Search Abstracts | Symposia | Slide Sessions | Poster Sessions
From Awareness to Articulation: Metacognition and the Neural Correlates of Learning to Produce Foreign Speech
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Aarni J. Seppälä1, Henna Tamminen1, Henry Railo1; 1University of Turku
Learning to produce new speech sounds is central to foreign language (L2) speech acquisition. Although learners vary in how well they acquire L2 sounds, the reasons for these individual differences remain mostly unknown. One possible source of individual variation is metacognition, the ability to monitor and adjust one’s own behavior. In the context of speech production, metacognition can be thought as the correspondence between speaker’s subjective confidence in their pronunciation and its actual acoustic accuracy. Despite its potential importance, the link between metacognition and novel speech sound learning has been neglected in research. This question is also intriguing from a neuroscientific perspective: while the automatic and unconscious tracking of self-produced speech via corollary discharges is widely recognized, it remains unknown how conscious perception of one’s own pronunciation takes place on neural level. The aim of our study is to take a cognitive neuroscientific perspective to understand how 1) metacognition mediates learning of L2 sounds, and 2) what event-related potential (ERP) components mediate metacognitive evaluation of one’s own speech. In the present study, Finnish-speaking adults (planned N = 80; data collection ongoing; current N = 13) produce native Finnish and non-native Estonian vowel sounds while 64-channel electroencephalogram (EEG) is recorded. After each production, participants report how confident they were in their pronunciation. Pronunciation quality is quantified through formant analysis (F1, F2) and ratings by native Estonian listeners. Analyses on EEG-data will examine ERP components after speech onset, with particular attention to the P2 (120–200 milliseconds post speech onset) measured over left frontal electrodes. Preliminary results on the behavioral data were conducted using linear mixed-effect regression. The formants of the produced foreign speech sound approached the formants of native target sound (β = -0.008, SE = 0.001, 95% CI [-0.011, -0.006], t = -6.82) indicating better pronunciation as the experiment progressed. In accordance, participants’ confidence improved across experimental blocks (β = 0.164, SE = 0.009, 95% CI [0.147, 0.182], t = 18.83). Although participants improved their pronunciation during the experiment, and their confidence improved, there was a dissociation between confidence ratings and speech production accuracy. Confidence did not predict accurate speech production (β = -0.003, SE = 0.003, 95% CI [0.147, 0.182], t = -1.10). The P2 component was positively correlated over left frontal electrodes with confidence, although it failed to reach significance with the current sample size (r(11) = .49, p = .087). Although participants improved their pronunciation during the experiment, and their confidence improved, there was a dissociation between confidence ratings and speech production accuracy. The further planned EEG analyses with P2 and other components will examine the neural signature of confidence in one’s own speech.
Topic Areas: Language Production, Speech Motor Control