Cognitive effort in comprehending non-native speech decreases with listening experience

Abstract

Listeners adapt to non-native accents, as shown by improved accuracy transcribing a speaker's intended utterance during testing (Baese-Berk et al., 2013; Bradlow & Bent, 2008; Sidaras et al., 2009). However, this improvement does not address whether cognitive effort decreases with accent experience, or whether accented speech continues to require effort to compensate for atypical features. The present study addresses this question using pupil dilation, a physiological measure indexing cognitive effort (Sirois & Brisson, 2014). In a matched-guise experiment, 48 participants heard 90 sentences spoken in either a General American Accent (GAA) or a non-native accent (NNA) while an eye tracker measured pupil diameter. An equal number of sentences with words predictable from context (e.g., “Sugar tastes very sweet.”) and without (e.g., “It can’t be brown.”) were included. Participants transcribed (typed) each sentence. Pupil dilation while participants listened to each sentence was measured as the difference between average diameter for 100 ms immediately before and after the sentence played. Dilation decreased significantly more in the NNA than in the GAA condition, indicating that perception of accented speech becomes less effortful with experience. No effects of semantic predictability were observed. The relationship between effort and transcription accuracy will be discussed.