Verb impairment in aphasia: A priming study of body-part overlap

Abstract

Background: Verb production impairments are well documented in aphasia, especially in non-fluent aphasias with lesions of the left frontal lobe. Evidence is inconclusive about whether the impaired verb production is accompanied by inefficient processing and comprehension of verbs. It is unknown if specific semantic features of verbs, such as knowledge of the effector (body-part used to implement the action), are automatically activated by aphasic individuals during verb processing. This question is especially relevant given the growing evidence that unimpaired speakers mentally simulate the action portrayed by a verb, and action simulation is an inherent part of language comprehension (embodied action semantics). There is additional evidence that reading or listening to verbs depicting hand, mouth, or leg actions activates frontal motor areas associated with implementing hand, mouth, and leg actions respectively in unimpaired individuals (semantic somatotopy). Aims: This study examined on-line verb processing in nonfluent aphasic participants who are impaired in production of verbs. We investigated (1) whether somatotopic (body-part) information is automatically accessed when processing verbs, (2) how the time course of this activation influences processing of other somatotopically related verbs, and (3) whether somatotopic information that is devoid of action context (as in line drawings of body-parts) can affect downstream verb processing. Methods & Procedures: A visual-visual lexical priming paradigm was used. In Experiments 1 and 2, the prime–target verb pairs were somatotopically congruent, incongruent, or neutral nonword (licking–kissing; licking–clapping, xxxx–clapping respectively), and the stimulus onset asynchrony (SOA) was short or long. Instead of verbs, the primes in Experiment 3 were line drawings of congruent or incongruent body-parts. Six nonfluent agrammatic aphasic individuals with verb retrieval difficulties participated in the study. Outcomes & Results: Aphasic participants were highly accurate in their responses. Lexical decision speed was influenced by the preceding primes in a specific pattern: somatotopically congruent primes caused a delay (interference) in response times, and this diminished with longer SOA. Incongruent primes had no effect on response times. Picture primes caused a general delay in lexical decision speed, but this was not a somatotopically specific effect. This pattern of results is identical to that observed with unimpaired participants. Conclusions: This group of verb-impaired aphasic individuals was able to automatically (and rapidly) activate somatotopic features of verbs, showing little evidence of impaired lexical-semantic representations. Hence verb processing and verb naming were found to dissociate. In addition, this study extends our understanding of language processing by showing that actions are simulated by the human brain, even when verbs are encountered as de-contextualised single words. Further, somatotopic information is necessary, but not sufficient, for action simulation.