Using N-grams to Analyse Content
Paste your text, choose a phrase length of two, three or four words, and this tool lists every phrase that appears more than once, ranked by frequency. Only repeated phrases are shown, since a list of every unique phrase in a document would be as long as the document.
An n-gram is simply a sequence of n consecutive words. Bigrams are two-word phrases, trigrams three, and so on. They are a standard tool in text analysis because meaning lives in phrases far more than in isolated words.
Why phrases beat single words
Single-word frequency tells you a page mentions “running” and “shoes” often. It cannot tell you whether those words appear together as a topic or scattered across unrelated sentences. The bigram “running shoes” answers that directly.
This matters for keyword work because almost all real search queries are phrases. A page optimising for individual words is optimising for something nobody searches, whereas phrase analysis shows the actual multi-word terms your content emphasises.
Longer n-grams reveal something different again. Repeated three and four-word sequences are usually either your genuine key phrases or, more often, verbal tics — the same construction reused across paragraphs.
Choosing a phrase length
- Bigrams (2) — best for identifying the key topics and keyword phrases a page targets.
- Trigrams (3) — good for finding specific long-tail phrases and the beginnings of repetitive phrasing.
- 4-grams — mostly surface writing habits: repeated transitions, stock openers, filler constructions.
Spotting repetitive writing
Longer n-grams are unexpectedly useful as an editing tool. If “it is important to” appears six times in an article, or every section opens with “when it comes to”, the four-gram list surfaces it immediately in a way rereading rarely does.
This is one of the more reliable ways to catch AI-assisted or hastily written content that reads acceptably sentence by sentence but repeats the same handful of constructions throughout. Readers register that repetition as monotony even when they cannot name what is wrong.
The fix is straightforward once you can see the pattern: vary the constructions, or cut them. Most repeated four-grams are filler that the sentence works better without.
N-grams for keyword research
Running competitor content through a bigram and trigram analysis is a quick way to see which phrases they emphasise. Comparing that against your own page shows where your coverage diverges.
It is also useful in reverse. If you have a large body of your own content — support tickets, product reviews, forum posts — n-gram analysis surfaces the phrases your actual audience uses, which are frequently different from the industry terms you would have guessed. Those phrases are keyword candidates worth checking.
For turning a list of such candidates into a plan, the Keyword Clustering Tool groups them by topic, and the Keyword Intent Classifier tells you what kind of page each group needs.
How the counting works
The text is split into words, lowercased, and stripped of punctuation. Every consecutive sequence of the chosen length is counted, including sequences that span sentence boundaries — so the last word of one sentence and the first of the next can form a phrase.
That is a deliberate simplification. Filtering by sentence boundary would be more linguistically correct but would miss phrases split across clauses, and in practice the cross-boundary noise is minimal and easy to ignore.
Phrases appearing only once are excluded. A unique phrase carries no information about emphasis or repetition, which is what this analysis is for.
Where n-grams mislead
- Boilerplate pasted in with your content — navigation, footers, calls to action — produces high-frequency phrases that say nothing about the article.
- Short texts produce thin results, since few phrases repeat in under a few hundred words.
- Necessary repetition looks like a problem: a product page legitimately repeating the product name is not badly written.
- It counts exact sequences, so “running shoes” and “shoes for running” are unrelated entries.
- High frequency does not mean importance — position in the page matters more, which the Keyword Prominence Checker measures.
Frequently asked questions
What is an n-gram?
A sequence of n consecutive words. A bigram is two words, a trigram three, a 4-gram four. They are used in text analysis because meaning and search queries live in phrases rather than in isolated words.
Why do I only see phrases that appear more than once?
A list of every unique phrase would be roughly as long as the text itself and would tell you nothing. Repetition is the signal this analysis exists to surface.
Which phrase length should I use?
Bigrams for identifying keyword topics, trigrams for long-tail phrases, and 4-grams for catching repetitive writing habits. Most people get the most value from bigrams and 4-grams for different reasons.
Do phrases spanning two sentences get counted?
Yes. The text is treated as one continuous word sequence, so a phrase can cross a sentence boundary. In practice this adds little noise and avoids missing phrases split across clauses.
Is repeated phrasing always bad?
No. Product names, technical terms and brand names legitimately repeat. What is worth fixing is repeated filler — stock transitions and constructions that add nothing and make writing feel monotonous.