Writing tone
Is your writing academic? Measure it.
Paste a paragraph. This page counts what it actually does — sentence length and spread, passive constructions, hedging, and the phrases that read as machine-written. No account, no upload, no AI.
Paste something above and the numbers appear here. Nothing is uploaded.
What the academic register actually means
A formal academic register is not ornate. It is impersonal where the field expects it, precise about who found what, and willing to say “appears to” when the evidence only supports that much. Long words are not the marker; unhedged claims are the failure.
The measurable signs are a steady sentence length with real variation inside it, hedges near the claims rather than sprinkled everywhere, and passives used where the actor genuinely does not matter — not as a default.
This page measures. It does not rewrite, and it is not a way to make AI writing pass as your own: nothing here changes your text, and the editor's rules against faking a human voice apply at every register.
How to read these numbers in academic prose
The meter counts; it does not grade. None of these figures has a target in the code behind this page, so the page gives none: good academic writing covers a wide range on every one of them, and the right comparison is your own earlier chapters. What follows is what each number is evidence of, and what to ask when one of them looks wrong.
- 1
Sentence length, and how much it varies
Average sentence and Spread are one reading, not two. A chapter whose sentences all sit within a word or two of the average reads as machine-set even when every sentence is correct, because the variation that comes from thinking at different scales has gone. Long sentences run to reports the ninetieth percentile, the length your longest tenth actually reach. That is more useful than a maximum: one sixty-word sentence in a chapter is a sentence, not a habit.
- 2
Sentences per paragraph
An average. Single-sentence paragraphs are legitimate in English academic prose for a transition or a point that has to land on its own, and a chapter made entirely of them is a slide deck. The figure worth distrusting is a flat three in every paragraph, which is what you get when structure has been produced rather than found.
- 3
Passive constructions
Counted per thousand words, because the rate is the thing and a raw count only tracks length. In a methods section the passive is usually right: the sample was stratified by year names no agent because the agent does not matter. In a discussion the same construction often hides whose claim is being reported. The count cannot tell those apart, so read where the passives fall rather than working the number down.
- 4
Sentences that hedge
The share of your sentences carrying a hedge: may, might, appears to, suggests, tends to. Hedging is not a weakness in research writing, it is how a claim is held to its evidence. The question is whether each hedge has a reason standing behind it. A result that may indicate a link, with nothing said about why the link is uncertain, is a sentence protecting itself rather than reporting a limit.
- 5
Em dashes
Counted per thousand words. The character is the measurement, so unlike every word-list signal on this page nothing can slip past it, and among the things counted here it is one of the loudest. A high rate is not an error, the em dash is good punctuation, but current model output reaches rates that human drafts of the same prose generally do not.
- 6
Reading ease
The Flesch reading-ease score, and the one measurement on this page that exists only in English: the syllable count behind it reads vowel groups, which makes the figure meaningless in Turkish, German, Spanish and French, so those pages print not measured instead of a precise-looking number about nothing. A low score is not a fault in a thesis. Technical vocabulary and subordinate clauses both push it down and both belong where they are. Read it against your own earlier writing, not against a general-audience target.
- 7
Words and sentences
The two counts are there so the rates above mean something. Below forty characters nothing is measured at all: a paragraph is the smallest honest unit, and a single sentence cannot show spread.
Every line above is a reading, not a threshold. Where your department publishes a rule, a word limit, a required citation style, a house preference about voice, that rule wins over anything on this page.
Why these seven, and when each one counts
Each pattern has a floor before it is reported, because every one of them is a legitimate move the first time. Two worn metaphors, three trailing participles, four inflated verbs: one is a sentence, four is a tic. Everything the page reports is quoted from your own text.
Trailing participial comment
The sentence finishes, then a comma adds a comment on what it just said: the data were collected in two waves, thereby highlighting the gap. The inference sits in the tail instead of in the sentence, which means nothing in it has to be shown. Reported from three.
