Abstract
Describing problems of relevance in clinical populations has been a subject of interest for both theoreticians and clinicians. The argument that conversational relevance is a product of listener interpretation is supported in the present paper. Study 1 examines the topic-shifting profiles of a social worker in conversation with normal elderly participants as compared to her profiles in conversation with participants with dementia. Study 2 compares semantic segments in conversations between this same social worker and a normal adult and a traumatically brain-injured adult. Qualitative differences were found in Study 1 in the types of topic shifts used, the possible reasons for these shifts, and the context to which the shifts related. In Study 2, differences were found in the types of relationships between semantic segments and in the relative proportion of implicit and explicit segments. Support is given for considering conversational partners' roles in determining relevance.