Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
The Role of Working Memory During Sentential Language Processing: Evidence from EEG
Poster B46 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port
Megan Nakamura1, Yanina Prystauka2, Eleonora Rossi1; 1University of Florida, 2UiT Arctic University of Tromsø
Recent work on neural oscillatory correlates of sentence processing suggests power decreases in alpha and beta can be associated with the processing of syntactic and semantic violations (Kielar et al., 2014). Some findings also suggest changes in alpha may reflect neural processes underlying Working Memory (WM) (Wianda & Ross, 2019). More recently, Rossi & Prystauka (2020) found that when individuals process grammatical information in their second language (L2), alpha and beta oscillations decline earlier in time than in native speakers. The authors hypothesized that processing in the L2 taxes WM resources, and that this reduction might be at the basis of the observed shortening of the oscillatory duration. The present study tests this theory by manipulating WM itself to create a ‘taxing condition’ while processing in the native language. Method: Participants: 55; (f =38; mean age = 19.7) native monolingual English speakers with no history of neurological and/or learning disorders. All participants completed a behavioral Operational Span (O-Span) (Turner and Engle; 1989) and completed an EEG Acceptability Judgment Task (AJT) which consisted of 60 correct sentences, 60 items with syntactic violations, and 60 items with semantic violations. Critically, participants were randomly assigned to two conditions: 1) Non-Memory condition: the AJT task was completed without the ongoing WM task. 2) Memory condition: the AJT task was completed with an additional ongoing WM task which required participants to memorize a sequence of three words at the beginning of each sentence, monitor the sentence for those words, and randomly recall if a specific word was present in the sentence. Preliminary Results: Linear mixed-effects model analysis was performed to examine mean amplitude EEG responses for the memory manipulation (non-memory, memory), sentence condition (correct, syntactic violation, semantic violation), and EEG ROI (region of interest). The model included fixed effects for Condition, Sentence Type, and ROI, as well as their interactions, while accounting for random effects as a within subject factor. The model was fitted using the REML criterion with the bobyqa optimizer. The results show significant main effects of Condition and sentence type and an interaction by Condition by sentence type. More specifically, the Non-Memory group shows overall larger P600s compared to the Memory group. For sentence type the data indicate a larger P600 for sentences containing syntactic violations. The interaction further demonstrated that the Non-Memory condition by syntactic violation overall had the largest mean amplitude. Additionally for the ROIs, the Central Posterior cluster was predictive of more positive mean amplitudes, which is in line with the typical scalp distribution of the P600. The Left and Right Anterior ROI as well as the Right Posterior ROI were predictive of overall more negative mean amplitude which is in line with the typical scalp distribution of the N400. The time-frequency analysis is currently underway, but we predict to observe the same decline in alpha and beta band for the memory condition in line with Rossi & Prystauka’s, (2020) L2 processing results, considering that this oscillatory signature may be a marker of differential WM resources during language processing.
Topic Areas: Syntax and Combinatorial Semantics, Control, Selection, and Executive Processes