Learning acoustically complex word‐like units within a video‐game training paradigm.

Abstract

Over the course of language development, infants learn native speech categories and word boundaries from speech input. Although speech category learning and word segmentation learning occur in parallel, most investigations have focused on one, assuming somewhat mature develop of the other. To investigate the extent to which listeners can simultaneously solve the categorization and segmentation learning challenges, we created an artificial, non‐linguistic stimulus space that modeled the acoustic complexities of natural speech by recording a single talker’s multiple utterances of a set of sentences containing four keywords. There was acoustic variability across utterances, presenting a categorization challenge. The keywords were embedded in continuous speech, presenting a segmentation challenge. Sentences were spectrally rotated, rendering them wholly unintelligible, and presented within a video‐game training paradigm that does not rely upon explicit feedback and yet is effective in training non‐speech and ...