Generate local unigram, bigram, trigram, 4-gram, or 5-gram frequency tables from segmented words.
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.
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.
By default n-grams are built inside detected sentences, so the last word of one sentence is not silently paired with the first word of the next. Case sensitivity and cross-sentence behavior are explicit controls.
Use the tool first, then apply these checks to verify inputs, interpret the result, and hand it off without displacing the primary workflow.
N-gram counts depend on how text is split and normalized. Case folding, punctuation removal, apostrophes, hyphens, Unicode normalization, and sentence boundaries can all change the resulting phrases. Keep the same settings when comparing two documents; otherwise a frequency change may reflect preprocessing rather than a real difference in language use.
Unigrams emphasize individual words, bigrams expose short associations, and larger n-grams capture more specific phrases but become sparse quickly. A high n can produce many one-off sequences, while n=1 loses phrase context. Run more than one n value when you need both broad vocabulary and repeated phrasing evidence.
A common phrase can dominate because of boilerplate, headings, navigation text, or a repeated template rather than because it is conceptually important. Inspect the source occurrences of high-frequency n-grams and consider stop words or section boundaries before using counts to make editorial, linguistic, or search conclusions.
Local analysis avoids intentionally uploading the pasted text, but you should still use appropriate handling rules for confidential material on the device. When exporting results, record n, case sensitivity, normalization choices, limit, and source version. That makes a later comparison reproducible instead of leaving only an unexplained frequency table.
WebToolArc