Category-Specific Lexical Dissociations

Abstract

Selective impairment or selective sparing of lexical categories is encountered repeatedly in the clinical examination of aphasic patients as well as in patients who have partially recovered from the effects of herpes encephalitis. These dissociations may affect a narrowly defined semantic category (e.g., body parts) or a broadly defined one (naturally occurring versus manmade objects). They may affect specific symbol systems (comprehension of letter-names or of number names), or psycholinguistically defined dimensions (particular parts of speech; proper vs common nouns, abstract versus concrete nouns). The dissociations observed in aphasic patients usually involve only a single modality (e.g., body part impairment only in auditory comprehension; grammatical morpheme impairment only in tasks of oral reading). Those observed in postencephalitic patients sometimes appear to span all aspects of semantic knowledge. The very disparity among these lexical dissociations suggests that they may arise from a number of different mechanisms. Other evidence indicates that certain subsets of dissociations arise from closely related mechanisms. For example, a given category (e.g., alphabet letters) may be selectively impaired in auditory comprehension testing, but selectively well preserved in naming. As a different example, impairments in the identification of animate objects are in complementary distribution with impairments in the identification of manmade ones. Such commonalities are suggestive as to where to look for underlying mechanisms. Nevertheless, most explanatory accounts to date are highly speculative. In some instances, empirical evidence relates the dissociations to anatomic regions of injury; in others, there may be selective impairment in the processing of critical semantic properties that are common to the members of a category. In this article I shall review the specific types of lexical category dissociations, imposing an organization that appears to me to be motivated by the likelihood of common underlying processes.