Slide Slam P13 Sandbox Series
Understanding implausible passives in aphasia: algorithmic and heuristic processes compete
Caitlyn Antal1, Djuly Trankarov1, Kyan Salehi1, Alexa Ray Falcone1, Laura Pissani1, Roberto G. de Almeida1; 1Concordia University
INTRODUCTION Given implausible passive sentences such as “The dog was bitten by the man”, native speakers sometimes report incorrectly that “the dog" is the doer of the action. Similar effects have been used to illustrate several overlapping accounts of language comprehension such as “pragmatic normalization” (Fillenbaum, 1974), “good enough” interpretation (Ferreira, 2003), and “rational inference” (Gibson et al., 2013). Yet, the nature of these errors in interpretation remains unclear: Is it a failure of grammatical—algorithmic—processes to yield a structure that is faithful to the correct interpretation of the sentence? Is it due to higher-order processes driven by heuristics (e.g., pragmatic) thus neglecting what grammar outputs? Or does the interpretation fail because of a combination of both algorithmic and heuristic processes? Some studies have shown greater difficulty to process implausible sentences in individuals with Broca’s aphasia (e.g., Caramazza & Zurif, 1976) given their difficulty with syntactic (algorithmic) computations. Others, however, suggest that “persons with aphasia in general” (Gibson et al., 2015) have difficulty with implausible sentences due to “greater noise”. We investigated dissociations between sentence type and aphasia etiology aiming to understand the source of implausible passive deficit and what they inform us about the unimpaired language comprehension system. METHOD Participants were 6 non-fluent [NF], 4 fluent [FL], 3 mixed but predominantly non-fluent [MN], 2 with mixed aphasia [MX], and 42 healthy controls. In each trial, a sentence was aurally presented and immediately followed by two pictures on a computer screen. Participants had to choose which picture best represented the sentence they heard. The 24 experimental quartets varied in voice (active, passive) and plausibility (plausible, implausible): (a) active plausible (The dog bit the man), (b) active implausible (The man bit the dog), (c) passive plausible (The man was bitten by the dog), and (d) passive implausible (The dog was bitten by the man). Pictures for the sentences above were that of a dog biting a man (correct for (a) and (c)) and a man biting a dog (correct for (b) and (d)). RESULTS Repeated measures ANOVAs showed a main effect of group, voice, and plausibility, as well as all first and second order interactions. Overall, group analyses showed that NF and MN individuals performed worse than controls on all sentence types. We also found an effect of plausibility, whereby NF and MF individuals had significantly more difficulty choosing the correct picture when presented with implausible sentences. These impairments were not found in FL individuals. Case-series analyses will also be presented. DISCUSSION Our group analyses suggest that the underlying nature of the impairments for NF and MN individuals may stem from a deficit building structures corresponding to event descriptions. The difficulty shown by the NF and MN groups with passive and implausible sentences suggests that they have an impairment computing the syntactic (algorithmic) processes fundamental to language comprehension. We further suggest that this deficit may be coupled with problems computing thematic structures for noncanonical sentences leading to erroneous choices based on heuristic (viz., pragmatic) processes.