General deficits of both linguistic and non-linguistic visual statistical learning in Chinese children with developmental dyslexia

Purpose: Statistical learning is crucial to language acquisition. Impairments in statistical learning likely lead to reading difficulties in individuals with developmental dyslexia (DD), as supported by recent advances. However, whether the statistical learning deficits in individuals with DD are specific to the linguistic domain remains elusive. We examined the domain specificity of statistical learning deficits among Chinese children with DD. Method: Twenty-four Chinese children with DD (11 females) and 24 age-matched controls (12 females) participated in this study. All children were third-grade and native Mandarin speakers. We used the triplet-learning paradigm with a familiarization phase (online learning) and a test phase (offline learning) to examine the visual statistical learning (VSL) of both linguistic and non-linguistic regularities. Results: Children with DD demonstrated significantly slower online learning speed and lower offline learning accuracy compared to controls in the linguistic VSL task. Similar results were observed in the non-linguistic VSL task, where children with DD also performed worse than the control with reduced offline learning accuracy. Critically, linguistic VSL explained unique variances in word reading abilities, even after controlling the effects of reading-related skills, such as phonological awareness, morphological awareness, and working memory. Conclusion: These findings together suggest that VSL deficits in Chinese children with DD are domain-general, and statistical learning plays a significant role in reading.