%22%2F%3E%3Ccircle%20cx%3D%2232%22%20cy%3D%2232%22%20r%3D%2223%22%20fill%3D%22%23fff%22%2F%3E%3Cg%20fill%3D%22none%22%20stroke%3D%22url(%23gu)%22%20stroke-width%3D%225%22%20stroke-linecap%3D%22round%22%3E%3Cpath%20d%3D%22M21.5%2024h21%22%2F%3E%3Cpath%20d%3D%22M21.5%2032h21%22%2F%3E%3Cpath%20d%3D%22M21.5%2040h11%22%2F%3E%3C%2Fg%3E%3C%2Fsvg%3E)
TextLabFree punctuation remover
Remove punctuation without breaking don't
Strip the marks out of a passage while the ones doing real work stay: the apostrophe in don't, the hyphen in well-known, the comma and point in 1,500.25. It is free, needs no signup and puts no limit on length. Each mark is judged by the characters either side of it rather than by which key typed it, so an apostrophe inside a word is treated differently from one being used as a quote — and the table underneath shows you every mark sorted by the job it was doing.
- 100% free
- No signup
- Contractions survive
- Compound words survive
- Every mark tallied
Every mark in your text, by the job it was doing
Paste something and each mark will be sorted by role — an apostrophe inside a word is counted apart from one being used as a quote.
How to remove punctuation from text
Three steps, and the count of what will go appears before you change a setting.
Paste the passage you want stripped back
Put the text in the upper box. Your clipboard also arrives by way of ⌘V, or Ctrl+V for Windows, pressed while the page is on screen. The headline number starts moving straight away: it counts the marks that will be removed against the ones the exceptions below are protecting, so you can see the trade before you make it.
Tick what should survive
Six exceptions, each phrased by what it protects rather than by which character it matches. The first three are on by default because they are almost always right — the apostrophe inside a word, the hyphen joining two words, and the comma and point inside a number. Turn on sentence endings and commas as well if you want readable prose with the decorative marks gone.
Check the tally, then copy
The table at the bottom sorts every mark in your text by the job it was doing, with a column for how many were removed and how many spared, so an apostrophe inside a word is counted separately from one being used as a quotation mark. Choose whether a removed mark leaves a space behind or closes the gap, then copy the bare words.
Technical specifications
| What counts as punctuation | Unicode's punctuation categories — around 800 characters including . , ! ? ; : ' " ( ) [ ] { } - – — / \ @ # % & * _ and the bullet |
|---|---|
| What is deliberately left alone | Symbols: currency signs, maths operators, comparison signs and arrows. Any found in your text are listed under the tool by name and count |
| How an apostrophe is judged | A letter on both sides makes it part of a word (don't, O'Brien); a closing one straight after an s is read as a possessive (dogs'); anywhere else it is a quotation mark |
| How a hyphen is judged | A letter or digit on both sides makes it a joiner (well-known, twenty-five); anything else makes it a dash. En and em dashes are always dashes |
| Roles counted separately | Eleven: sentence endings, commas, apostrophes inside words, quotation marks, joining hyphens, standalone dashes, brackets, colons and semicolons, slashes, separators inside numbers, and everything else |
| Replacement | The gap closes by default; switch to a space and this/that becomes two words, with the resulting double spaces tidied up automatically |
| Speed and limit | No cap. A 10,000-word chapter is scanned character by character and re-tallied in about 4 ms, so the table keeps up with typing |
| Price and privacy | Free, no signup; the passage is processed in this tab and never uploaded |
Frequently asked questions
Will this turn don't into dont?
Not unless you ask it to. The apostrophe exception is on by default and it protects any apostrophe with a letter on both sides, so don't, it's, we're and O'Brien all come through whole while the apostrophes being used as quotation marks are removed. Blanket punctuation stripping is what produces dont and cant, and it is the single most common complaint about tools that do this job — which is why the distinction is made by position in the word rather than by the character itself.
What happens to hyphens in compound words?
A hyphen with a word character on each side is treated as part of the compound and kept, so well-known, mother-in-law and twenty-five survive intact. A hyphen with a space beside it, or one standing alone as a bullet, is a dash rather than a joiner and is removed. Em and en dashes are always removed regardless, because they are punctuation between phrases rather than glue inside a word. The tally counts the two roles on separate rows.
