Statistical Learning in Children With/Without Dyslexia: Impacts of Working Memory and Item Characteristics

• In Event: Learning How the Brain Acquires Statistical Regularities: Implications for Education

Abstract

Objectives: This study examined whether statistical learning difference in children with and without development dyslexia (DD) is associated with working memory, and if so, how this association varies in terms of item characteristics (levels of consistencies or statistical regularities).

Theoretical Framework: The statistical learning and reading (SLR) model assumes that working memory dynamically interacts with long-term memory systems, both of which compute statistical properties of input items. Meanwhile, item characteristics, such as levels of consistency or statistical regularities, influence the involvement of different memory systems. For instance, processing of novel and unexpected items requires working memory. Increasing evidence suggests that children with DD exhibit limited working memory capacity and specific difficulties in processing low consistency and irregular items. Thus, we hypothesize that difficulties in processing irregular items displayed by children with DD are associated with their under-utilization of working memory during statistical learning.

Methods: By testing hypothesis, this study administered a verbal working memory span task and a statistical learning (SL) of artificial orthography experiment to 150 Chinese second- and third-graders with DD (Mean age = 8 years 4 months, SD = 9 months) and 150 age-matched typically developing controls (Mean age = 8 years 3 months, SD = 8 months).

Materials: In the learning phase of the SL experiment, participants were implicitly exposed to invented left-right and top-bottom-structured characters comprising a control and a target radical carrying high (100%), moderate (80%), or low (60%) positional consistency. The testing phase consisted of a recognition test in which participants were asked to decide whether a newly presented character appeared in the learning phase (i.e., old tested form) or not (i.e., new tested form). Besides, based on whether the target radical appeared in its primary or secondary position, each logographic character was categorized into a regular or irregular item, respectively.

Results: Logistic regression analyses revealed that despite their comparable recognition accuracy with typically developing children, there was a lack of significant association between working memory and false recognition on the irregular item with moderated positional consistency and new tested form for DD children. In contrast, regardless of reading status (i.e., dyslexia or not), working memory was positively associated with recognition accuracy on high and low consistency items. These results indicate that children with DD might exhibit deficits in utilizing working memory to cope with irregular items during the acquisition of quasi-regularities.

Significances: These findings suggest that working memory plays a critical role in acquiring quasi-regularities of print input, and that statistical properties of words, especially the quasi-regular items should be prioritized for teaching children with dyslexia.

Authors

• Puyuan Zhang, University of Hong Kong
  Presenting Author

• Shelley Xiuli Tong, University of Hong Kong
  Non-Presenting Author