Counts and percentages together
A count tells you how often; a percentage tells you how heavily. Seeing that one verb accounts for half a per cent of every word in a chapter is far more persuasive than a bare number ever is.
Every writer has crutch words you cannot hear in your own drafts, because they sound like your voice, and that is what this word frequency analyzer is for. It ranks every word by count and percentage, filters out stop words, and flags the phrases you reach for again and again.
Paste text, then adjust the filters. The table rebuilds instantly with every change.
Frequency analysis is the oldest trick in editorial practice and still one of the sharpest. Sort a chapter by how often each word appears and the top of that list stops being a curiosity. It becomes a map of your defaults, showing which nouns you lean on, which verbs you reuse and which qualifier keeps creeping back in paragraph after paragraph.
The first pass will be dominated by the machinery of English: the, and, of, to. Those reveal nothing about your writing, which is why the stop-word filter exists. Switch it on and what surfaces instead are the terms carrying your actual meaning, ranked by how hard each one is working.
Phrases matter as much as single words. A repeated verb might be invisible, but a construction such as she felt a wave of appearing eleven times across a chapter is the kind of thing readers register subconsciously and reviewers mention explicitly. The two and three-word counts pull those out of the prose where you can see them.
A count tells you how often; a percentage tells you how heavily. Seeing that one verb accounts for half a per cent of every word in a chapter is far more persuasive than a bare number ever is.
Around two hundred structural words are held back by default, including articles, pronouns, auxiliaries and the most common prepositions. Turn the filter off whenever you want the unedited picture.
Raising the length threshold is the fastest way to skip past small connective words without maintaining an exclusion list. Six letters or more usually leaves only vocabulary you chose deliberately.
Two and three-word sequences are counted separately, and only sequences appearing at least twice are listed. That keeps the results focused on genuine habits rather than incidental word pairs.
Send over the revised file and we will return interior layouts ready for print and retail upload.
Unique words divided by total words. The figure falls naturally as a text gets longer, so always read it alongside the sample size.
| Ratio | Reads as | Typical of | What to check |
|---|---|---|---|
| Above 60% | Very varied | Samples under 300 words | Too short to conclude anything; paste more |
| 45 - 60% | Varied | Short stories, essays, single scenes | Nothing; this is healthy for the length |
| 35 - 45% | Comfortable | Chapters of 2,000 to 5,000 words | Scan the top twenty for one dominant verb |
| 25 - 35% | Typical | Full-length novels and nonfiction books | Normal; repetition is structural, not lazy |
| 18 - 25% | Repetitive | Long books, children's writing, technical manuals | Whether repeated terms are deliberate anchors |
| Below 18% | Very repetitive | Early readers, controlled-vocabulary texts | Intentional in some categories, a warning in others |
Any word appearing far more often than its meaning justifies. Common culprits include just, really, actually, somehow, suddenly and back, along with a favourite verb quietly doing the job of six different verbs. If a word is in your top twenty after stop words are removed and it is not central to the subject, it deserves a second look.
Because English is dominated by tiny words. Roughly a quarter of everything anybody writes is made up of a few dozen function words. Switch the stop-word filter on, or raise the minimum length, and the list will immediately start reflecting your real vocabulary choices.
It is unique words divided by total words, expressed as a percentage. On its own it is a weak signal, because it drops automatically as texts get longer. It becomes useful when you compare chapters of similar length and one comes back noticeably flatter than the rest.
No. Run, runs and running are counted separately because no stemming is applied. That is deliberate: seeing the exact forms you favour helps revision more than a merged root would, and stemming introduces errors of its own.
A phrase occurring once is just a sentence. The lists are filtered to sequences appearing two or more times, which is what turns a pile of word pairs into a set of habits you can act on.
A single chapter is the sweet spot. It is long enough that patterns emerge and short enough that you can trace each result back to specific pages. Whole manuscripts work too, but findings become harder to act on once you lose the sense of where the repetition sits.
Yes, and it is genuinely useful for that. Running your finished chapters through the analyser shows which terms your book actually emphasises, which you can then compare against the phrases readers type when they look for a book like yours.
There is no hard cap. Very long texts of several hundred thousand words may pause briefly while the counts build, since everything is processed in your browser, but the tool will get there. The displayed table is capped at your chosen number of rows for readability.
We take your finished manuscript and typeset it properly, with scene breaks, chapter openers, drop caps, running heads and a matching ebook build that behaves on every device your readers own.
See Fiction Formatting