Writers cannot see their own tics. You know what you meant to say, so your brain glides over the fifth “just” in a paragraph and the third “very” on a page. A word frequency counter does not glide — it counts, coldly and completely, and the resulting list is often the most honest feedback a draft will ever get.
This article explains what word frequency analysis reveals about writing: the crutch words hiding in your prose, what lexical density means, why keyword stuffing backfired as an SEO strategy, and how to turn a frequency list into a practical editing pass.
Your crutch words are showing
Every writer has filler words they lean on without noticing: just, very, really, quite, actually, basically, literally. In speech these are harmless softeners; on the page they dilute. A frequency count makes them visible — when “just” appears 47 times in a 2,000-word draft, the pattern is undeniable.
The fix is rarely to delete them all. Some instances earn their place. The value of the count is that it forces a decision per instance: does this word add meaning, or is it throat-clearing? Most crutch words, interrogated one by one, turn out to be removable — and each removal tightens the sentence.
Common offenders vary by person, which is why generic advice (“never use very”) is less useful than your own data. Run your last three pieces through a counter and your personal top ten will emerge. Those are your words to watch.
- just
- very
- really
- quite
- actually
- basically
- literally
- stuff / things
- a lot
- in order to
Lexical density and variety
Linguists measure vocabulary richness with the type-token ratio: unique words (types) divided by total words (tokens). A 500-word text with 200 unique words has a ratio of 0.4. Higher generally means more varied diction; lower means more repetition. It is a blunt instrument — it falls naturally as texts get longer — but as a comparative tool between drafts, it is revealing.
Related is lexical density: the proportion of content words (nouns, verbs, adjectives, adverbs) to total words. Dense academic prose packs in content words; chatty prose pads with function words (the, and, of, to). Neither is inherently better — a thriller and a textbook have different jobs — but knowing your density helps you match style to purpose.
The practical use is spotting monotony. If your type-token ratio is low for the genre, you are probably repeating the same nouns and verbs. A thesaurus is the wrong fix; the right fix is usually more specific observation — “walked” repeated twelve times wants “strode,” “shuffled,” or a scene rewritten so the walking matters less.
Keyword stuffing vs natural writing
In the early 2000s, search engines ranked pages partly by keyword frequency, so pages repeated target phrases absurdly: “Our cheap hotels in Paris are the cheapest Paris hotels for cheap Paris hotel stays.” It worked, briefly, and reading the web was miserable. Modern search engines use semantic understanding and actively penalise stuffing.
The irony is that frequency still matters — just naturally. A good article about sourdough will mention “starter,” “flour,” and “fermentation” often, because the topic demands it. Search engines now look for that topical vocabulary (related terms, entities, depth) rather than raw repetition of one phrase. Write thoroughly about the subject and the frequencies take care of themselves.
For writers, the takeaway is to check frequency for accidental over-optimisation as well as under-variety. If your target keyword appears in every other sentence, readers — human and algorithmic — will notice, and not fondly.
Editing with frequency data
Here is a practical workflow. First, run your draft through a word counter and look at the top 20 words, ignoring the obvious function words. Second, for each content word high on the list, ask: is this repetition deliberate (a theme, a motif) or lazy? Third, search each suspect and read every instance in context — frequency tells you where to look, but only context tells you what to cut.
Fourth, watch for accidental echoes: the same vivid adjective twice in three paragraphs, a character who “smiles” in every scene. Readers notice repetition far more than writers do, because writers remember intentions while readers only see words. Fifth, re-run the count after editing; the list should look calmer.
This pass takes twenty minutes on a long article and routinely removes the exact weaknesses beta readers would have flagged — before any human ever sees the draft.
What frequency can't tell you
Numbers are a guide, not a verdict. Frequency cannot judge rhythm — a repeated word can be anaphora, the deliberate drumbeat of great rhetoric (“I have a dream” works because of repetition, not despite it). It cannot judge voice — genre fiction, comedy and spoken-word styles repeat differently than essays. And it cannot judge meaning — “not” appearing often might be a careful argument, not a tic.
The failure mode is editing by spreadsheet: deleting every repeated word until the prose is varied but lifeless. Use the count to find candidates, then decide with your ear. If a repetition sounds good when read aloud, it stays — the counter is an advisor, and you are the editor.
Used that way, frequency analysis is one of the cheapest serious editing tools available: no course, no coach, just a list of what you actually wrote versus what you thought you wrote.
Frequently asked questions
What's a good unique-word ratio?
There is no universal target — it depends on length and genre. As a rough guide, short texts often sit between 0.4 and 0.6, and the ratio naturally drops as texts lengthen. Compare your drafts to each other and to writing you admire in the same genre, rather than chasing an absolute number.
Do longer, fancier words make writing better?
No. Variety helps, but precision helps more. “Utilise” is longer than “use” and worse in almost every sentence. Frequency analysis should push you toward more exact words, not more impressive ones — “shuffled” beats “walked slowly,” not because it is fancier but because it is truer.
Can frequency analysis detect AI-generated writing?
Partially. AI text tends toward average diction — common transitions, balanced vocabulary, few true rarities — and detectors use statistical patterns including frequency distributions. But it is an arms race: good human editing and newer models both blur the signals, so frequency alone is never proof either way.
How do I find my personal crutch words?
Run several of your pieces through a word frequency counter and look at the content words in the top 20. The ones that surprise you are the crutches — the words you did not know you used. Everyone's list is different, which is exactly why personal data beats generic banned-word lists.
Conclusion
Counting words sounds mechanical, but it reveals the most human thing about a draft: the habits its author could not see. Find your crutch words, check your variety, respect deliberate repetition — and let the cold arithmetic make your warm prose better.