This research program proposes to apply a connectionist model of visual word recognition (the """"""""triangle"""""""" model of Seidenberg and McClelland, 1989, hereafter SM89) to various aspects of developmental and acquired dyslexias, and explore the impact of various remediation techniques. Contrary to more traditional models which posit the accessing of entries from a mental lexicon, this model proposes that word recognition is a process in which patterns of activation are computed over orthographic, phonological and semantic representations. This model raises a natural theoretical question, namely, what is the division of labor between """"""""direct"""""""" access to meaning and """"""""phonologically mediated"""""""" access to meaning from print. The division of labor within the model is crucial to accounts of the role of phonology in reading acquisition, and to developmental and acquired dyslexias. To date, various subcomponents of the full model have been computationally explored in isolation (spelling to sound, SM89; and spelling to meaning to sound Plaut and Shallice, 1993), and factors influencing the division of labor have been explored with respect to normal reading (Harm, 1998). But no large scale version of the full model has been applied to developmental or acquired dyslexia. Lacking explicit simulations of the role of phonology in both normal and impaired reading, it is difficult to posit effective remediation techniques for abnormal reading in the face of damage. Further, lacking an explicit demonstration of the triangle model's accounts of various dyslexias, it is difficult to assess the validity of the theory as a whole. I propose to combine and extend the various implemented models within this framework into a larger model capable of testing specific theoretical accounts of the breakdown and recovery of literacy. Empirical experiments will be used to test predictions of earlier modeling efforts and constrain model development. This work will significantly advance theories of the acquisition, skilled use, breakdown and recovery of literacy.