Does it remove the dollar sign, the plus sign and the equals sign?
No, and this surprises people. Unicode sorts characters into categories, and it files currency signs, maths operators and comparison signs as symbols rather than as punctuation — so $, €, +, =, <, > and ~ are all outside the definition this page works to. A note under the tool lists any it left behind so you are never left wondering. If you want those gone as well, that is a broader operation than removing punctuation.
What about the comma and the full stop inside 1,500.25?
They are protected by their own exception, which is on by default, and a comma or period is spared whenever it has a digit on both sides. Without that rule a price list becomes unreadable — 1,500.25 collapses into 150025, and every date, version number and decimal in the document goes the same way. Turn the exception off if you deliberately want the digits run together, which is occasionally what a data-entry field wants.
Why did this/that become thisthat?
Because a removed mark closes the gap by default rather than leaving a space. That is right for an apostrophe and wrong for a slash, and there is no way to guess which you meant, so the choice is a tick box: leave a space where each mark was, and this/that becomes this that with two separate words. The follow-up tidy is on as well, so you will not be left with double spaces wherever two marks sat side by side.
Are curly and straight quotes both handled?
Yes — the twelve quotation characters in common use are all recognised, including the curly pairs Word inserts, the low German opening quote, and the French angle marks. They are counted together as quotation marks in the tally. The one genuinely ambiguous case is an apostrophe at the edge of a word: in 'tis and '90s it is part of the word, but it is written identically to an opening single quote, so it is treated as a quote and removed. A closing quote straight after a word ending in s goes the other way and is kept, because dogs' bowls is far more common than a quotation ending in s.
Why would anyone want punctuation removed at all?
Three reasons, and they want slightly different settings. Word games and puzzles need bare letters, so everything goes. Counting and text analysis needs the marks out so that word. and word are not counted as two different words, but wants contractions kept whole. Data entry into a field that rejects punctuation needs the marks replaced by spaces rather than deleted, so the words on either side do not fuse. The exceptions above cover all three without needing three different tools.
The same apostrophe doing two different jobs
The character that ends a quotation and the character inside don't are the same character. So are the hyphen in well-known and the dash in the meeting — finally — ended. Any tool that works by matching characters therefore has to choose between destroying contractions and leaving quotation marks behind, and most of them destroy the contractions: you ask for punctuation to be removed and you are handed dont, wellknown and OBriens, which is a worse text than the one you started with. The only way out is to stop looking at the character and start looking at its neighbours. An apostrophe with a letter on both sides is doing morphological work inside a word; the identical character with a space on one side is punctuating a sentence. That single test is the argument for this page existing separately from a tick box that says “remove punctuation”.
English makes the test harder than it sounds, and it is worth knowing where the remaining ambiguity sits. A plural possessive puts the apostrophe at the end of the word, so the dogs' bowls has a letter on one side only and would fail a naive both-sides rule; it is protected here by a second rule that reads an apostrophe immediately after an s as possessive. The cost of that rule is that a quotation ending in an s-word keeps its closing mark. Going the other way, 'tis and '90s begin with an apostrophe that is genuinely part of the word but is written identically to an opening quote, and nothing short of understanding the sentence can tell them apart — so they lose it. Two edge cases in a rule that handles every contraction correctly is a trade worth making, and stating them is more honest than pretending the problem is solved.
The other surprise is where punctuation ends. Unicode splits the characters that are not letters or digits into punctuation and symbols, and the split is not where intuition puts it: the percent sign, the ampersand and the at sign are punctuation, while the dollar sign, the plus sign and the greater-than sign are symbols. This page works to the punctuation category and leaves symbols entirely alone, listing any it found so the omission is never silent — if you need those gone too, that is the job of removing special characters, which draws the line in a different place. And if the reason for stripping the marks is analysis rather than presentation, the word frequency counter already ignores punctuation when it counts, so word. and word land in the same bucket without any of this.
Where the stripping happens
Character by character, inside this browser tab, on code that arrived with the page. The passage you paste is not uploaded and not retained, and neither is the tally the tool builds from it — both are rebuilt from scratch on every keystroke and gone when you close the tab. There is no account, so there is no history to clear.