Text Frequencies

Analyse pasted text or a UTF-8 file. Letters are Unicode graphemes containing a letter, so composed accents count together; emoji, digits and standalone marks are excluded from the letter total. Words use Intl.Segmenter word-like boundaries, including numeric tokens and language-specific scripts without spaces. Punctuation separates words instead of joining them.

NFC merges canonically equivalent text; NFKC additionally converts compatibility forms such as ligatures. None keeps the original forms. Ignore case uses lowercasing in the selected text language, not full linguistic case folding. Segmentation depends on the browser’s ICU version and selected text language.

Percentages use the full letter or word total. Search, sorting and pages do not change the denominator. Tables show 100 entries per page; CSV exports all entries in the current view, independent of search/page. The optional chart shows the top 30 letters. No silent 500-word truncation.

Computation runs locally in a cancellable worker. Changes invalidate old results, file loading does not analyse automatically, and no text/results are saved. Limit: 1000000 UTF-16 code units, 20000 distinct entries per type, files 4 MB. Older browsers without Intl.Segmenter are unsupported.

Long entries are shortened only in the display; CSV retains the complete text.