Duplicate Word Checker

Find doubled words like "the the", words repeated too close together, phrases you reuse and sentences pasted twice, with every threshold measured from real edited prose rather than guessed.

Your writing

Text or a live page

Five problems wearing one name

Search for a duplicate word checker and you will find a dozen tools that do the same thing: count your words, list the ones that appear more than once, and give you a slider to hide short words. That is a frequency counter, and it answers a question almost nobody is asking. Of course “the” appears forty times. It is supposed to.

The thing people are usually looking for is the doubled word. You typed “the the”, your eye read it as “the”, and it is still there three drafts later. That is a different problem, and so are the three or four others that get filed under the same heading. This tool separates them, because each one has a different fix:

Doubled words. The same word twice in a row with only space between it. A typing slip, and the most embarrassing kind because it looks like nobody read the page.
Close repeats. The same word used again a few words later. Not an error. It just makes the sentence clunk.
Words you lean on. One word carrying too much of the whole piece, usually a filler you cannot hear yourself using.
Repeated phrases. Several words reused together, which is how a turn of phrase becomes a verbal tic.
Repeated sentences. The same sentence twice, word for word. Almost always a paste that went wrong.

The doubled word hides at the end of a line

This is the one worth having a tool for, because it defeats proofreading. When the first word ends a line and the second begins the next one, your eye treats the two as a single phrase and reads straight past it. Spellcheckers do not catch it, because both words are spelled correctly. It survives in printed books and in copy that four people signed off.

So a doubled word here means the same word twice with nothing but whitespace between them, and a line break is whitespace. The “nothing but” matters as much as the rest: “No, no” has a comma and is deliberate, and “I saw it. It was fine” has a full stop and is two sentences. Neither is reported.

Some doubled words are correct, and we checked which

“We had had enough” is correct past perfect. A tool that marks it as an error teaches you to click past its warnings, and then it misses the real one. Most checkers either flag everything or quietly suppress a list of exceptions somebody copied from a blog post.

We measured it instead. Across 1.56 million words of edited prose — eight books from Project Gutenberg — there are only seventeen distinct words that ever appear doubled, and the list is dominated by a handful of real constructions:

had had, past perfect, by far the most common
that that, conjunction followed by pronoun: “the fact that that happened”
you you, object then subject: “I told you you would win”
in in, particle then preposition: “they rushed in in a crowd”
her her and it it and there there, the same pattern again
is is, the double copula: “what it is is a mistake”
will will and can can, a noun followed by a modal: “the will will be read”
deliberate repetitions: very very, so so, long long, ha ha

Those are listed separately, with the construction named, so you can glance at each one rather than trust it or fight it. Everything else is reported as a slip, because across that entire corpus nothing else turned up. One more number from the same measurement: a doubled word occurs about once every 8,200 words of published writing. If your draft has four in a thousand words, they are not all grammar.

Then we ran the finished tool back over the same 1.6 million words to see how often it would cry wolf on prose that had already been edited and printed. It called two things a slip, which is one every 800,000 words, and neither of them is English: a French causative (devait faire faire) and a line of written dialect. That is the number worth knowing before you trust a proofreading tool, and it is the number nobody publishes.

What “too close together” actually means

Every style guide tells you not to repeat a word too close to itself. None of them says how close. So we measured that too, over the same corpus, looking only at content words because “the” and “of” repeat constantly in all good writing and always will.

Only 1.2% of repeats happen within five words of each other.
2.6% within ten words.
4.9% within twenty.
10.5% within fifty.

In other words, published prose almost never puts the same word twice in a breath, which is exactly why it jars when you do. Twenty-five words is the default here. Narrow it to ten if you only want the worst offenders; widen it to a hundred if you are polishing something that matters.

Different forms of a word count as the same word, so “process” and “processes” are matched, and both forms are shown so you can see what triggered it. The matching is deliberately shallow — it strips only endings it can be sure about, and only when enough of the word is left — because a stemmer that merges words a reader hears as different would invent repeats that are not there.

