Analyze distinct words locally with total words, unique words, repetition count, and a frequency list.
Counts run locally in your browser. When available, word, sentence, and grapheme segmentation uses the browser’s Unicode-aware Intl.Segmenter; fallback rules are used on older browsers.
Separate hapax words from repeated terms and expose the vocabulary distribution.
| Word | Count | Share |
|---|
Start with the broad counter that matches the question, then move to focused frequency, reading-time or syllable analysis. Specialist readability, repetition and writing-goal tools remain available from the full hub.
The result distinguishes Unicode graphemes, code points, word/sentence boundaries, line endings or normalization where those details change the meaning of the count. Browser-local analysis stays live without treating approximate language heuristics as exact facts.
Case-sensitive counting treats Word and word as different tokens, while many text-analysis tasks want them combined. Choose the setting that matches the question and document it when comparing counts across files or revisions.
Hyphens, apostrophes, URLs and Unicode punctuation can change where one word ends and another begins. Inspect representative tokens from the source when the exact unique count matters instead of assuming every counter uses the same tokenization rules.
A longer document usually contains more unique words simply because it contains more words. If the goal is comparing lexical variety, pair the unique count with total word count or a normalized measure rather than ranking documents by raw unique words alone.
Lowercasing, trimming punctuation or stemming may combine tokens that were originally distinct. That can be useful for search, deduplication and vocabulary analysis, but it changes the question being answered. Keep the normalization policy explicit.
Use a short string such as red blue red BLUE and predict the result before running the tool. Under case-insensitive counting there should be two unique words; under case-sensitive counting there should be three. A tiny known sample is the fastest way to verify the chosen settings.
WebToolArc