Psychophysiological indices of listening effort due to noise masking and nonnative-accent

Abstract

Physiological responses linked to listening effort may differ depending on whether the increased effort is due to masking vs. distortion of the speech signal (Francis, et al. 2016; Zekveld, et al. 2014), suggesting that listeners may engage different cognitive strategies to compensate for different sources of reduced intelligibility. Here, we assessed physiological and behavioral indices of effort in participants listening to short stories in two conditions: (1) nonnative-accented speech in quiet and (2) native-accented speech masked by speech-shaped noise. Compared to previous studies, stories were used instead of sentences to investigate effort over longer durations, while accented speech was used to reduce intelligibility more naturally than noise vocoding or text-to-speech synthesis. Masker noise levels were adjusted for each participant to ensure comparable word recognition performance across conditions. Individual characteristics including age, hearing thresholds, vocabulary, selective attention, working memory capacity, personality traits, and noise sensitivity were also measured. Smoothing spline ANOVAs and growth curve analysis will be used to determine overall effects of condition, as well as which traits best predict patterns of effort-related responses over time. Different sources of listening effort are predicted to induce different patterns of physiological response reflecting engagement of different cognitive processes, possibly in a trait-dependent manner.