Word Counter With Reading Time
Count words, characters, and sentences and estimate reading time for any text as you type.
Your draft stays on your device
Nothing to delete afterwards
Safe for embargoed work
Works in every modern browser
How it works
- 1
Paste or type your text
A paragraph, an article, an essay, or a full manuscript.
- 2
Read the counts
Every figure updates live, so you can see the effect of an edit immediately.
- 3
Trim or expand to fit
Work against the limit you were given without leaving the page.
Why use this tool
Words, characters, sentences, paragraphs
The four figures editors and briefs actually ask for, updated as you type.
Characters with and without spaces
Two different numbers that different limits mean, shown side by side rather than one at a time.
Counts what a reader sees
A character with a combining accent counts as one, not two, which naive counters get wrong.
Reading time estimate
Based on a stated words-per-minute figure you can see, not a hidden constant.
Nothing is uploaded
The text is measured in your browser — an unpublished draft never reaches a server.
Free, no account
No signup, no watermark, no cap on how much you count.
What counts as a word depends on the language
Almost every word counter works the same way: split the text on whitespace and count the pieces. For English that is close enough to right, and for a great many languages it is simply wrong. Chinese, Japanese, and Thai do not put spaces between words at all, so a five-hundred-word Chinese passage split on whitespace returns a count of one. Counting those languages properly requires segmentation — a dictionary-driven guess at where one word ends and the next begins — and even native tools disagree about the result. This is why a Japanese publisher specifies a character count rather than a word count, and why a translation quote for those languages is priced per character. If your text is not written with spaces between words, the word figure is not a measurement, and the character figure is the one to use.
The same content produces different counts in different languages
Even among languages that do use spaces, a word is not a constant unit, which matters whenever a limit crosses a translation. Arabic attaches its conjunctions, articles, and prepositions to the following word — و, ال, بـ, لـ all fasten on — so a sentence that takes eight words in English can take six in Arabic while saying exactly the same thing. Turkish does it through agglutination, packing into one word what English needs four for. German runs the other way with compounds, though its overall count sits close to English, while Spanish and French typically run about ten percent longer than English for the same meaning. The practical upshot is that a nine-hundred-word English brief translated faithfully will not be nine hundred words, and holding the translator to that number produces padding or cuts. Set limits per language, or set them in characters.
A character is not one thing either
Character counts look like the simple, unambiguous alternative, and they are not, because "character" has at least three defensible meanings. The letter é can be stored as a single code point or as a plain e followed by a combining accent — visually identical, one grapheme to a reader, but one or two units depending on how you count. Emoji make it stark: a flag is two code points, an emoji with a skin-tone modifier is two, and a family emoji can be seven joined together, so the same visible symbol counts as 1, 2, or 7 depending on the method. This is why a message that looks like 279 characters gets rejected by a field with a 280 limit. Counting graphemes — what a reader would call a character — is the behaviour that matches human expectation, and it is what this tool reports.
Which number your limit actually means
Before trimming, it is worth knowing which figure the limit refers to, because the same text has several. Characters with spaces and characters without differ by roughly a sixth in ordinary prose, and style guides use both without always saying which. Academic word limits usually exclude the bibliography and often the footnotes and abstract, so the figure in your editor is not the figure being assessed. SMS counts segments of 160 characters, but only for the basic alphabet — one accented letter or emoji switches the whole message to a 70-character encoding, which is why a message can suddenly cost three segments instead of one. Meta descriptions are truncated by pixel width rather than character count, so a description of capital letters is cut shorter than one of lowercase. Ask which count before you cut.
Reading time is an estimate, and a shakier one than it looks
The familiar figure is 200 to 250 words per minute for silent reading of ordinary English prose, and every reading-time badge you have seen is that number divided into a word count. It is a reasonable average and a poor prediction for any specific case. Technical writing with unfamiliar terms is read far slower; a list is scanned rather than read; anything with code or equations breaks the model entirely, because readers stop and re-read. And the rate is language-specific — reading speeds differ measurably between scripts, so applying an English figure to Arabic or Chinese text compounds the word-counting problem already described. Treat a reading-time estimate as a rough signal of length for your audience, not as a measurement, and be sceptical of it on anything that is not continuous prose.
Why counting locally matters for this kind of text
Think about what is actually in the box when someone counts words. It is a draft — an article before publication, a dissertation chapter, a cover letter, a legal submission, a manuscript, a press release under embargo. By definition it is work that has not been released yet, and often work whose early release would cause real harm: a paper that must stay unpublished before review, a client document under confidentiality, an announcement with a date attached. Pasting it into a hosted counter sends the entire unreleased text to a third party in exchange for a number. This tool measures the text in the page and transmits nothing, which you can confirm in the Network tab of your developer tools. For a task performed on unpublished writing, that is the correct default.
Common mistakes to avoid
- Counting words in Chinese, Japanese, or Thai. Those languages do not separate words with spaces, so a whitespace split returns a meaningless number — use the character count instead.
- Applying one word limit across translations. Arabic and Turkish express the same content in fewer words and Spanish and French in more, so a fixed number forces padding or cuts; set limits per language or in characters.
- Assuming every visible character counts as one. An accented letter can be two code points and a family emoji seven, which is how a message that looks like 279 characters fails a 280-character field.
- Trimming before checking which count the limit means. Characters with and without spaces differ by about a sixth, and academic limits usually exclude the bibliography and often the notes.
- Pasting an unpublished draft into a hosted counter. You are sending an entire embargoed or confidential document to a third party in exchange for a number; count locally.
How it compares
| Aspect | This tool | Online counters | A word processor |
|---|---|---|---|
| Content sent to a server | Never | Usually yes | Depends on the app |
| Grapheme-aware characters | Yes | Rarely | Varies |
| Both character counts shown | Yes | Sometimes | Yes |
| Works without opening a document | Yes | Yes | No |
| Account or signup | Not needed | Often required | Often required |
| Price | Free | Free / paid tiers | Licence or subscription |
Features
Live counting
Every figure updates as you type, so you can edit against a limit rather than checking afterwards.
Both character counts
With and without spaces, because form fields and style guides disagree about which they mean.
Grapheme-aware character count
Accented letters and emoji count as what a reader sees rather than as their underlying code points.
Sentence and paragraph counts
Useful when a brief specifies structure rather than length.
Handles long documents
Whole manuscripts count without an upload or a size cap.
Arabic and RTL text
Right-to-left text is counted and displayed correctly, not reversed or mangled.
Nothing to install
No word processor, no runtime, no dependencies — it runs on the web page.
Eight interface languages
The full interface in eight languages, including right-to-left Arabic.
Secure by default
Served over HTTPS, with no content tracking and no third-party upload.
Who uses it
Writers
Working an article to a commissioned length without switching to a word processor.
Students
Checking an essay against a limit, knowing which parts the limit excludes.
Translators
Comparing source and target length when a rate is quoted per word.
Marketers
Fitting a meta description or an ad field before the platform truncates it.
Frequently Asked Questions
No. Everything runs locally in your browser — your text is never uploaded, stored, or shared.
Yes — completely free, with no account and no limits.