| |
Abstract:
Abstract: Strong behavioral evidence links deficits in
phonological processing, to impairments in learning to read,
raising important questions about effective paths to remediation. A
connectionist model of, developmental dyslexia (Harm and
Seidenberg, 1999) was used to investigate how endogenous
phonological impairments affect beginning reading. Phonological
impairments in the model led to the development of less
componential internal representations, which impairs learning
spelling-sound correspondences. We then examined the effects of
behavioral interventions that are being used in large-scale studies
specifically directed at remediating phonological processing. These
interventions had little impact on model performance. However, a
remediation scheme targeting spelling to sound correspondences by
emphasizing the componential structure of words (McCandliss et al.,
1999) produced improvements in the model's nonword reading
comparable to those observed in studies of children. The
effectiveness of the intervention was linked to a restructuring of
the internal, "hidden" representations that map visual words to
their phonological representations: these "hidden" representations
became more componential in nature. The model explains why
interventions involving phonological training alone may not be
successful in addressing reading impairments, and how such efforts
might be improved by interventions that emphasize the systematic,
componential properties inherent in the writing system.
|
