Word & Character Counter

How the word counter works

The counter reads your text locally and runs four small passes over it. First it splits the text into words. Where the browser exposes Intl.Segmenter — Chrome, Edge, Safari and Firefox all ship it today — the counter uses it with granularity: "word", the same Unicode algorithm that powers find-as-you-type in your operating system. Each segment carries an isWordLike flag, so punctuation, emoji and stray symbols are skipped while genuine words and numbers are kept. In a browser without that API the counter falls back to a Unicode-aware regular expression, which gives the same answer for ordinary Western prose.

The remaining passes are deliberately simple and deterministic. Sentences are counted at full stops, question marks, exclamation marks and ellipses; paragraphs are counted as blocks separated by at least one blank line; unique words are counted on a case-folded form of each word, so Word and word count once. Average word length is the mean number of characters in a word. Reading time divides the word count by 200 words per minute and speaking time by 130 words per minute.

The frequency table groups words by their lower-cased form, drops a short English stopword list unless you re-include it, and sorts by count. Percentages are shares of all words in the text, including the stopwords that the table itself hides, so the column always adds up against the same denominator.

Accuracy and limits — read this before you quote a number

Word counting is only well defined for languages that put spaces between words. That covers English and most European languages, but not Chinese, Japanese or Thai, which write without spaces. For those scripts a correct count needs a dictionary-based segmenter, and browsers do not guarantee one: Intl.Segmenter may return a whole sentence as a single segment, or split it at arbitrary points. The regex fallback behaves similarly, because its word-boundary rules have nothing to latch onto. Treat any count for a no-space script as approximate and verify it in a tool that ships a proper dictionary before you publish the figure or bill against it.

• Hyphenated and apostrophe words. well-known and don't are single words in the browser's Unicode rules, but some editorial house styles count them as two. If your style guide differs, expect a gap of a few percent.
• Abbreviations. A sentence counter that keys on full stops reads e.g. and Mr. as sentence ends. That is why the sentence figure can run slightly high on academic or legal text.
• Numbers and symbols. Standalone figures such as 2025 count as words; a lone & or + does not. Maths-heavy text therefore reads as having fewer words than a human would say it has.
• Characters are counted in UTF-16 code units. An emoji outside the basic plane — most of them — counts as two. The "no spaces" figure removes every Unicode whitespace character, not just the ordinary space.
• Reading and speaking speeds are estimates. 200 and 130 words per minute are widely used conventions, but a dense technical passage can drop below 150 wpm while a rehearsed script can exceed 160. Use the numbers to compare drafts, not to promise a duration.
• Stopwords are English. The filter list contains English function words only. For another language, turn the toggle on and read the raw distribution instead.

How to use the counter

1. Paste or type your text. Click into the box and start writing, or paste from anywhere. Every statistic updates as you type; there is nothing to submit.
2. Read the panel. Words, characters with and without spaces, sentences, paragraphs, unique words, average word length and the two time estimates all sit in one grid, so you can compare them at a glance.
3. Inspect the vocabulary. The table below the grid lists the ten most frequent words with a count and a percentage. Tick Include common stopwords to see the raw distribution, or leave it off to see which content words dominate.
4. Copy the summary. Copy summary puts a plain-text report — every statistic plus the top words — on your clipboard, ready to paste into an email, a brief or a spreadsheet.
5. Start over. Clear empties the box and resets every figure to zero without reloading the page.

A practical tip: for a length-limited piece, write first and count afterwards. Trimming to hit a count is easier than padding, and the unique-word figure is a fast signal that you are repeating yourself — a high word count with a low unique count usually means the same point has been made twice.