An investigation of argument structure processing in normal and aphasic participants: a test of the argument structure coplexity hypothesis (Thompson 2003)

The study was conducted so as to test the mental organization of verb argument structure (VAS) and the generalization of the predictions based on the Argument Structure Complexity Hypothesis (ASCH; Thompson, 2003), which explains the pattern of impairment of the agrammatic patients’ verb production as a function of the argument structure complexity, i.e. both in terms of the number of arguments taken by a verb (transitive vs. unergative verbs) and of the presence of syntactic movement (unaccusative vs. unergative verbs). The aim was i) to test the effect of the number of arguments in a task tapping lemma access bypassing overt production, in both neurologically unimpaired (Experiment 1 and 2) and aphasic participants (Experiment 3), and ii) to test the effect both of the number of arguments and of syntactic movement in a patient with deep dyslexia (Experiment 4). Moreover, an additional goal was to inform about the deficit underlying the verbspecific impairment that characterizes nonfluent aphasia, by testing the hypothesis of a defective lemma access as compared to a deficit at the level of grammatical encoding (Experiment 3). A third goal was to test the assumption of a separate level of representation for semantic and VAS information, as suggested by models of contemporary psycholinguistics (Experiment 2). In Experiment 1, 2 and 3 participants were asked to perform a sentence completion task by choosing among two verb options that differed either in the VAS (unergative vs. two-place transitive verb) or in the verb semantic content (Condition 2 of Experiment 2 only). Experiment 4 provided instead a deep dyslexic patient with a word naming task that was performed both on verbs belonging to different categories (unergative, unaccusative, transitive) and on nouns. Results from Experiment 1, 2 and 3 demonstrate that access to VAS information is faster (for healthy subjects) and easier (for aphasic patients) for unergative than for transitive verbs, thus suggesting the possibility that the ASCH more generally reflects an aspect of normal language processing. In addition, results from Experiment 4 offer support only to the first prediction of the ASCH, suggesting that the effect of the number of arguments and the effect of syntactic movement arise at different level of processing. Finally, data from Experiment 2 demonstrate that VAS and semantic information can be accessed independently, in line with models of contemporary psycholinguistics.