Phrases, reported at full length

Look for repeated four-word phrases inside “at the end of the day” and you will be told, three times, that you have repeated a four-word phrase. The useful answer is that you used one six-word phrase twice. So the search runs from the longest match downwards and drops a shorter phrase when every place it occurs already sits inside a longer one.

How many is too many? Measured the same way: a typical 1,000 words of edited prose repeats about four distinct four-word phrases. Eleven would put you in the top 10%. Three-word phrases repeat about thirteen times per thousand words, which is why the default here starts at four — below that you are reading noise. The tool shows the expected figure for your word count next to the result, so the number has something to be compared against.

Two things nobody else reports

Sentences that appear twice. Word for word, once punctuation and capitals are set aside. This is never a style choice. It is a paste that went wrong, usually when a section was moved, and it is invisible when the two copies are four paragraphs apart.

Sentence openers that get stuck. Three or more sentences in a row starting with the same word. “The road asks for cushioning. The trail asks for grip. The treadmill asks for very little.” Each sentence is fine. Together they drone. No other checker mentions it, and it is one of the easiest things to fix once somebody points at it.

Is repetition bad for SEO?

No, and the opposite worry does more damage. There is no penalty for using your subject repeatedly, and reaching for awkward synonyms to avoid it reads worse and achieves nothing. Google has not needed word counts to work out what a page is about for a very long time.

What repetition costs you is attention. A doubled word makes a page look unproofread, and a reader who notices one starts looking for others instead of reading your argument. An echo makes a sentence clunk. A phrase used six times becomes something the reader is tracking rather than your point. Fix those because the writing is better for it.

Which tool do I want?

This one is a proofreading tool. It looks for mistakes and tics in a draft before it goes out.
The Keyword Frequency Counter answers what a page is about and how that compares with a competitor’s page. Same arithmetic, completely different question.
The Keyword Density Checker audits a published page: whether your terms reach the title, the headings, the URL and the alt text.
The Free Readability Checker scores how hard the writing is on seven formulas, which repetition does not affect at all and readers very much do.
The Sentence Length Checker finds the sentences that run too long, and shows the rhythm of a page in the order you wrote it.
The Word Counter is for while you are drafting, with counts updating as you type.

Is anything stored?

No. Pasted text is checked entirely in your browser and never leaves the page, which is why findings appear as you type. The one exception is checking a page by address: a browser is not permitted to fetch another site’s pages directly, so the address is sent to our server, which reads the page and returns the text. Nothing is written to a database or a log either way, and the report exists only in the open tab. Export the CSV first if you want to keep it.

Five kinds of repetition, not one

Most tools with this name count your words and list the ones that appear more than once. That is a frequency counter wearing a different hat, and it misses the thing people are usually looking for: the doubled word their eye slid straight over.

Doubled words, line breaks included

"The the" is easy to type and almost impossible to see, especially when the first word ends a line and the second starts the next one. Those are found as readily as the ones sitting side by side, which is where most checkers give up.

It knows "had had" is fine

Some doublings are real English. We measured which ones over 1.6 million words of edited prose, and those are listed separately with the construction named. Run back over the same corpus, the finished tool calls two things a slip, which is one every 800,000 words, and neither of them is English.

Close repeats, with a real threshold

Using the same word twice in a breath is what makes writing sound clumsy. In edited prose only 2.6% of repeats fall within ten words and 4.9% within twenty, so the distance you set means something rather than being a round number.

Whole phrases, at full length

Search for repeated four-word phrases inside "at the end of the day" and you find three overlapping fragments. This reports the longest match once, so the finding is the phrase you actually wrote.

Pasted sentences and stuck openers

A sentence that appears twice word for word is a paste that went wrong, not a style choice. Three sentences in a row opening with the same word is a rhythm problem no other checker mentions. Both are reported.

Shown where it happens

Every finding comes with the words either side of it, and the whole text is shaded so you can see the clusters. A list of words with numbers next to them tells you nothing about where to edit.

