The goal of this article is to illustrate the application of self-organizing dynamics in the design of a model of lexical access. We focus particularly on the mapping of sound structure on to the lexicon and the influence of that structure on lexical access. The approach is tested in a series of two sets of simulations that explicate how lexical access might occur in normal subjects and aphasic patients. Both sets of simulations address the behavioral effects of both phonological and phonetic variability of prime stimuli on the magnitude of semantic priming. Results show that the model can successfully account for the behavioral effects associated with several kinds of acoustic manipulation, competitor presence, and the unfolding of those effects over time—primarily because it balances three important control parameters: resting lexical activation, positive feedback, and negative feedback. These simulations are offered as support (in the form of an existence proof) that deficits in the degree of lexical activation can account for the lexical processing impairments shown by Broca's aphasics who have reduced lexical activation, and Wernicke's aphasics who have increased lexical activation. Overall, results suggest that the present approach promises to offer a coherent theoretical framework within which to link empirical evidence in language processing and cognitive neuroscience in terms of possible underlying mechanisms.