decimal place.
The constant in the French formula is given as 207 by some sources and 209 by others, and we show both
The French adaptation of the Flesch index was published in 1958 by Kandel and Moles in the Cahiers d'études de radio-télévision. We could not get to the original text, and the secondary sources contradict one another on that one constant. Rather than choosing in silence, the tool computes both and says so. The gap is two points on a scale of a hundred: it does not change the verdict, but it changes what we can honestly claim.
A readability index does NOT measure the quality of a text
It only looks at sentence length and syllables per word. A perfectly incoherent text made of short sentences and simple words scores excellently. The criticism is published and recognised: these formulas were built to grade school textbooks, and they ignore the structure of a text, its navigation aids and the reader's knowledge. Use the index as a signal about your long sentences, not as a mark.
Counting visible signs takes a function older browsers do not have
Cutting a text into visible signs means knowing the Unicode rules that group a character with its accents, a flag, or a family of emoji. The native function that does it arrived in Chrome in 2020, in Safari in 2021 and in Firefox in 2024. When it is missing, the tool does not guess: it writes that the count is not available on this browser, which is better than a wrong figure.
Splitting into sentences is a heuristic, not a certainty
A full stop separates two sentences, but it also ends an abbreviation, a decimal number, a web address. The tool requires an ending mark followed by a space or the end of the text, and it never cuts between two digits. Even so, “Mr. Smith arrived.” will be counted as two sentences: that is the price of a rule you can explain in one line rather than a list of abbreviations per language.
Reading times are conventions, and we write them down
We use 200 words a minute for silent reading and 130 for reading aloud. Those are the usual orders of magnitude, not measurements made on you: technical reading goes well below, a fast reader well above. We show the convention next to the figure so you can correct it in your head.
Questions about this tool
Is my text sent anywhere?
No, and that matters especially here: what people paste into a word counter is a dissertation, a cover letter, an article not yet published. Check it in thirty seconds: open the inspector with the F12 key, the “Network” tab, clear the list, then paste your text and count it. Not one line appears. Cut your connection and do it again: the tool works the same.
Why three different figures for characters?
Because the question has three right answers, and the gap is not trivial. A simple emoji takes two code units in a computer's memory, one Unicode code point, and one sign on screen. A family emoji takes up to seven code units and is still one sign. Forms and databases nearly always count in code units: if your message has to fit a field of 280, that is the figure that decides. A human proofreader counts visible signs.
Can I trust the readability index?
As a signal, yes. As a mark, no. It only measures how long your sentences are and how many syllables your words have: a perfectly incoherent text made of short words scores excellently. These formulas date from the 1940s and were built to grade school textbooks; their misuse has been criticised in the scholarly literature since 1981. The most useful thing on this page is probably not the index, it is the longest sentence in your text, shown at the bottom.
Why do you show two constants for the French formula?
Because we do not know which is right, and saying so is better than choosing in silence. The French adaptation of the Flesch index was published in 1958 by Kandel and Moles. We could not get to the original article, and the sources that cite it give the constant as 207 in some places and 209 in others. The gap is two points out of a hundred: it does not change the verdict, but it changes what we can honestly claim.
Does the word count match my word processor?
Nearly always, within a few words. The gaps come from the edge cases: a number written with a space, a word hyphenated at a line break, a web address. We count as one word any run of letters or digits, including its internal hyphens and apostrophes: “porte-monnaie” counts as one, “don't” as one. If your editor counts two, it is splitting on hyphens.
Is there a size limit?
Two hundred thousand characters, roughly a three hundred and fifty page book. The limit is not there to spare a server, since there is none: beyond that, counting syllables takes a visible moment in the tab, and a tool that looks frozen is a bad tool. Split it by chapter if you go over.
“How many characters?” has more than one right answer, and that is the first thing this tool shows you: one emoji is two code units, one code point, and a single visible sign. All three counts are shown separately, with what each of them is for.
VerifyThe proof, step by stepThe network inspector check, explained one gesture at a time, with screenshots.
The cornerThe text and data cornerCount, compare, format, clean. The small jobs that cost an hour when you do not have them at hand.