Home Resources Readability checker
Free toolProse readability checker
Paste a chapter. Out come its reading ease, its sentence lengths, how much of it is dialogue, and the words it leans on. No account, no upload, no model reading your book.
A .docx, a .pdf, a .txt or a .md, up to 25MB. It is opened in this tab and never sent anywhere.
Reading ease
This runs entirely in your browser. Nothing you type is sent anywhere, and there is no account to make.
How the scores are worked out
Flesch reading ease is 206.835 minus 1.015 times the average sentence length minus 84.6 times the average syllables per word. It runs from about 0 to about 100 and it rewards short sentences and short words. Flesch-Kincaid grade is the same two inputs rearranged to give a US school year instead of a score.
Syllables are counted by heuristic, because English has no rule. Vowel groups are counted, a silent final e is dropped, and nothing returns zero. It is wrong on a handful of words in every thousand, which is the tolerance both formulas were built with.
Sentences are split on full stops, question marks and exclamation marks, after Mr, Dr, initials and ellipses have been put out of reach. Getting that wrong is how a page reports twice as many sentences as the chapter has and halves every average on it.
Dialogue is measured by quotation marks, and only the convention your book actually uses. If the text is set in curly doubles, single quotes are ignored, because otherwise every apostrophe in a contraction becomes a candidate for an opening quote and a chapter comes back 90% dialogue. Where those marks go is on the dialogue punctuation sheet.
Adverbs are words ending in -ly, minus a list of words that end in -ly without being adverbs, which is why family and reply are not in the count. Filter words are a fixed list of perception and cognition verbs. Repeats are words over three letters that appear more than once, with the common function words taken out first.
None of these are errors. There is no score, no grade and no verdict, because a chapter of stripped dialogue and a chapter of description have different numbers in the same well-made book. What the page is for is noticing when this chapter is unlike the rest of yours.
What it refuses. A .docm, because that is the macro format. A .doc, .rtf, .odt, .pages or .gdoc, each of which has an export that gets you a .docx. HTML. Anything over 25MB. And a .docx that unpacks to far more than its size suggests, which is how a zip file gets used to hang a browser tab. It is the same guard the Shunn format generator uses, because it is the same reader underneath.
Galley runs the same kind of pass across a whole manuscript rather than one pasted chapter, in revision, next to the overused word check and the pacing view.
For working things out, not for signing anything. Printers and retailers set their own rates and specifications, change them without notice, and none of them is affiliated with me or with Galley. I checked these figures by hand in September 2026 against Rudolf Flesch's reading ease formula . Confirm on the official page before you commit money to a price, a print run or a contract.
Questions
Does my chapter get uploaded anywhere?
No. There is no server behind this page. Whether you paste the text or open a file, it is read and counted by the browser you are reading this in. Nothing is kept between visits, so a reload empties the box.
What files can it open?
A .docx, a .pdf, a .txt or a .md, up to 25MB. Word, Google Docs, Pages, Scrivener and LibreOffice all export a .docx. Or paste the text straight in, which is what most people do for one chapter.
What reading ease should fiction have?
Most commercial fiction sits between 70 and 85, and literary fiction runs lower. The number is a shape, not a grade. A chapter of dialogue scores higher than a chapter of description in the same book by the same person, and neither is better writing.
Should I delete all my adverbs?
No. The count is here because a chapter that is 6% adverbs usually has weak verbs underneath, not because -ly words are wrong. Look at what the adverb is propping up. Sometimes the answer is that the adverb was doing the job.
What are filter words?
Verbs that put a narrator between the reader and the scene: she saw, he felt, they noticed, I realised. "She saw the lantern go out" is one step further from the reader than "the lantern went out". Some are necessary. Counting them shows you how often you reach for them.
How reliable is the passive voice count?
Roughly. It matches a form of "to be" followed by something that looks like a past participle, which catches most real passives and some innocent sentences too. Read the ones it finds before you change anything.
Is there AI in this?
No. Every number on this page comes out of counting, and sums you could do by hand with enough time. Nothing reads your book, nothing scores it, and nothing is generated.
Next
More at the resources index.
Put all this into practice when you write your novel in Galley.
Launch Galley