Research addressing cognitive deficits in developmental dyslexia is divided between phonological and visual processing camps, with little crosstalk between the two. However, reading- the activity impaired in dyslexia - is the faculty to access spoken language infonnation in written fonn, i.e., visual input. Therefore, understanding of how reading operates, and may be disrupted, requires the study of phonological and semantic processing from orthographic infonnation. Based on the theoretical literature on dyslexia as a reading deficit (Chapter One), and what we have learnt about nonnal reading from event-related potentials (ERPs; Chapter Two), the present thesis describes a programme of research investigating orthographic and phonological processing during reading with ERPs to characterise their interplay in dyslexia over time. The aim is to examine the extent to which these processes interact in dyslexia and if group differences are attributable to perceptual processes, attentional processes or both. Throughout the studies reported, phonological content of written words appears to be processed similarly in dyslexic and control adults until post-perceptual processing stages. In Study 1, which tested orthographic and phonological priming effects during phonological decisions on pseudoword-word pairs, phonological priming did not differ between dyslexic and control readers until a late stage associated with stimulus reprocessing, indexed by the P600 wave. The earliest orthographic similarity responses (Nl amplitude) were reduced, however, and interacted with phonological processing at a later stage (N2-P3 ERP range). Study 2 showed that, even in a highly-constraining sentence context, phonological priming can override orthographic mismatches in skilled readers since homophones and pseudohomophones of hi ghly expected words are
integrated similarly to best completion words during early phonological analysis (attenuated N2) and later semantic processing (attenuated N400). Following the rationale of Study 2, Study 3 compared the performance of dyslexic and control readers in a similar paradigm. Early phonological integration, indexed by N2, and conflicts between orthographic and phonological information, indexed by P2, were similar in the two groups but ERP modulations specific to phonologically acceptable misspellings failed to capture dyslexic participants attention to the same extent as it did in control participants in the P3a range. Thus, phonological processing differences seem to emerge at a late stage of orientation of attention towards phonological information after a ' normal' stage of perceptual decoding. Study 4 tested the reliability and generality of differences in attentional orienting responses using an adapted oddball design prone to elicit P3a effects. Replicating Study 3, responses to pseudohomophones were significantly attenuated in dyslexic readers, while P3a amplitudes elicited by deviants in a nonverbal control task were similar between the two groups of participants. Overall, a consistent pattern has emerged from these studies: Detection and integration of phonological information from written words is early and automatic in both nonnal and dyslexic readers. In dyslexics, the deficit in reading appears to emerge at the stage of attentional orientation to phonological information, particularly when it conflicts with orthographic information. It remains unclear however, whether this effect is accompanied by an overall reduced sensitivity to orthographic form. In sum, deficient interactions between phonological, orthographic, and attentional processes appear to be at the core of the reading deficit in dyslexia, and these cannot really be considered separately.