Sladjana Lukic ^1* Jiang Fei ² Maria Luisa Mandelli ^3,4 Ting Qi ⁵ Sarah A Inkelis ^3,4 Emily Rosenthal ^4,6 Zachary Miller ^3,4 Emma Wellman ^3,4 Silvia A Bunge ^4,6,7 Maria Luisa Gorno Tempini ^3,4 Christa Pereira ^3,4

• ¹ Ruth S. Ammon College of Education and Health Sciences, Department of Communication Sciences and Disorders, Garden City, Kansas, United States
• ² Department of Epidemiology and Biostatistics, Medical Center, University of California, San Francisco, San Francisco, California, United States
• ³ Memory and Aging Center, Medical Center, University of California, San Francisco, San Francisco, California, United States
• ⁴ Berkeley Schwab Center for Dyslexia and Cognitive Diversity, University of California, San Francisco, San Francisco, California, United States
• ⁵ Beijing University of Posts and Telecommunications (BUPT), Beijing, Beijing Municipality, China
• ⁶ Department of Psychology, University of California, Berkeley, Berkeley, California, United States
• ⁷ Helen Wills Neuroscience Institute, College of Engineering, University of California, Berkeley, Berkeley, California, United States

Introduction: Most studies of dyslexia focus on domains of impairment (e.g., reading and phonology, among others), but few examine possible strengths. In the present study, we investigated semantic fluency as a cognitive strength in English-speaking children with dyslexia aged 8-13. Methods: Ninety-seven children with dyslexia completed tests of letter and semantic verbal fluency, standardized measures of reading and cognitive functions, and task-free resting-state functional magnetic resonance imaging (rs-fMRI). First, we adjusted performance on semantic fluency by letter fluency and created a residual score that was used to separate participants into high (residual > 0) or average (residual < 0) semantic performance groups. We then employed a psycholinguistic clustering and switching approach to the semantic fluency task and performed dynamic task-free rs-fMRI connectivity analysis to reveal group differences in brain dynamics. Results: High and average semantic fluency groups were well-matched on demographics and letter fluency but differed on their psycholinguistic patterns on the semantic fluency task. The high semantic fluency group, compared to the average semantic fluency group, produced a higher number of words within each cluster, a higher max cluster size, and a higher number of switches. Differential dynamic rs-fMRI connectivity (shorter average dwell time and greater brain state switches) was observed between the high and average groups in a large-scale bilateral frontal-temporal-occipital network. Discussion: These data demonstrate that a subgroup of children with dyslexia perform above average on semantic fluency tasks and their performance is strongly linked to distinct psycholinguistic patterns and differences in a task-free resting-state brain network, which includes regions previously implicated in semantic processing. This work highlights that inter-individual differences should be taken into account in dyslexia and reveals a cognitive area of strength for some children with dyslexia that could be leveraged for reading interventions.