← All reviews

Simplify or Help? Text Simplification Strategies for People with Dyslexia

Luz Rello, Ricardo Baeza-Yates, Stefan Bott, Horacio Saggion · 2013 · Proceedings of the 10th International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/2461121.2461126

Summary

This paper evaluates two automatic lexical simplification strategies designed to make text more accessible for people with dyslexia. The research addresses a key gap: while previous tools for dyslexic readers focused on text presentation (fonts, spacing, colors), none had tackled simplifying the actual text content. The authors used LexSiS, an automatic lexical simplification system for Spanish, and tested two strategies. The first, SubsBest, follows the standard approach of automatically replacing complex words with simpler synonyms. The second, ShowSyns, is a novel interactive strategy that highlights complex words and lets readers tap or click to see up to three simpler synonyms on demand, without altering the original text. The study involved 96 participants — 47 with a confirmed dyslexia diagnosis and a control group of 49 without dyslexia — making it the largest user study of its kind at the time. Participants read scientific texts from the Spanish edition of Scientific American, either in original form, automatically simplified, interactively assisted, or manually simplified by language experts (gold standard). The researchers used eye-tracking technology to measure reading time and fixation duration, comprehension tests with inferential questions, and Likert-scale easiness ratings covering readability, understandability, and memorability. The ShowSyns strategy was tested across three devices — laptop, tablet, and smartphone — to demonstrate device independence.

Key findings

Participants with dyslexia rated texts presented with the ShowSyns (interactive synonym) strategy as significantly more readable (p=0.015 vs original) and more understandable (p=0.001 vs original, p=0.013 vs SubsBest) than all other conditions. The automatic substitution strategy (SubsBest) performed poorly: participants without dyslexia found SubsBest texts significantly harder to understand than all other options, and harder to remember than manually simplified text. Neither simplification strategy produced significant effects on objective reading time, fixation duration, or comprehension scores. However, the subjective perception of readability strongly correlated with actual reading speed for dyslexic participants — they perceived texts as more readable when they read them faster. Crucially, only about 10% of words were modified by SubsBest, which the authors suggest may explain the lack of measurable objective effects. The study revealed an important asymmetry: people with dyslexia perceived the objectively more readable options (ShowSyns, gold standard) as more readable, while people without dyslexia perceived the opposite — the options they read faster were rated as less readable.

Relevance

This research has significant implications for how we design reading tools and content systems for people with dyslexia. The key takeaway is that giving users control over simplification — showing synonyms on demand rather than automatically substituting words — is more effective than fully automated text modification. Automatic word substitution can actually make texts harder to understand by introducing awkward word combinations or losing contextual meaning. For web developers and content creators, this suggests that interactive glossary or synonym features may be more valuable than automated text rewriting. The finding that subjective perception of readability matters even when objective measures show no difference is also important: if readers with dyslexia feel that text is easier to read, they may be more willing to engage with reading activities, helping to break the cycle where slower reading leads to less reading practice. The study is limited to Spanish-language texts and lexical-level simplification only, leaving syntactic simplification and other languages as areas for future work.

Tags: dyslexia · text simplification · readability · cognitive accessibility · eye tracking · lexical simplification · natural language processing · user study