The Role of Entrained Oscillations in Segmenting Rhythmic Sentences during Foreign Language Listening

Abstract

When hearing a foreign language, listeners often have difficulties segmenting the continuous speech signal into individual words The mechanisms by which word segmentation occurs are largely unknown. Recently, neural oscillatory entrainment to the speech envelope has been proposed as a possible mechanism underlying speech processing. In the current study, we investigated whether the rhythm structure of foreign speech provides cues for word segmentation and whether neural oscillatory entrainment is involved during parsing of the signal. English natives listened to highly rhythmic and highly nonrhythmic Dutch stories. As an index of neuronal entrainment, we computed coherence between the neural activity measured with magnetoencephalography (MEG) and the speech envelope. In order to quantify word segmentation ability, participants performed a forced-choice lexical decision task after each story. They were prompted to recognize a word that had been repeated within each story and distinguish it from a novel word. Further, as a concurrent measure of successful word segmentation, we analyzed event-related fields (ERFs) in response to familiarized versus novel words. High response accuracies as well as a suppression of the ERFs in response to familiarized versus novel words indicate that participants were able to segment words in the foreign speech stream. Moreover, we observed a significant increase in speech-brain coherence for rhythmic versus non-rhythmic speech input. Contrary to our predictions, we found no effect of rhythmicity on word segmentation. Thus, our results suggest that word segmentation of foreign speech does not solely rely on rhythmic regularities in the signal. Further research is required to elucidate the underlying mechanisms of processing both rhythmic and non-rhythmic speech and to integrate the role of neuronal entrainment. Keywords: Neural entrainment, Oscillations, Foreign language listening, Word segmentation