Towards the Prediction of Dyslexia by a Web-based Game with Musical Elements
Maria Rauschenberger, Luz Rello, Ricardo Baeza-Yates, Emilia Gomez, Jeffrey P. Bigham · 2017 · Proceedings of the 14th International Web for All Conference (W4A) · doi:10.1145/3058555.3058565
Summary
This paper presents DysMusic, a web-based serious game prototype that aims to predict the risk of dyslexia in pre-reading children using musical elements rather than linguistic tasks. Current dyslexia screening tools rely on reading and writing abilities — linguistic elements like letter knowledge, phonological awareness, and reading comprehension — meaning they can only be used with children who have already acquired some reading skills, typically after the first year of school. This often results in late detection, delaying critical early intervention. The authors exploit research showing that people with dyslexia have auditory and visual perception difficulties linked to short-term memory problems, and that differences in musical perception (rise time, duration, intensity/loudness, and frequency/pitch) have been found between children with and without dyslexia. DysMusic adapts the classic visual Memory card game into an auditory version: instead of matching pictures, children flip digital cards and try to find pairs with matching sounds. The game has four tasks corresponding to four musical parameters (frequency, length, rise time, and rhythm), each with subtasks using 4 or 6 cards. Musical samples are generated as simple sinusoidal waveforms using Audacity, with carefully controlled parameters. Cards have no visual distinguishing features to avoid confounds like color blindness.
Key findings
A usability test with 10 participants (5 children aged 3-9 and 5 parents aged 35-40) using the think-aloud protocol on an Android tablet revealed that the game was well-understood and engaging — children showed more enjoyment than parents, smiling and laughing while playing. All participants found the first subtask of each musical element equally difficult regardless of the parameter, likely because it was their first exposure; by the second subtask they had familiarized with the parameter and could often name it. Length and Frequency were the most difficult musical elements to distinguish, while Rhythm was easier. The youngest child (age 3) had major difficulty with 6-card subtasks and could not find matching pairs, suggesting that younger children should only use 4-card versions. Key design changes from usability feedback included switching from HTML5 video tags to YouTube iframe embedding to prevent loading issues, adding spoken positive feedback after each matched pair, and reducing the footer size. The next planned steps include adding visual elements to the game and conducting experiments with 30 participants across English, Spanish, and German to validate whether musical and visual elements can distinguish between children with and without dyslexia.
Relevance
This research addresses a critical gap in dyslexia detection: the inability to screen pre-reading children who have not yet developed the linguistic skills that current screening tools require. The language-independent approach through musical perception is particularly significant, as it could theoretically be deployed globally without translation — only the game interface needs localization, not the assessment itself. For accessibility practitioners, DysMusic exemplifies how web-based serious games can serve as accessible, scalable screening tools that are more engaging and less stressful than traditional clinical assessments. The connection between music perception and phonological processing (which underlies dyslexia) is well-established in the literature but had not previously been applied to pre-reader screening in a game format. Key limitations include the early prototype stage with no dyslexia classification results yet, the small usability sample, and the need to validate whether the musical tasks actually discriminate between children with and without dyslexia risk. The work builds on the same research group's Dytective game, which achieved 83% accuracy for detecting dyslexia using linguistic elements.
Tags: dyslexia · screening · serious games · gamification · music · auditory perception · pre-readers · children · phonological awareness · web-based assessment · usability testing · language independence · early detection · learning disabilities