Worn metaphor
Cornerstone, backbone, bedrock, linchpin, the holy grail, the tip of the iceberg. Each was an image once and now fills a slot. Reported from two. Gold standard and paradigm shift are deliberately not on the list: both are technical terms, in methodology and in the philosophy of science, and marking a randomised trial described as the gold standard would simply be wrong.
Two items dressed as a range
From policy to practice, from teachers to administrators: two items set out as though they were the ends of a spectrum. A real range is excluded on both sides, so from 2020 to 2024 and from baseline to week twelve are not marked. Numbers, dates and measured quantities make a range real; two nouns do not. Reported from three, and English only.
Stacked hyphenated modifiers
A data-driven, evidence-based, student-centred approach. Three hyphenated modifiers inside one sentence, and two such sentences before anything is reported. A compound modifier is not the problem; three stacked in one sentence is a habit rather than a description. English only, because the construction is.
Straw man setup
One might argue that the sample is too small, followed immediately by the answer. An objection nobody raised is built and then knocked down, and the argument looks won without having been had. Reported from three, because anticipating a real objection is part of a discussion section and the move is legitimate once.
Inflated verb
Serves as, acts as, stands as, boasts, represents a testament to. A layer of verb where a plain one would do: the chapter serves as a bridge between the two studies, rather than the chapter bridges them. Reported from four.
Talking about the text
This study examines, The analysis indicates, This chapter will show. The sentence puts the text in front of the finding. This one carries a measurement rather than an opinion: across seven completed theses, 107,000 words of body prose, the construction appears 0.19 times per 10,000 words. In our own English drafts it appeared 7.48 times, forty times the rate of the writing it was imitating. The future-tense form ran at 5.24 per 10,000 words in those drafts and 0.00 in the theses, which is the clearer finding of the two: finished work does not announce in its body what the analysis is about to do, it does it. Reported from three.
This is not a style guide, and the page rewrites nothing on its own. A phrase caught here is worth re-reading, not swapping for a synonym: replacing cornerstone with linchpin changes the word and leaves the sentence saying as little as before. What the counts are good for is locating the paragraphs you wrote on autopilot.
Want it rewritten in this register?
Measuring is free and happens in your browser. Rewriting is done in the editor, where the text has its sources, its citations and its section around it — that context is what keeps a rewrite from quietly changing what you claimed.
The editor rewrites in the register you picked and leaves your sources and your citations where they are. It works on the prose, sentence shape, the habits above, the register you chose, and not towards any score: nothing in it is aimed at how some other tool will read the result.
Open the editorQuestions
Is my text stored or sent anywhere?
No. The measurement runs on this page, with no account, no upload and no AI model involved in producing the numbers. Rewriting is a separate step in the editor and happens only if you ask for it.
If nothing is flagged, will an AI checker clear my text?
That is not a question this page can answer, and it is not what the patterns are for. Checkers are not published, they disagree with each other, and they change. What the seven patterns are evidence of is a habit in your own prose; how some other tool reads the result is outside what can be measured here.
Is the passive voice wrong in academic writing?
No, and a count is not a verdict. In a methods section the passive is usually the better choice, because naming the agent would add nothing. In a discussion it often hides whose claim is being reported. Same construction, different job, which is why the page reports the rate and leaves the reading to you.
Why does the page not tell me what the numbers should be?
Because it does not know, and neither does the code behind it: no measurement here has a target range. Average sentence length in a finished thesis depends on the discipline, the chapter and the supervisor. A page printing aim for 22 words would be inventing a standard your department has not set. Compare against your own earlier chapters, or against your faculty's style guide.
Which of these only work in English?
Two of the seven patterns and one of the measurements. The false range and stacked hyphenated modifiers are English constructions, so they are looked for here and not on the other language versions of this page. Reading ease is English-only as well, because the syllable heuristic behind the Flesch score reads vowel groups and the figure means nothing in Turkish, German, Spanish or French.
Write it in this register from the start
In the editor, the register is a document setting: every section, edit and continuation comes back in it. You can also give Grongy a piece of your own writing and have it match your rhythm — your habits, from your own sentences.