The specific nature of the statistical learning deficit in developmental dyslexia: Impaired temporal and spared...

While extensive research highlights the importance of statistical learning in language acquisition and processing, its relationship to developmental dyslexia, a common developmental language disorder, remains unclear. Studies of statistical learning in dyslexia often rely on specific tasks that most likely do not cover the scope of statistical learning as a multi-componential construct, making it difficult to discover what aspects of learning are affected in dyslexia. The present study provides the first systematic examination of two major facets of statistical learning in dyslexia. We investigated statistical learning performance among individuals with and without dyslexia using a 2 × 2 factorial design that systematically manipulated both the type of statistical relation between elements (non-spatial vs. spatial) and the structure of statistical input (conditional vs. distributional). Neurotypical participants demonstrated learning across all tasks, whereas individuals with dyslexia showed significantly reduced performance when statistical regularities involved non-spatial relations, though not when they involved spatial relations. This deficit was apparent regardless of whether tasks involved conditional or distributional statistics. The findings suggest that rather than being generally weak in statistical learning, individuals with dyslexia face challenges in learning environments involving non-spatial statistical regularities that require integration of statistical computations across multiple exposures. Given the involvement of non-spatial aspects in printed and spoken language, these deficits can add to the challenges faced by individuals with dyslexia in acquiring linguistic representations. Additionally, the dissociation observed in dyslexia supports the notion that statistical learning comprises distinct sub-processes rather than a single unitary process.