HOW IT WORKS
How N-grams are counted
The analyzer tokenizes text into normalized words, then builds overlapping windows of size n. For n=2, every consecutive pair becomes one row; for n=3, every consecutive triple, and so on.
(phrase count ÷ total n-gram windows) × 100(phrase count ÷ total n-gram windows) × 1,000Window count is the denominator in every row, so a page with repeated high-frequency function words will have lower per-1,000 values as windows increase.
PRACTICAL SETTINGS
Choose n and filtering according to your analysis goal
N-gram size
Use n=1 for lexical repetition, n=2 for short phrases, and n=3+ for more specific linguistic patterns.
Stop words
Turn off stop words to focus on content-bearing phrase fragments and reduce repeated filler phrases.
Minimum count filter
Filter out one-off windows to keep long outputs usable while preserving full exports.
Local vs URL mode
Text mode runs immediately in-browser; URL mode fetches readable webpage text and applies the same calculations.
LIMITATIONS
N-grams are diagnostic, not semantic
An n-gram table shows surface repetition, not intent, meaning, or topical coverage. Use it with unigram and bigram comparatives to validate editorial patterns, and review edits manually before publishing.