You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster A18, Wednesday, November 8, 10:30 – 11:45 am, Harborview and Loch Raven Ballrooms

Syntactic Constituent Rate Effects in EEG

Ellen Lau1, Mina Hirzel1, Natalia Lapinskaya2, Jeffrey Lidz1;1University of Maryland, College Park, 2McMaster University

Recent findings using MEG suggest that the phase reset of neural oscillations encodes hierarchical linguistic structure in the brain [1]. In two experiments, we demonstrate this entrainment manifests regardless of stimulus presentation rate. Our data suggest that further work with EEG could provide insights into the mechanisms that give rise to these effects (e.g. lexical vs. syntactic, [2]). In experiment 1, eight adults listened to lists of simple sentences with identical syntactic structure. One-syllable English words were synthesized independently using Neospeech. Each word was truncated or padded with silence at the end to a duration of 320ms. Sixty unique sentences were formed with the sentence structure [[Adj Noun][Verb Noun]] (e.g., ‘sick-Adj ducks-N hate-V pools-N’) by concatenating each syllable with no acoustic gaps added. Participants listened to 30 blocks (22 test blocks, 8 filler blocks) of 12 sentences each. Their task was to indicate at the end of each block whether they heard an ungrammatical sentence within that block, and eyes were kept closed to minimize blink artifacts. Following [1], we predicted that cortical tracking of abstract linguistic structure would manifest as high amplitude peaks at frequencies corresponding to sentence (1Hz), phrase (2Hz) and word (4Hz) presentation rates. For each participant, epochs were extracted spanning the length of each of the 22 12-sentence trials and averaged. As in [1], we observed peaks at the sentence (t(7) = 2.0, p < .05) and syllable rates (t(7) = 7.2, p < .01) in the frequency domain. However, the expected peak at the phrasal rate (t(7) = 1.7, p = .07) was not as prominent. The syllable rate activity had a central distribution, while the sentence rate activity had a frontal distribution. If this effect reflects the encoding of syntactic constituency, we would expect to observe it independent of presentation rate. In experiment 2, we slowed the presentation rate to 520ms per syllable by padding 200ms after each syllable; experiment 2 was otherwise identical to experiment 1. In the frequency domain, we observe clear frequency peaks at the syllable rate (t(7) = 6.0, p < .01) and phrasal rate (t(7) = 2.6, p < .05), but the sentence rate peak was somewhat diminished in strength (t(7) = 1.0, p = .16). The slower presentation rate makes it easier to examine correlates of the frequency effects in the time-locked evoked response. Here we observed after each word a positive deflection followed by a negative deflection peaking at about 500ms. The amplitude of this negative peak appeared larger after words 2 and 4 (the nouns), a potential correlate of the phrasal rate. In a potential correlate to the sentence rate, frontal responses were more positive between 150-450ms after word 1, which could reflect either early processes related to beginning a sentence, or late processes related to completing a sentence (670-970ms after word 4). References: [1] Ding, N., Melloni, L., Zhang, H., Tian, X., & Poeppel, D. (2016). Nature neuroscience. [2] Frank, S. & Yang, J. (2017). CUNY Sentence Processing Conference.

Topic Area: Grammar: Syntax

Back to Poster Schedule