Perceptual learning in the laboratory versus real-world conversational interaction

Abstract

Understanding perceptual learning effects under novel acoustic circumstances, e.g., situations of hearing loss or cochlear implantation, constitutes a critical goal for research in the hearing sciences and for basic perceptual research surrounding spoken language use. These effects have primarily been studied in traditional laboratory settings using stationary subjects, pre-recorded materials, and a restricted set of potential subject responses. In the present series of experiments, we extended this paradigm to investigate perceptual learning in a situated, interactive, real-world context for spoken language use. Experiments 1 and 2 compared the learning achieved by normal-hearing subjects experiencing real-time cochlear implant acoustic simulation in either conversation or traditional feedback-based computer training. In experiment 1, we found that interactive conversational subjects achieved perceptual learning equal to that of laboratory-trained subjects for speech recognition in the quiet, but neither group generalized this learning to other domains. Experiment 2 replicated the learning findings for speech recognition in quiet and further demonstrated that subjects given active perceptual exposure were able to transfer their perceptual learning to a novel task, gaining significantly more benefit from the availability of semantic context in an isolated word recognition task than subjects who completed conventional laboratory-based training.