Severity of impairment or specific types of aphasia: an empirical investigation

This study examined empirical means of clustering aphasic patients into relatively homogenous and identifiable groups. The method included: (1) examining the intercorrelations of the subtests of the Neurosensory Comprehensive Examination for Aphasia by the means of factor analysis; (2) allowing the subjects to group themselves in homogenous subgroups by means of hierarchical cluster analysis; and (3) comparing the profiles of these groups on the derived dimensions. Four factors accounting for 79.3% of their total test variance were derived which reflected attention/memory; reading ability; verbal-naming ability; and elementary writing skills. With five groups the hierarchical grouping analysis indicated that only 21.5% of the total variability could be ascribed to within group variability. An analysis of variance of the four factor scores found that 66%, 93%, 81% and 79% of the between group variance of each of the respective factors was associated with group membership and the accuracy of membership prediction was 94.1%. Four groups appeared to be distributed on the basis of a severity model of impairment. One group representing 6.3% of the total sample demonstrated a pattern of specific impairment involving a mild level of impairment on attention/memory tasks, a moderate level of impairment on verbal naming tasks and writing skills, and a profound level of impairment on measures of reading skills. The high inter-factor correlations indicated the general nature of language functioning. The results were discussed in terms of the concept of aphasia as a general cognitive deficit.