Prosodic variation in intelligible L2-accented speech: Using acoustic measures to predict listening effort

Abstract

Despite the relative ease with which listeners understand spoken language in their daily lives, speech perception is a complex task that becomes challenging in certain conditions, such as listening to speech in an unfamiliar second language (L2) accent. In the current study, we use pupillometry data from McLaughlin and Van Engen (2020) to ask whether prosodic variation predicts listening effort for L2-accented speech. In that study, L1 speakers listened to L1- and L2-accented English sentences. The pupillary response was larger for the L2-accented speech, even though it was fully intelligible. We hypothesize that prosodic differences in L2- versus L1- accented speech may account for such differences in listening effort for fully intelligible speech. To test this, we apply four acoustic measures to each of the stimuli used in the experiment. The first two measures, wiggliness and spaciousness (Wehrle et al., 2018), quantify the number and degree of F0 contour changes across each trial. The remaining two measures are relative word duration (Baker et al., 2009) and vowel pitch (Li & Shuai, 2011). We will use these trial level measures as fixed effects in Growth Curve Models to test the extent to which prosodic deviation predicts pupillary responses.