Flesch Reading Ease — score, CEFR and plain-language check

Why does the Flesch Reading Ease still matter?

Three professions deal daily with the gap between what a text says and who can actually read it. Government employees writing a letter to a citizen. Marketing leads balancing conversion against skip-rate. Language teachers picking a textbook for the right proficiency level. For all three, the Flesch score is the first quick orientation — a number on a scale from zero to one hundred that hides a substantive linguistic statement behind it.

This tool bundles four such numbers on one page. The headline score is the classic Flesch — Rudolf Flesch’s English 1948 original or Toni Amstad’s German 1978 adaptation. Alongside it, the Wiener Sachtextformel, LIX and Gunning-Fog recompute the same sample independently. When the four agree, the diagnosis is solid. When they diverge, the per-sentence table shows where the discrepancy comes from.

How does the Flesch formula work?

The English Flesch Reading Ease follows the formula FRE = 206.835 − 1.015 × ASL − 84.6 × ASW. ASL is the average sentence length in words, ASW is the average number of syllables per word. Two levers, that is all — and that parsimony is also the formula’s strength. Long sentences lower the score. Multi-syllable words lower the score. Anyone wanting to raise the score gets a direct instruction: shorter sentences, shorter words.

The German Amstad adaptation rescaled both coefficients in 1978 to match the structurally longer German lexicon. German compounds add syllables that English avoids, and German sentences are on average a few words longer. Without the Amstad recalibration, a German text systematically scores too low — a common bug in tools that do not detect the input language.

A look at the original Flesch publication makes the maths transparent enough to redo on a napkin. That transparency is also why the formula has survived as a teaching tool for 78 years, despite repeated calls to replace it with neural language models.

When is the Flesch score alone not enough?

Three situations call for a parallel look at other indices. First, technical writing with many loan words — Flesch treats them as English while the reader may experience them as foreign. Second, very short samples where statistical variance makes any single number unreliable — the Wiener Sachtextformel tends to be more stable. Third, texts with conspicuously many long but transparent words (“telecommunication”, “interconnection”) — LIX is more honest here because it counts letter length, not the linguist’s guess at syllable boundaries.

For exactly this reason, the tool displays four indices simultaneously: Flesch (syllable-based), Wiener Sachtextformel (syllable-based, German-calibrated), LIX (character-based, language-neutral) and Gunning-Fog (syllable-based, English-calibrated for US grade level). When all four point in the same direction, the verdict is robust. An index that diverges sharply from the rest usually flags a language or sample problem rather than a text judgement.

What does the plain-language traffic light check?

On 28 June 2025 the German Barrierefreiheitsstärkungsgesetz (Accessibility Strengthening Act, BFSG) came into force. It obliges many businesses to make their consumer-facing communication accessible to people with reading and comprehension difficulties; German authorities have been bound by the related Barrierefreie-Informationstechnik-Verordnung 2.0 since 2002. The substance is the same: short sentences, simple words, sparse use of passive voice, few nested clauses.

The traffic light in this tool checks four factors against empirically grounded thresholds from the Hamburg Comprehensibility Model and recommendations of the German Netzwerk Leichte Sprache. Green (“Suitable”) means: average sentence length ≤ 15 words, complex words (≥ 3 syllables) < 10 %, passive ratio < 15 %, nested-sentence ratio (> 2 commas) < 10 %. Orange (“Borderline”) means the text is in plain-language territory but not quite Leichte Sprache. Red (“Not suitable”) signals legal or academic prose that would not be acceptable for citizen communication.

Important: the traffic light is a first orientation, not a legal opinion. Genuine BFSG conformance also requires images, layout, screen-reader compatibility and manual review by people from the target group. Sentence statistics are one building block among several.

How does the CEFR bridge help language teachers?

The Common European Framework of Reference for Languages ranks language competence in six levels — A1, A2, B1, B2, C1, C2. Teachers of English as a second language need daily information about which text suits which level. A pragmatic conversion table from L2-teaching practice maps higher Flesch scores to lower levels — because lower levels need easier text.

The thresholds in this tool: Flesch 90 or above is A1-compatible, 80 to 90 is A2, 70 to 80 is B1, 60 to 70 is B2, 40 to 60 is C1 and below 40 is C2. These are rules of thumb, not scientifically exact boundaries — the CEFR scales themselves are defined through communicative can-do statements, not through readability scores. But for a first textbook scan, the thresholds work.

A useful side-effect: if you target a marketing text at B1 (internationally communicable) and the tool measures B2, you see immediately where to cut. The per-sentence table then highlights the exact sentences that are too long or too complex.

How do I read the per-sentence table?

The table shows every sentence with its word count, its individual Flesch score and a hint tag. Tags are “long” (more than 25 words), “nested” (more than two commas), “many long words” (three or more words with ≥ 3 syllables) and “passive” (auxiliary “is”/“was” with a past participle).

To rewrite a text, sort the table by “hardest first”. The top ten to fifteen sentences are usually where the biggest gains live. To lift a 70-point text to 80, you can split three over-long sentences and replace two nominal constructions — and the score climbs visibly.

The table caps at the first 60 sentences in any sort order; longer documents see a note about hidden rows. That keeps scroll fatigue in check while still surfacing the difficulty hotspots.

Where does the syllable heuristic fall short?

Syllables are a non-trivial problem. English follows broadly known patterns: count vowel groups, subtract a silent final E, restore the syllable for “-le” after a consonant. German has its own logic: each vowel group is a syllable, diphthongs (ei, ai, au, eu, äu, ie) count as one, umlauts and ß count as letters.

Both heuristics are robust but imperfect. Proper names, abbreviations and loan words can deviate — “PowerPoint” reads as two syllables to the tool but a native speaker hears three. For aggregate statistics over 100 or more words, the error averages out. For individual words, the count is approximate; the authoritative reference shows how non-trivial accurate syllabification really is.

How fast can an adult read this text?

The reading-time estimate in the statistics box uses 250 words per minute. That is the median for silent adult reading of well-formatted material, based on a large meta-study by Marc Brysbaert in 2019. Actual reading speeds range from about 175 words per minute on difficult text to over 350 on easy fiction.

For marketing copy and web content, 250 is the safe default. Anyone writing a letter and reading “reading time: 2 min 30 s” knows immediately whether the format fits. A citizen letter with 90 seconds of reading time has a chance of being read in full; one with 8 minutes does not.

Frequently Asked Questions

All answers are also in the body text — the FAQs pick up the most common search queries directly.

Which Text Tools Are Related?

These tools complement the readability analysis:

• Character counter — count letters, words and reading time live as you type.
• Caesar cipher — encrypt, crack and learn the classical Caesar shift.
• Morse code translator — convert text to and from Morse with audio and a learn mode.
• Text diff — compare two text versions line by line.
• Lorem ipsum generator — produce placeholder text for layout testing.