Word and Letter Frequency Calculator

Letter frequency in English

E is the most common letter in English, turning up about 12.7% of the time. The rough order after that runs E T A O I N S H R D L C U M W F G Y P B V K J X Q Z, which is where the old printer’s nonsense phrase “etaoin shrdlu” comes from. Counting letters this way is much older than it looks: a 9th-century Arabic treatise on deciphering messages, written by the mathematician al-Kindi, is the first known description of the technique. The Morse code story often told alongside it is a separate episode, and the counting there was done by Alfred Vail in the 1830s, who tallied the type in a Morristown printer’s case so the commonest letters could get the shortest codes.

Word frequency

“The” is the most common English word, at roughly 7% of all running text. The top 100 words account for about half of everything written in English, and the top 1,000 cover somewhere near 80%. That lopsidedness is Zipf’s law: word frequency falls off roughly in proportion to 1 divided by the word’s rank.

Zipf’s law

Named after George Kingsley Zipf, who popularised it in the 1930s, though Jean-Baptiste Estoup had noticed the same pattern in French shorthand texts around 1916. The claim is simple: the most frequent word appears about twice as often as the second, three times as often as the third, and so on down the list. It holds across essentially every human language, and it holds surprisingly well on short samples too, which is why the chart here draws the predicted curve next to your actual counts.

Where this gets used

Frequency analysis is the standard first move against a substitution cipher. If the ciphertext’s commonest letter is Q, Q probably stands for E, and a few more matches usually crack the rest. Beyond codebreaking, editors use word frequency to catch a word they have leaned on too heavily, and the ratio of unique words to total words, shown here as vocabulary richness, is one of the oldest measures in authorship studies. Be careful with that one: it drops as the sample gets longer, because common words keep repeating while new ones run out. Comparing richness across two texts is only fair when they are about the same length.