Easy-to-read (ETR)

Easy-to-read, also called easy read or ETR, is a set of writing and layout rules for people with intellectual disabilities. In Europe the reference is "Information for all", a standard written by Inclusion Europe and partner organizations in the Pathways project. Inclusion Europe also owns the European easy-to-read logo, which publications that follow the guidelines may carry.

The rules are concrete. Use words people know, and explain every difficult word you cannot avoid. Use the same word for the same thing throughout. Avoid metaphors, words from other languages, special characters and abbreviations such as "e.g." Start each sentence on a new line and never split a word across two lines. Write numbers as digits, never use Roman numerals, and prefer the present tense. Use a clear sans-serif font, align text to the left and do not justify it.

The standard also sets a process rule: people with intellectual disabilities should be involved in making the information, including checking how easy it is to understand. A text that follows every writing rule but was never read by anyone from the target group has skipped part of the standard.

Easy-to-read is stricter than plain language. ISO 24495-1:2023 sets governing principles and guidelines for plain language documents, and its widest use is for documents intended for the general public. ETR aims at readers with intellectual disabilities and often also helps people with low literacy, older readers and people reading in a second language.

Software that produces ETR automatically, usually with a language model, has to turn the standard into requirements that can be tested:

• Meaning is preserved. The simplified text must not add facts or drop obligations, dates and amounts. A model that invents the expansion of an acronym changes the meaning for a reader who cannot check it.
• Writing rules become automatic checks: sentence length, paragraph length, one sentence per line, digits instead of number words, no Roman numerals, no abbreviations.
• Difficult words left in the text are explained. A dictionary of words and acronyms needs the context of the sentence, because the same word can mean different things.
• Readability formulas are a trend indicator. Most were designed for English and count syllables and sentence length, so they cannot tell whether the meaning survived.
• Readers from the target group check the output, at least on a sample, as the standard expects. The system should make that review easy and record its results.
• The interface is accessible as well. ETR readers use the buttons, forms and error messages, so the app needs to meet WCAG and keep its own labels in easy-to-read language.
• Reading aloud needs text normalization. Abbreviations, dates and numbers in the source have to be expanded and inflected correctly before speech synthesis.

WCAG 2.2, Success Criterion 3.1.5 Reading Level (level AAA), requires supplementary content or a simpler version when text needs reading ability above the lower secondary education level, after removing proper names and titles. ISO/IEC 23859:2023, published in July 2023, sets international requirements and recommendations for making written text easy to read and understand in any user interface, covering how such text is created, adapted and evaluated; it gives no language-specific guidance. Article 9 and Article 21 of the UN Convention on the Rights of Persons with Disabilities oblige states to make information accessible to persons with disabilities. In Poland, the 2019 Act on ensuring accessibility for persons with special needs (ustawa o zapewnianiu dostępności osobom ze szczególnymi potrzebami) requires public bodies to publish on their websites a description of the scope of their activities in easy-to-read text, alongside a machine-readable file and a recording in Polish Sign Language (Article 6(3)(c)).

For PSONI (Polskie Stowarzyszenie na rzecz Osób z Niepełnosprawnością Intelektualną, the Polish association for people with intellectual disabilities) we are building Generator ETR, a web and mobile app that simplifies pasted or uploaded documents into easy-to-read Polish with a self-hosted model. The app also has a dictionary of difficult words, reads the text aloud while highlighting the current fragment, keeps a history of simplifications and exports the result to DOCX. The project specification requires WCAG 2.2 level AAA.

The first round of model fixes addressed three problems: invented expansions of acronyms, including PSONI itself; formatting, with headings, bullet points and blocks of at most 4 to 6 lines; and difficult words left in the text, such as konflikt zbrojny (armed conflict). A "simplify more" option produces one further level of simplification. The dictionary covers difficult words and acronyms, and for words with several meanings it uses the context of the sentence. For reading aloud, PSONI reported errors in abbreviations and numbers, and in October 2026 we are adding a normalization step that, for example, makes Polish numerals agree with the nouns they count before synthesis.