Statistical Learning from Acoustic Simulations of Cochlear Implants

Project Summary Here, in a laboratory simulation of language learning, we propose to investigate whether it is possible for normal-hearing infants and adults to segment words from spectrally degraded speech. The goal of this research is to test the hypothesis that fine frequency resolution is required for successful auditory statistical learning. Participants will first listen to an artificial language used widely in research on word segmentation, which consists of a pause-free, monotone stream of trisyllabic nonsense words (Lew-Williams & Saffran, 2012). In each experiment, speech will be (1) clear, (2) noise-vocoded, or (3) sinewave-vocoded. Then, participants will be tested on their ability to extract word boundaries in a headturn preference procedure (infants) or a forced-choice task (adults). The proposed research will be conducted in the CSD dept at NU. The principal investigators are Tina (director of the HLL) and Casey (director of the LLL). Project staff will include Hillary and an RA.