Character Frequency

The text is broken into items, counted into a tally, and ranked. What counts as an item is the whole question, which is why there are three modes rather than one.

Words matches runs of letters, digits and apostrophes, so don't stays one word and punctuation is ignored. Everything is lowercased first, so The and the are counted together — which is almost always what you want, and occasionally is not.

Letters only counts alphabetic characters and ignores spaces, digits and punctuation. This is the mode for the classic frequency analysis: in English, E is reliably the most common letter at around 12%, followed by T, A and O.

Every character counts literally everything except line breaks, spaces included and labelled. Spaces normally top that ranking at roughly 15–18% of a typical English text, which surprises people the first time they see it.

Percentages are of the total items counted in that mode, so they always sum to 100 within rounding. The top forty are shown, with the total number of distinct items reported above.

You have written a 900-word article and suspect you have been leaning on one word too heavily.

Paste it and leave the mode on Words. The top of the ranking is predictable — the, a, to, of — and tells you nothing. Scroll past the function words and the useful part appears: essentially appearing 9 times in 900 words is a tic, and now you can see it.

This is also a keyword check. If you wrote an article about compressing images and the phrase does not appear in the top thirty, the page may not be about what you think it is about, at least as far as a search engine can tell.

Now switch to Letters only on a decent chunk of English and compare against the known distribution: E around 12%, T around 9%, A around 8%. A text that deviates sharply is either not English, is very short, or is unusual on purpose — the novel Gadsby was written without the letter E at all, and this mode would show that instantly.

Every character is the mode for spotting the invisible. A count of non-breaking spaces or an unexpected tally of curly versus straight apostrophes tells you the text came from somewhere with its own typography — which matters if it is going into a system that only handles ASCII. Cleaning that up is a find and replace away once you know it is there.