Readability Score Calculator – Measure How Hard Your Text Is to Read
Readability formulas do something narrow and useful: they estimate the reading level a passage demands, from nothing but how long its sentences are and how long or how common its words are. That is a crude model of comprehension and a remarkably durable one — the same handful of equations has rated naval manuals, insurance policies, patient leaflets and newspaper front pages for seventy years. This readability score calculator runs nine of them at once, reports a consensus grade level, and shows the sentences and words that produced it.
The nine formulas and what each one actually measures
Flesch Reading Ease (Flesch, 1948) is the one style guides quote. It runs 0–100 with higher meaning easier, and most general-audience writing aims for 60 or above. Flesch–Kincaid Grade Level rewrites the same two inputs as a US school grade. Both use only average sentence length and syllables per word, which means a short rare word counts exactly the same as a short common one — their central blind spot.
Gunning Fog (1952) adds the share of words with three or more syllables, excluding proper nouns, easy compounds and words that only reach three syllables through a regular -es, -ed or -ing ending. SMOG (1969) counts polysyllables scaled to a thirty-sentence sample and is the standard in healthcare communication. Coleman–Liau and the Automated Readability Index count letters instead of syllables, which makes them reproducible and insensitive to pronunciation guesses. Linsear Write scores a hundred-word sample, weighting long words three times as heavily as everything else.
Dale–Chall is the interesting outlier. It is the only formula here that asks whether a word is familiar rather than merely short, by checking it against a published list of words a fourth-grade reader knows. That catches what the syllable-based formulas miss: gross, writ and tort are all one syllable and none of them is easy. Spache applies the same idea to primary-grade text, and reads low on adult prose because that is not what it was built for.
How to read the consensus grade level
The formulas disagree, sometimes by four or five grades, and that spread is information rather than noise. A high Gunning Fog with a low Coleman–Liau means long words in short sentences — dense terminology, briskly delivered. The reverse means plain vocabulary strung into sprawling sentences. The consensus grade level is the median of the grade-based formulas, which is more stable than any single one, and the grade-level ladder plots each formula as a tick so you can see the spread at a glance. A US grade converts roughly to reader age by adding five.
For target levels: general web writing and journalism usually aim at grades 7–9; consumer health information and instructions at grades 6–8; plain-language guidance for public-facing documents typically asks for grade 8 or below. Pick an audience preset and the tool grades the text pass, borderline or fail against that ceiling and lists what to fix first.
Turning a score into a rewrite
A number on its own changes nothing, so the tool breaks the passage down. The sentence breakdown scores every sentence and sorts worst first, with word count, syllables per word and a difficulty tag; the sentence-length histogram makes a tail of thirty-word sentences obvious. The hard-word finder lists the long and unfamiliar words with a plainer alternative where a safe one exists — utilize to use, approximately to about, terminate to end. Long sentences, apparently passive constructions and filler intensifiers are flagged separately, and a rewrite checklist is built from the problems this particular text has.
In practice, sentence length is the lever that moves fastest: splitting the three longest sentences usually drops the consensus grade further than any amount of word substitution, because every formula here weights average sentence length heavily. Use compare mode to put a draft and a rewrite side by side and watch each metric move.
Counting choices that affect the result
Sentences are split on full stops, question marks and exclamation marks, with an abbreviation list so Dr., e.g. and U.S. do not create false breaks, and a rule that a period between digits — 3.14 — never ends a sentence. Syllables come from a table of spelling traps plus a vowel-group heuristic, so they are an estimate rather than a lookup. Headings, bullet markers, code blocks and bare URLs are stripped by default: a heading carries no full stop, so left in place it merges into the sentence that follows and inflates every sentence-length average. Bullet items get a full stop added so each counts as its own sentence.
Everything runs in your browser — no upload, no request — so unpublished drafts stay private, and results export as TXT, CSV, JSON or a Markdown report.