What this tool does
"Make it more readable" is not an instruction until you have a number, but a single number is misleading, because five formulas disagree about the same paragraph on purpose — they were built for different audiences and different languages. This tool runs Flesch Reading Ease, Flesch-Kincaid Grade, Gunning Fog, SMOG and Coleman-Liau at once, shows the plain-language statistics underneath, and flags the sentences costing you the most.
How it works
Every formula here was designed for English prose, and the differences between them are differences of assumption rather than of arithmetic. Flesch Reading Ease is 206.835 minus 1.015 times the words per sentence minus 84.6 times the syllables per word, on a scale where 90-100 is very easy and 60-70 is plain standard English. Flesch-Kincaid turns the same two inputs into a US school grade level. Gunning Fog uses words per sentence plus the share of complex words, and it is the one with a stated exclusion list: proper nouns, hyphenated compounds counted as their parts, and two ordinary words run together are all excluded, so a proper noun that starts a sentence gets counted as complex because capitalisation gives no evidence either way. SMOG is built on the count of three-syllable words and has a 30-sentence minimum below which it is not meaningful. Coleman-Liau uses letters per 100 words instead of syllables, which makes it the only one of the five that never has to guess how many syllables a word has.
The honesty problem is syllable counting, and it is worth being blunt about it. There is no reliable way to count syllables in a word from its spelling, so every implementation uses a heuristic, and published implementations disagree with each other on roughly 5-10% of ordinary English words. The counter here counts vowel runs with y as a vowel, subtracts one for a silent trailing e unless it is a bare "le" after a consonant, returns 1 for a word with no vowel at all, and caps the result at 3 because the formulas only distinguish one, two and three-plus. So "queue" reads as 1 rather than 1, "business" reads as 3, and any loanword whose spelling does not follow English rules is simply wrong. A score printed to two decimal places is therefore false precision, and a score without a stated margin of error is worse than one that admits it.
The underlying statistics are here because they are more actionable than the composite scores. Words, sentences, paragraphs, characters with and without spaces, average words per sentence, average syllables per word and lexical diversity give you the parts, and every one of the formulas is a combination of them. The most useful number on the page is usually average words per sentence, because it is exact — it is a division, not an estimate — and because long sentences are the most common reason a paragraph reads as hard. Sentence splitting handles decimals, domain names, file extensions, a list of common abbreviations and single-letter initials, so "Dr. Rao paid 3.5 lakh in March." is one sentence rather than three.
The most important limitation is script. None of these formulas were built for Devanagari, Tamil, Telugu, Bengali or any other Indic script, because they assume vowels and syllable structure that those scripts do not have, and because a text in one of those scripts has no comparable English grade level to be measured against. The tool detects when a meaningful share of the letters are Indic — more than 30% — and refuses to present a meaningful score rather than printing a number that means nothing. A stray Sanskrit word inside an English paragraph is fine and does not invalidate the result. Reading time is included at both 200 words per minute for silent reading and 130 for text read aloud, and it is a division, so unlike the readability scores it is exact.
Worked example
A 84-word paragraph of technical prose that a new reader is meant to follow without rereading it.
- 84 words, 5 sentences, 596 characters, 513 without spaces, 157 syllables
- Average 16.8 words per sentence and 1.87 syllables per word — both exact divisions, and the first is the number worth attacking
- Flesch Reading Ease 31.66, which is difficult: 30-40 is the band for dense professional or academic writing
- Flesch-Kincaid grade 13.02, Gunning Fog 19.58, SMOG 16.4, Coleman-Liau 18.35 — the spread across five formulas is the point
- 27 words of three or more syllables, which is what Gunning Fog is reacting to
- Reading time: 0.42 minutes silent at 200 words per minute, 0.65 minutes spoken aloud at 130
One paragraph reading as roughly a first-year undergraduate text, with the word count, sentence count and syllable counts exposed so the specific cost can be located rather than guessed at.
Accuracy and limitations
- None of these formulas were designed for Indic scripts, and the tool refuses to report a score for text that is more than 30% Devanagari, Tamil, Telugu or another Indic script. That is a refusal, not a failure — a number for that text would be arithmetic on assumptions that do not hold.
- Syllable counting is a heuristic and published implementations disagree on 5-10% of ordinary English words. Treat the grade levels as a band of roughly a year, not as a precise measurement, and prefer the exact statistics over the composite scores when deciding what to edit.
- SMOG has a 30-sentence minimum and is not meaningful below it, and Gunning Fog's proper-noun exclusion relies on capitalisation, which gives no evidence at the start of a sentence. A short sample or one full of proper nouns will move the Fog figure for reasons that have nothing to do with readability.
Frequently asked questions
- What is a good Flesch Reading Ease score?
- For general business writing, 60 to 70 is the usual target, which is the band most style guides call plain standard English. 90 to 100 is very easy, close to what a child reads, and 30 to 40 is dense professional or academic prose. Below 30 is genuinely hard going, and above 90 usually means the text is choppy rather than accessible.
- Why do the five formulas give different grade levels?
- Because they are built on different assumptions, not because one of them is broken. Flesch uses syllables, Coleman-Liau uses letters per 100 words and never has to guess at syllable counts, and Gunning Fog and SMOG use different definitions of a complex word. A spread of three or four grade levels between them on the same paragraph is normal, and it is the reason quoting a single score is not very informative.
- Can I use a readability score for Hindi, Tamil or Telugu text?
- No, and the tool will tell you so rather than print a number. All five formulas were built for English and assume vowel structure and syllable counts that these scripts do not have, so a score would be arithmetic on assumptions that do not hold. The word, sentence and character counts are still shown, since those are script-independent, but the readability and grade-level figures are suppressed for more than 30% Indic letters.
- How accurate is the syllable count?
- Not very, and any tool that says otherwise is overselling. Syllables cannot be reliably derived from spelling, so every implementation uses a heuristic, and published implementations disagree with each other on roughly 5-10% of ordinary English words. Read a grade level as a band of about a year, and use the exact statistics — words per sentence, character count, complex words — when deciding what to actually change.
- What is the fastest way to make a paragraph easier to read?
- Shorten the sentences. Average words per sentence is an exact number rather than an estimate, it is the largest single term in both Flesch formulas, and long sentences are the most common reason a paragraph reads as hard. The tool scores each sentence individually so you can see which specific ones are costing you, instead of guessing where the problem is.
- Is my text uploaded anywhere?
- No, everything runs in your browser tab. For a tool where people paste unpublished drafts, internal documentation or unreleased announcements, that is not a nice-to-have — it is the condition on which the tool is usable at all.