Abstract

An understanding of language processing in humans is critical if realistic computerised systems are to be
produced to perform various language operations. The examination of aphasia in individuals has provided a large
amount of information on the organisation of language processing, with particular reference to the regions in the
brain where processing occurs and the ability to regain language functionality despite damage to the brain. Given the importance played by aphasic studies an approach that can distinguish between aphasic forms was devised by using a
Kohonen self-organising network to classify sentences from the CAP (Comparative Aphasia Project) Corpus. We demonstrate that the different distributions of words in aphasics types may lead to grammatical systems which inhabit different areas in self-organising maps.