Semantic fluency difficulties in developmental dyslexia and developmental language disorder (DLD): poor semantic structure of the lexicon or slower retrieval processes?

BACKGROUND: Children with dyslexia and/or developmental language disorder (hereafter children with DDLD) have been reported to retrieve fewer words than their typically developing (TD) peers in semantic fluency tasks. It is not known whether this retrieval difficulty can be attributed to the semantic structure of their lexicon being poor or, alternatively, to words being retrieved more slowly despite semantic structure being intact. AIMS: To test two theoretical models that could potentially account for retrieval difficulties in semantic fluency tasks, namely, the Poor Lexical-Semantic Structure Model and the Slow-Retrieval Model. Both models predict that children with DDLD will retrieve fewer items compared with TD children. However, while the Poor Lexical-Semantic Structure Model predicts a less sophisticated network of semantic connections between words in the lexicon, as evidenced by smaller clusters of related items in children with DDLD, the Slow-Retrieval Model predicts intact inter-item associations in the lexicon, as evidenced by the two groups' clusters being of a similar size. The groups' semantic fluency performance was therefore compared. How semantic fluency performance related to children's language, literacy, and phonological skills was also investigated. METHODS & PROCEDURES: A total of 66 children with DDLD aged 7-12 years and 83 TD children aged 6-12 years, all monolingual Greek speakers, were tested on semantic fluency, using the categories 'animals', 'foods' and 'objects from around the house'. The numbers of correct and incorrect responses, clusters and switches, and the average cluster size were computed. Children were also assessed on non-verbal IQ, language, literacy and phonological tasks. OUTCOMES & RESULTS: In both groups, productivity in semantic fluency tasks correlated strongly with the numbers of clusters and switches, but not with average cluster size. The DDLD group produced significantly fewer correct responses and fewer clusters compared with the TD group, but the two groups showed similar switching and average cluster size. Children's language, literacy and phonological skills significantly predicted the number of correct responses produced, beyond the significant effect of age. CONCLUSIONS & IMPLICATIONS: We conclude that poorer semantic fluency performance in children with DDLD results not from a lexicon with poor semantic structure, but rather from slower retrieval processes from a lexicon with intact semantic structure. The underlying causes of slow lexical retrieval still need further investigation.