More than adaptation--evidence for training-induced perceptual learning of time-compressed speech

The identification of time-compressed speech improves significantly following short-term exposure, but it is not clear whether additional practice yields additional learning. The goal of the experiment reported here was to determine whether 30-40 minutes of training, during which listeners practiced the identification of 100 different time-compressed sentences, yielded additional learning to that induced by a single brief exposure to 20 sentences. We also asked if this learning generalized to novel sentences and to a new speaker. Training resulted in more learning than a single brief exposure, and this learning generalized to a new speaker but not to new tokens. Brief exposure to 20 sentences did not result in any significant increases to performance when compared to naive listeners. We conclude that a prolonged learning phase exists for time-compressed speech, but that learning during this phase does not fully transfer to new, untrained tokens.