What Character Frequency Analysis Can Tell You
Counting how often each character appears in a block of text sounds like a curiosity and is a genuinely useful diagnostic. The Character Frequency Counter gives you the tally.
What normal English looks like
In ordinary English prose, the space is the most common character by a wide margin, followed by e, t, a, o, i, n, s, h and r. The letter e alone is around twelve per cent of the letters. At the other end, q, j, x and z together account for under one per cent.
Knowing roughly what normal looks like is what makes the abnormal visible, and that is the whole value of the exercise.
What the counts expose
A simple substitution cipher. Replace every letter with another letter and the frequencies travel with them, so the commonest character in the ciphertext is probably standing in for e. This is the classic application and it is how such ciphers were broken long before computers.
Padding and filler. A suspiciously high count for one character usually means a run of it - a line of dashes, a column padded with zeros, a block of underscores someone used as a separator.
Encoding faults. Finding a question mark where no question is asked, or a scattering of odd accented characters in English text, is the signature of text that has been through the wrong character encoding. A high count of one unexpected symbol is often the replacement character software substitutes when it cannot represent what it was given.
Invisible characters. A tally shows you what a read-through cannot: non-breaking spaces, tabs mixed with spaces, zero-width characters pasted in from a web page. If a count does not add up to the length you expected, something invisible is in there, and the Whitespace Trimmer will take it out.
Whether a password is random. A genuinely random string has a flat distribution. One where a few characters dominate was produced by a human, whatever it looks like - which is an argument for the Password Generator rather than inventing one.
Reading it honestly
Frequency analysis needs volume. A tweet tells you nothing - with two hundred characters, which letter comes out on top is close to chance. A few thousand characters start to look like the language. Below that, treat any pattern as noise.
It also only ever tells you what is there, not why. A high count is a question to investigate rather than an answer.
Common questions
Does it count spaces and punctuation?
Yes, and they are usually the most revealing part - particularly the space, whose count is a good estimate of the word count.
Does case matter?
Usually it is worth normalising with the Case Converter first, so A and a are counted together, unless you are specifically looking at capitalisation.
Can it identify the language?
Roughly, on enough text. Different languages have distinctly different letter distributions - a high count for w and z suggests Polish, for instance.
Is my text uploaded?
No. The counting happens in your browser.
Got text you want to look inside?