Robin van Rijthoven

36 Chapter 2 The present study To sum up, it has been made clear that orthography, phonology, and semantic abilities are of main importance for learning to read words. Throughout the grades, children develop full word representations in order to become fluent readers. Although children with dyslexia are in need for more compensation in phonological recoding, it remains unclear for this specific group of readers whether the full semantic lexicon (depth and broadness of the lexicon) could partly compensate for a weak phonological component in building orthographic representations. It is still unclear whether the role of semantic abilities is in assigning phonology to new orthographic representation as tapped with pseudoword reading or in addressing phonology in direct word recognition. An indirect effect of semantic abilities via phonological awareness and rapid naming on word and pseudoword decoding could be predicted in the former case, a direct effect of semantic abilities on word decoding in the latter case. Therefore, in the present study, we investigated the direct and indirect effects of semantic abilities on pseudoword and word decoding within a group of Dutch children with dyslexia, taking into account phonological awareness and rapid naming. In contrast with previous research, in the present study, semantic abilities is defined as much broader than just vocabulary. It is generally known that semantic abilities involves more than vocabulary alone, and especially because the specificity and redundancy of the lexical representations seems to be important (Perfetti & Hart, 2002), a broad operationalization definition of semantic abilities was used, including both lexical knowledge and comprehension skill. Starting from the question to what extent semantic abilities in children with dyslexia would foster their decoding skills, we expected to find pseudoword and word decoding to be both directly and indirectly (via phonological awareness and rapid naming) predicted from semantic knowledge. Method Participants Participants were Dutch children diagnosed with dyslexia who received an in‐service reading and/or spelling intervention in a clinic for assessment and intervention for children with learning difficulties. For the purpose of this study, 99 files of Dutch children were collected from the clinic. Due to missing data and different instruments, a group of 55 children (36 boys and 19 girls) with Dutch as their first language remained for this study. For an anticipated medium effect size (in terms of Cohen’s f 2) of 0.15 for each path including three predictors (semantic abilities, phonological awareness, and rapid automatized naming), a representative group of 55 participants was needed to have sufficient power (>0.80) at the .05 level (two tailed) according to calculations as recommended by Faul and colleagues (2007).