The proposed research seeks to evaluate two recent hypotheses about the source of phonological errors in the speech of fluent aphasics: (1) Wernicke's aphasics exhibit disrupted processing within the phonological lexicon (Ellis, 1985), and (2) conduction aphasics exhibit disrupted phonemic string construction (Nespoulous et al., 1986). While each of these hypotheses appears to account for some aspects of phonological dysfunction in fluent aphasia, their generality is called into question by the variability of clinical symptoms associated with these syndromes. The proposed research will use longitudinal data to evaluate the above hypothesis. Wernicke's, conduction, and transcortical sensory aphasics will be examined with a battery of phonological encoding and decoding tests administered several times over the first 6 months post onset. To create more homogeneous experimental groups, each subject will be given a subdiagnosis on the basis of spontaneous speech (e.g., Wernicke's aphasia with semantic jargon). A clinically-based coding system for classifying production errors will be used both to determine this subdiagnosis and to identify the portion of the production data that will submitted to detailed phonological analysis. A model of word production will serve as the basis for generating the phonological analyses and for interpreting all data. It is anticipated that, by examining changes in phonological error patterns over time, the core behavior(s) needed to more precisely diagnose fluent aphasics will be identified, and, competing interpretations of error patterns will be disentangled. Better understanding of the evolution of fluent aphasic symptomatology will also provide invaluable information for the prognosis and rehabilitation of aphasic patients.
