Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Adapting to Foreign-Accented Speech: ERP Evidence of Interactions Between Phonetic and Semantic Processing
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
This poster is part of the Sandbox Series.
Jiaxuan Li1, Weijie Xu1, Judith Kroll1, Xin Xie1; 1University of California, Irvine
Language comprehension is guided by expectations; however, perceptual input in noisy linguistic environments may deviate strongly from one’s expectations, leading to a need to correct for perceptual errors or to reconstruct meaning. Two distinct mechanisms have been proposed to shape the processing of error-prone noisy linguistic signals. On the one hand, altered or degraded phonetic input may disrupt bottom-up mapping to categories and cascades to affect the retrieval of the intended lexical item, resulting in increased N400 (Romero-Rivas et al., 2015). On the other hand, prior top-down expectations may be favored over bottom-up information given a noisy signal. Comprehenders often tolerate or ignore perceptual errors and interpret the sentence as a more plausible alternative, resulting in reduced N400 (Hsin et al., 2023, Goslin et al., 2012). Both predictions have found empirical support in the literature, suggesting that listeners may engage in both strategies depending on various factors. We hypothesize that listeners can dynamically adjust bottom-up and top-down strategies, but over distinct timescales. The discrepant evidence may arise due to different studies capturing different stages of the adaptation processes. Specifically, listeners are predicted to adapt to the perceptual input rather rapidly, overcoming initial difficulty in phonetic processing. Over a slower timescale, listeners may gradually switch to a strategy that upweight top-down prior expectations to counteract the low-level perceptual noise; this strategy may persist even after the bottom-up driven difficulty to lexical access is resolved. To test this hypothesis, we track N1/P2 (phonetic encoding) and N400 (semantic processing) to examine how comprehenders adapt strategically over time to Foreign-accented speech (FAS)—compared to native speech (NS)—in an exposure-test paradigm. Method. In an Exposure phase, participants will hear only native or foreign-accented speech, depending on group assignment (NS vs. FAS; n = 24 each). Critical items consist of 128 high-constraint sentences predicting a specific word (e.g., “He played the trumpet in the school…”), completed by the expected word (“band”), a cohort competitor (“bank”), a rhyme competitor (“sand”), or an unrelated word (“mist”). In the Test phase, all participants hear 40 novel FAS sentences, 30 of which ending with an expected word and 10 with a phonological competitor. Comprehenders will answer yes/no comprehension questions following 1/3 of trials in both phases. Predictions. Baseline perception: NS group will exhibit large N1/P2 responses and graded N400 amplitudes (expected < competitors < unrelated). FAS group will show an initial attenuation or delay of N1/P2 responses compared to the NS group due to increased phonetic uncertainty. We predict that bottom-up perceptual errors may increase lexical access difficulty, which in turn may lead to increased N400 for expected words, yielding less differentiation between expected words and other conditions. Adaptation: FAS group will adapt to phonetic variation in FAS rapidly, as indexed by increased N1/P2 over exposure. At the same time, we predict that the FAS group will gradually adapt their strategy to rely more on top-down prior, resulting in reduced or more selective N400s (driven by upweighted top-down prior) over the course of exposure.
Topic Areas: Speech Perception, Meaning: Lexical Semantics