Frontiers in Psychology (May 2015)
Electrophysiology of Sentence Processing in Aphasia: Prosodic Cues and Thematic Fit
Abstract
Introduction: The building of semantic and syntactic representations during sentence processing can be influenced by probabilistic cues such as context, plausibility, and prosody (Garnsey, Pearlmutter, Myers & Lotocky, 1997; Kjelgaard & Speer, 1999). Consider the following: 1. While the band played the song pleased all the customers. This sentence contains a temporary direct object (DO)/sentential complement (SC) syntactic ambiguity because the verb “played” is an optionally transitive verb so the listener does not initially know whether the NP “the song” is the DO of the verb “played” or whether it is the subject of the main clause. This can be immediately disambiguated with the addition of a pause after the verb “played” signaling the presence of a syntactic boundary. Lexical-semantic information may also interact with syntax during sentence processing. For example, several studies have found that healthy controls are sensitive to verb transitivity and thematic fit plausibility (Mitchell, 1987; van Gompel & Pickering, 2001) such that processing is momentarily disrupted when a verb is followed by an implausible NP. Using event-related brain potentials (ERPs), we aim to examine how persons with aphasia (PWA) use prosodic and lexical-semantic information. Prosody was manipulated to either facilitate (Sentence 2a & 2b) or disrupt (Sentences 2c & 2d) syntactic processing. Thematic fit conflict between the verb (e.g. “played”) and the structurally ambiguous NP (e.g. “song”/“beer”) was also manipulated. 2a. [While the band played] the song pleased all the customers. 2b. [While the band played] the beer pleased all the customers. 2c. [While the band played the song] pleased all the customers. 2d. [While the band played the beer] pleased all the customers. * [ ] Indicates prosodic contour Methods: Twenty-four healthy college-age control participants (YNCs) and ten adults with a Broca’s aphasia participated in this study. Each sentence was presented aurally to the participants over headphones. ERP Data Recording & Analysis. ERPs were recorded from 32-electrode sites across the scalp according to the 10-20 system. ERPs were averaged (100ms prestimulus baseline) from artifact free trials time-locked to critical words (i.e., the point of disambiguation “pleased” in the prosodic comparison, and the NP “the song”/”the beer” in the semantic comparison). Mean amplitudes were calculated in two windows: 300-500ms for the N400 effects and 500-1000ms for the P600 effects. Results: The data from our YNCs revealed a biphasic N400-P600 complex in the prosody comparison (Figure 1A). We also found an N400 effect immediately at the NP in the incongruent relative to congruent thematic fit comparison. For the prosodic comparison in the PWA group, a delayed N400 effect was found one word downstream relative to the YNC data in the prosody comparison (Figure 1B). Additionally, an N400 effect was observed in the thematic fit comparison. Discussion: The results suggests that PWA possess a delayed sensitivity to prosodic cues, which then may affect their ability to recover from misanalysis from an incorrect parse. The results also indicate that PWA are sensitive to thematic fit information and have the capacity to process this information similarly to YNCs.
Keywords