Questions about repetition

What counts as a duplicate word?

It depends which problem you have, which is why this tool separates them. A doubled word is the same word twice in a row with only space between: "the the". A close repeat is the same word used again a few words later, which is a style problem rather than an error. A word you lean on is one that carries too much of the whole piece. A repeated phrase is several words reused together. A repeated sentence is usually a paste error. They need different fixes, so lumping them into one list of words with counts beside them helps nobody.

Why does it say "check it" instead of flagging "had had"?

Because "we had had enough" is correct past perfect, and a tool that calls it an error teaches you to ignore its warnings. We went through 1.56 million words of edited prose and collected every doubled word that actually occurs in it. There are very few: "had had", "that that", "you you", "in in", "her her", "it it", "there there", the double copula in "what it is is", and a handful of deliberate repetitions such as "very very".

Those are listed with the construction named, so you can glance at each one and move on. Everything else is marked as a slip, because in that whole corpus nothing else turned up. A doubled word appears roughly once every 8,200 words of published writing, so if your draft has four of them in a thousand words, they are not all grammar.

We then ran the finished tool back over the same corpus to count how often it would be wrong. It called two things a slip in 1.6 million words of published prose, one every 800,000 words, and neither was English: a French causative and a line of written dialect.

How close is too close for a repeated word?

Every style guide says not to repeat a word too close to itself and none of them says what "too close" is, so we measured it. Across the same corpus, only 1.2% of content-word repeats happen within five words of each other, 2.6% within ten, 4.9% within twenty and 10.5% within fifty. Published prose simply does not do it often.

Twenty-five words is the default here because it catches the genuinely awkward ones without burying you. Narrow it to ten if you only want the worst, or widen it to a hundred if you are polishing. Function words such as "the" and "of" are left out entirely, because they repeat constantly in all good writing and always will.

Is repeating a word bad for SEO?

No, and the opposite worry is the more common mistake. There is no penalty for using your subject repeatedly, and swapping in awkward synonyms to avoid it reads worse and helps nothing. Google works out what a page is about from far more than word counts.

What repetition costs you is attention. A doubled word makes a page look unproofread, which costs trust. An echo makes a sentence clunky. A phrase you use six times becomes a verbal tic the reader starts noticing instead of your argument. Fix those because they make the writing better, not because of a ranking factor.

Why is my main keyword listed under "words you lean on"?

Because it is the word you use most, which is exactly what you would expect on a page about that subject. That list is sorted by count, not by fault. An article about running shoes should say "shoe" a lot.

The useful part is further down: the word you did not realise you were leaning on. "Important", "really", "various", "essentially" and the like turn up four or five times in a draft and add nothing any of those times. The per-thousand-words figure is there so the number means the same thing on a short page and a long one.

How is this different from your keyword tools?

Completely different jobs that happen to involve counting. The Keyword Frequency Counter answers "what is this page about, and how does that compare with a competitor". The Keyword Density Checker audits where your terms land on a published page: the title, the headings, the URL, the alt text. Both are SEO research tools.

This is a proofreading tool. It is looking for mistakes and tics in a draft before you publish it, and most of what it reports has nothing to do with keywords at all.

Does it handle different forms of the same word?

Yes, for close repeats and for the words you lean on: "process" and "processes" are counted together, and both forms are shown so you can see exactly what matched. The matching is deliberately shallow, stripping only endings it can be certain about and only when enough of the word is left, because a stemmer that merges words a reader hears as different would report repeats that are not there.

Doubled words are matched exactly, ignoring capitals, since "The the" at the start of a sentence is the same slip as "the the" in the middle of one.

Is my text sent anywhere?

No. Pasted text is checked entirely in your browser and never leaves the page, which is why the findings appear as you type. The one exception is checking a page by address: a browser is not permitted to fetch another site's pages directly, so the address is sent to our server, which reads the page and returns the text. Nothing is written to a database or a log either way, and the report is gone when you close the tab.