Remove Punctuation

Punctuation helps readers but gets in the way of analysis. Count words in a text full of quotes and dashes and “word” and “word,” become different entries. Build a keyword list and stray brackets and slashes creep in. Feed text to a word cloud and punctuation shows up as clutter. This tool removes punctuation while keeping the parts of words and numbers that only look like punctuation.

Punctuation and symbols

Unicode sorts marks into categories. Punctuation includes the familiar . , ; : ! ? along with quotation marks of every language (“ ” ‘ ’ « » „ 「 」), dashes, brackets, slashes, the ellipsis, inverted ¿ and ¡, and the full-width marks used in Chinese and Japanese. Symbols are a separate category: currency signs, maths signs, arrows, ©, ™, degree signs and emoji. Punctuation is always removed; symbols only when you turn on “Also remove symbols”, because $ and % often carry meaning.

Words stay whole

An apostrophe between letters is part of a word: don’t, O’Brien, rock’n’roll. A hyphen between letters joins a compound: well-known, mother-in-law. Both are kept by default, while apostrophes used as quotation marks and dashes between words are removed. Turn the options off if your analysis should split such words.

Numbers keep their value

Points and commas between two digits are decimal points or thousands separators, and a colon between digits is part of a time, so 4.50, 1,250, 3.14 and 9:30 stay exactly as written. Without that rule, 4.50 would become 450. A full stop at the end of a sentence after a number is still removed, because it is not followed by a digit.

Keeping what you need

“Keep . , ! ? : ;” removes quotes, brackets and other marks but leaves sentence punctuation, which keeps text readable. The “Never remove” box takes any characters you want to protect, such as # and @ for social media text or & for company names. Removed marks can be replaced with nothing or with a space; replacing with a space prevents “red/green” from becoming “redgreen”.

See what was removed

The panel under the result lists every mark that was removed and how often, with its code point on hover. It is a quick way to discover unexpected characters such as curly quotes, non-standard dashes or invisible symbols in pasted text.

Uses

Students and researchers clean text before counting words or running a concordance. Marketers prepare keyword lists and tag clouds. Developers normalise user input before comparing it, and data teams clean survey answers before coding them. The text is processed in your browser and never uploaded.

A worked example

The sentence “Wait—what?!” It’s a well-known café, $4.50 (est. 1998). becomes Waitwhat It’s a well-known café $4.50 est 1998 with the default options. Choose a space as the replacement and “Wait—what” becomes “Wait what” instead, which is better for word counts. Turn on “Also remove symbols” and the dollar sign goes too, leaving 4.50.

Before and after analysis

Keep a copy of the original text. Punctuation carries meaning that analysis may later need, such as the difference between a question and a statement or the quotes that mark dialogue. A common workflow is to clean a copy for counting and keep the punctuated version for quoting, so any figure you report can be traced back to the sentence it came from.

Related tools

Remove emoji only with Remove Emojis, strip accents with Remove Accents, count the cleaned words with the Word Frequency Counter, or fix spacing and line breaks with the Text Cleaner.