Abstract
Recent research suggests that the integrity of the right hemisphere is necessary for the processing of language in context (e.g. Foldi, 1987). A specific case of such processing is the ability to draw inferences from utterances. Studies show that patients with right hemisphere damage (RHD) have reduced ability to draw correct inferences from texts (e.g. Bryan, 1989). Currently there is little agreement as to the cause of this inability. This study adopts a Relevance Theory approach (Sperber and Wilson, 1986) to create a model of utterance processing, from which the cause of difficulties with inference might be deduced. The model serves as the basis for devising 60 bridging inferences on which six RHD subjects and 12 control subjects are tested. The inferences are of three types: textual, textually reinforced and encyclopaedic. The RHD subjects in this study performed worse overall, and in addition performed differentially on the three inference types. The implications of these results are discussed in the light of Relevance Theory, and it is suggested that RHD patients' problems with inference stem from an inability to use linguistic information from the text in the deductive system, coupled with an over-reliance on encyclopaedic information.