Readability Score Calculator — Flesch, Fog and Coleman–Liau

Background.

Open the readability statistics in Microsoft Word and you get two scores that both go back to Rudolf Flesch. Reading Ease, which he published in 1948, runs on a scale where higher means easier. The Flesch–Kincaid grade, worked out for the US Navy in 1975, turns the same inputs into a school grade, where higher means harder. Neither formula reads a word for meaning. Each takes two averages — how many words your sentences hold and how many syllables your words hold — and feeds them through coefficients fitted to how real readers performed on graded passages.

Two other formulas count different things. Robert Gunning's Fog index, from his 1952 book on clear writing, keeps sentence length but swaps syllable density for the share of words with three or more syllables. Meri Coleman and T. L. Liau wanted a formula a computer could score, so in 1975 they dropped syllables entirely and counted letters, which a machine counts reliably. Run all four on one passage and they will rarely agree; on the worked example below they span almost three grades. That spread is information. When Fog sits well above the Kincaid grade, long words are doing the damage. When every score falls as you split sentences, length was the trouble.

You enter counts here rather than pasting text. Whether "e.g." ends a sentence, whether "3.5" is a word, whether "every" has two syllables or three and "fire" one or two: automatic checkers settle these silently, and two tools can disagree by a grade on the same paragraph for no reason except tokenising. Counting, or at least checking a tool's counts, keeps those choices visible, so a before-and-after comparison measures your revision instead of a change in counting rules.

The scores answer a narrow question well: compared with the last draft, are the sentences shorter and the words plainer? Whether a reader will understand the text, trust it or act on it is a different question, and none of the four can answer it.

What is readability score calculator?

A readability formula is a regression equation. It predicts how hard a passage will be from surface features that can be counted, with coefficients chosen so its predictions matched the results of real readers on test passages: Flesch fitted Reading Ease to the McCall–Crabbs reading lessons used in American schools, and Kincaid's team re-fitted the formulas to comprehension results from Navy enlisted personnel. The outputs come in two kinds. Reading Ease is a score intended to run from about 0 (very difficult) to 100 (very easy). The other three are grade levels in the US school system, so 9.91 points to text suited to a reader around the end of ninth grade, and 12 to the final year of high school. Both kinds describe the text, never a particular reader.

How to use this calculator.

1. 01Pick the passage you want to judge and remove what is not running prose: headings, captions, tables, reference lists and navigation text. For a long document, take several samples of 100 words or more from the beginning, middle and end instead of one paragraph you already suspect.
2. 02Count Words and Sentences under rules you write down: whether numbers count as words, whether a bulleted item without a full stop is a sentence, and that abbreviations such as "Dr." or "e.g." do not end one. In a 100-word sample, a single miscounted sentence boundary moves Reading Ease by three to five points.
3. 03Enter Syllables as the total for the whole passage, checking a dictionary's syllable breaks when unsure. For Complex words, count words of three or more syllables after Gunning's exclusions: proper nouns, compounds of short words such as "bookkeeper", and verbs that reach three syllables only through an -ed or -es ending, like "created".
4. 04For Letters, count alphabetic characters only, leaving out digits, spaces, hyphens and punctuation. Syllables and letters can never be fewer than the words, and complex words can never exceed them; if the calculator rejects a value, one of the counts probably came from a different version of the text.
5. 05Read the three averages under the scores before the scores themselves. Words per sentence, syllables per word and the complex-word percentage tell you which feature to change. Revise, recount with the same rules, and compare the new figures with the old.

The formula.

Everything starts from ratios of five counts: W words, S sentences, Y syllables, C complex words and L letters.

Reading Ease = 206.835 − 1.015 × (W ÷ S) − 84.6 × (Y ÷ W) Flesch–Kincaid grade = 0.39 × (W ÷ S) + 11.8 × (Y ÷ W) − 15.59 Gunning Fog = 0.4 × [(W ÷ S) + 100 × (C ÷ W)] Coleman–Liau = 0.0588 × L₁₀₀ − 0.296 × S₁₀₀ − 15.8, where L₁₀₀ = 100 × L ÷ W (letters per hundred words) and S₁₀₀ = 100 × S ÷ W (sentences per hundred words).

The constants are fitted, not derived. 206.835 is simply where Flesch's regression line crosses zero sentence length and zero syllables; nothing anchors it to 100, so simple enough text scores above 100 and dense enough text below zero.

The two Flesch formulas use identical predictors in different proportions. In Reading Ease, an extra tenth of a syllable per word costs 8.46 points, the same as lengthening every sentence by about 8.3 words. In the Kincaid grade that tenth of a syllable adds 1.18 grades, which equals only about 3 extra words per sentence. Relative to word length, Kincaid's re-fit gives sentence length about 2.75 times the weight Flesch gave it, which is why the two can rank a pair of passages differently.

Fog is the simplest: average sentence length plus the percentage of complex words, multiplied by 0.4, so every ten percentage points of complex words adds 4 grades. Gunning's 0.4 scales the sum onto school grades.

Coleman–Liau is governed by word length. One extra letter per word adds 5.88 grades, while one extra sentence per hundred words removes only 0.296. A passage averaging 5.5 letters per word instead of 5, with everything else as in the worked example, rises from 12.12 to 15.06.

No result is capped or rounded to a whole grade, and no term accounts for how familiar the words are, how the page is laid out or what the reader already knows.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Kincaid et al. (1975), Navy Research Branch Report 8-75, reports the Flesch–Kincaid coefficients and related formulas.
2. [2]Flesch (1948), A New Readability Yardstick, is the original Reading Ease calibration.
3. [3]CDC Clear Communication Index evaluates communication features beyond readability formulas.
4. [4]U.S. Plain Language guidelines cover audience, organisation, wording and design.
5. [5]CMS Written Materials Toolkit warns against treating a readability formula as a complete communication assessment.