Verb argument structure effects in aphasia are different at single-word versus sentence level

Abstract

ABSTRACTBackground: Many aphasia assessments and therapies select and/or sequence verbs based on linguistic complexity of their verb argument structure (VAS). However, further empirical testing is needed to fully understand whether and how VAS parameters affect the cognitive difficulty of verb processing in different tasks and contexts.Aims: The study investigated whether more linguistically complex VAS universally implies more cognitively difficult verb processing, as predicted by the Argument Structure Complexity Hypothesis. We hypothesized that this would only be the case at sentence level, whereas in single-word tasks more linguistically complex VAS would facilitate lexical access via lexico-semantic associations with potential arguments, contrary to the Argument Structure Complexity Hypothesis.Methods and Procedures: The effects of three VAS parameters (number of arguments, number of valency frames, canonicity of thematic role marking) were tested in two tasks (single-word naming and cued sentence production) in two aphasia types (fluent and non-fluent, 20 participants per group). We analyzed how VAS parameters affected accuracy and latency in naming and canonicity and well-formedness in sentence production.Outcomes and Results: As hypothesized, VAS effects were different at the sentence versus single-word level. In sentence production, one VAS parameter (the number of arguments) showed the expected negative effect of greater linguistic complexity on sentence canonicity. No other comparisons were significant, likely due to ceiling effects. In contrast, in single-word naming, VAS effects were mixed. The number of valency frames showed the predicted facilitatory effect of linguistic complexity. The number of arguments showed a non-significant statistical trend in the same direction. For canonicity of thematic role marking, greater complexity had a negative effect on naming, possibly because it does not affect the number of lexico-semantic associates of the verb. All findings pertained to both non-fluent and fluent aphasia.Conclusions: When accounting for VAS parameters in selection or sequencing of verbs for assessment or therapy, the cognitive difficulty of verb processing should be estimated for a particular task. In sentence-level grammatical processing, most cognitively difficult verbs are those with more linguistically complex VAS. In contrast, in single verb retrieval, verbs with more linguistically complex VAS may be cognitively simpler if their richer lexico-semantic associations with potential arguments provide extra routes of lexical access.