Electrophysiological signatures of Working Memory during Sentential Language Processing
Poster D40 in Poster Session D with Social Hour, Friday, October 7, 5:30 - 7:15 pm EDT, Millennium Hall
Megan Nakamura1, Beverly Cotter2, Yanina Prystauka3, Eleonora Rossi1; 1University of Florida, 2University of California Davis, 3University of Tromsø-UiT The Arctic University of Norway
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 linguistic information in their second language, alpha and beta oscillations decline earlier in time. In this study we investigated the role of WM during language processing by manipulating WM itself to create a ‘taxing condition’ while processing in the native language. Method: Participants (n=34; f =28; mean age = 19.5) were native monolingual English speakers with no history of neurological diagnosis and/or learning disorders. All participants completed a behavioral Operational Span (O-Span) (Turner and Engle; 1989) and were randomly assigned to complete an EEG Grammaticality Judgment Task (GJT) either with an additional WM component i.e., “memory load” or without “no-memory load”. Participants in the no-memory group were instructed to complete the GJT consisting of (180) sentences: (60) correct–without violations, (60) with syntactic violations, and (60) with semantic violations. The memory group did the same but with the additional ongoing WM component which required participants to memorize a sequence of three words, to monitor the sentence for those words, and to randomly recall if a specific word was present in the sentence. All sentences were presented in rsvp. Preliminary Results: The data demonstrate a main effect of sentence condition; a larger P600 for sentences containing syntactic violations and a larger N400 for sentences containing semantic violations. Currently, the main effect of group (memory vs no-memory) has not reached statistical significance. However, the data demonstrates a potential qualitative difference between the two groups. For both groups, there is a biphasic negative to positive pattern only in the syntactic violation condition. This early negativity (between 300-500ms) with a left anterior distribution is indicative of ELAN/LAN and is larger for the non-memory group. The presence of ELAN/LAN for the syntactic condition is inline with previous findings reporting it as a marker of first-pass parsing (e.g. Neville et al., 1991; Münte et al., 1997; 1999; Friederici & Frisch, 2000; Friederici et al., 2002). It is thus possible that the manipulation of working memory is reflected in the shape and distribution of this biphasic signature. However, bearing in mind that this preliminary analysis has only considered grand averages, it is also possible that this apparent ELAN/LAN is more so indicative of averaging across individuals who elicit more N400-like ERPs during syntactic processing. For this reason, the final analysis will critically take into account individual variability in both ERPs and WM capacity as measured by the O-span. The time-frequency analysis is currently underway, but we predict to observe the same decline in alpha and beta band for the memory-loaded group as seen in Rossi & Prystauka’s, (2020) L2 processing condition, considering that this oscillatory signature may be a marker of differential WM resources during language processing.
Topic Areas: Syntax, Control, Selection, and Executive Processes