Spoken word to picture matching from PALPA: A critique and some new matched sets

Background: PALPA (Psycholinguistic Assessments of Language Processing in Aphasia; Kay, Lesser, & Coltheart, 1992) is a widely used clinical and research tool. Subtest 47, Spoken word-picture matching, requires the individual with aphasia to listen to a spoken word and correctly choose from five distractor pictures (target, close semantic, distant semantic, visually related, and semantically unrelated). It contributes diagnostically to the clinical evaluation of semantic processing. The authors claim that, first, errors on this test indicate that a semantic comprehension problem is present, and second, that distractor choice reflects the semantic specificity of the problem. For accurate clinical assessment the validity of these claims must be evaluated. Aims: This paper aims to evaluate the internal validity of PALPA spoken word-picture matching. It addresses two questions; first, is the relationship between the target and distractors what the authors claim it to be? Second, what is the relationship between the target and distractor stimuli in relation to a number of psycholinguistic variables? In addition it allows the clinician to examine the effects of individual variables on performance by including matched subsets of stimuli from this test (matched across five psycholinguistic variables: frequency, imageability, number of phonemes, semantic and visual similarity, word association). Methods and Procedures: Target and distractor relationships were investigated (in terms of semantic and visual similarity and word category) and psycholinguistic variables (including word frequency, word association, imageability, number of phonemes, semantic and visual similarity). Outcomes and Results: Analysis revealed a number of confounds within this test: close semantic distractors were not only more semantically similar but also more visually similar to their targets than distant semantic distractors; the semantic and visual (SV) close semantic distractors were more semantically similar to their targets than the non-SV close semantic distractors; targets and distractors did not bear a consistent categorical relationship to their targets, and there were significant intercorrelations between variables for these stimuli (e.g., frequency and length; semantic/visual similarity and length). Conclusions: The authors' claim that this test assesses semantic comprehension is certainly still tenable. Individuals making errors on this test have a high probability of some semantic processing deficit. However, this study shows that the test fails to assess the nature of the semantic processing deficit, as error patterns are subject to the effect of confounding factors. In its current form clinicians should exercise caution when interpreting test findings and be aware of its limitations. The development, here, of matched subsets of stimuli allows performance to be re-evaluated in terms of the influence of semantic and visual similarity, imageability, frequency, word length, and word association.