Count Words/Characters - Stats Tool

Words, Sentences, Paragraphs, Lines and Character Units

Measure word-like segments, sentence segments, paragraphs, lines and several character units without changing the draft. These are documented counting methods, not an exact match for every word processor or platform limit. Compare drafts under one counting method, see why code-unit length and visible clusters differ, or get an approximate reading time from a stated rate.

1. Paste text or choose Sample.
2. Read the labeled units and top keyword table.
3. Use Copy Stats for the complete summary, including keywords. On failure, select the summary manually.

How Each Number Is Counted

Intl.Segmenter uses locale en for word, sentence and grapheme boundaries. Word-like segments include Unicode text and numbers. Results depend on browser Unicode/ICU data; there is no universal linguistic accuracy promise. Paragraphs are nonempty blocks separated by blank lines. CRLF, LF and lone CR are recognized as record endings for line counts; a final ending creates a blank last line. Empty input has zero lines.

Keyword ranking uses lowercase word-like segments longer than two code points and a small English stop list. Map-based counts safely include names such as constructor. Percentages divide by all word-like segments, including excluded stop words, so displayed percentages need not total 100. Reading time is words divided by 200 and rounded up to minutes, not a measured reading speed.

One Text, Three Character Counts

For Hello. Unfinished ending, the en segmenter counts three words and two sentence segments, including the unfinished final fragment. An emoji outside a word is not a word. A supplementary emoji can occupy two UTF-16 units but one code point and one grapheme cluster.

Counts That Differ From Other Applications

The no-whitespace metric removes all JavaScript whitespace, including line breaks and tabs, not just spaces. UTF-16 count does not certify SMS segments or social-platform posting limits. Abbreviations, decimals and language-specific boundaries can differ from another application; use the required application's own rules for submission.

Input is limited to 1,000,000 UTF-16 units. Unsupported Intl.Segmenter browsers show an error rather than silently changing algorithms. The counting runs in your browser; the Privacy Policy describes how the site as a whole handles data.

Keep the original draft and record the counting unit when comparing results. MDN explains locale-sensitive Intl.Segmenter behavior.