An Auditory-Perceptual Rating of Connected Speech in Aphasia

Abstract

Purpose: The goal of this study was to develop a novel tool for connected speech analysis in aphasia, so that spoken output can be characterized in a data-driven and explanatory manner. Method: We designed a multidimensional rating scheme called the Auditory-Perceptual Rating of Connected Speech in Aphasia (APROCSA), in which 27 common features were each rated on a 5-point scale. Three researchers and twelve student clinicians rated 24 connected speech samples from the AphasiaBank database. Results: Ratings conducted by both researchers and student clinicians demonstrated good-to-excellent reliability and strong concurrent validity with AphasiaBank measures derived from transcriptions, clinical measures, and subscores from the Western Aphasia Battery (WAB). Factor analysis revealed that four underlying factors—Paraphasia, Logopenia, Agrammatism, and Motor speech—accounted for 79% of the variance in the connected speech profiles. Examination of individual patient scores showed considerable diversity of factor scores among patients of any given aphasia subtype. Conclusions: The APROCSA proved to be a reliable, valid, and efficient tool for research or clinical purposes. The preliminary findings of the factor analysis suggest a parcellation of non-fluency into three distinct profiles—Logopenia, Agrammatism, and Motor speech—which may occur in conjunction with other non-fluent profiles or with the fluent profile