| Popis: |
In their seminal article Castles and Coltheart (1993) argued that the group of disabled readers may not be a homogenous group. They found evidence for the existence of two subtypes, phonological and surface dyslexics. These findings were, however, criticized on methodological grounds. It has been argued (e.g. Stanovich, Siegel, & Gottardo, 1997) that comparing disabled readers to a normative sample of children of the same age, as in Castles and Coltheart, may yield processing trade-offs between sublexical and lexical paths that depend on the overall level of word reading skill. To this end, reading-level match controls were used in a new series of subtyping studies. In this contribution we show that the regression method of Castles & Coltheart as well as the reading-level match method lack reliability, and additionally the latter method is shown to be biased. We propose an alternative method based on state trace analysis that is both reliable and unbiased. Method Disabled readers' individual state trace curves were based on the reading performance on word and nonword reading lists across several conditions differing in difficulty. Word difficulty was manipulated by varying the number of syllables, number of consonant clusters and body frequency. To determine subtypes parameters of individual state trace curves were compared to the overall distribution of parameters. Results and Discussion Individual state trace curves of 73 Dutch disabled readers were calculated. Although the majority of disabled readers could not be categorized within one or the other subtype, the disabled readers belonging to a subtype showed a symmetrical pattern (as many phonological subtypes as surface subtypes). |
