Presented at the 26th International Congress of Applied Psychology. Athens, July 16–21

Languages differ, among other things, in the speech sounds used to make meaningful distinctions. Infants learn to distinguish and categorize the sounds of speech in their linguistic environment, and subsequently show diminished sensitivity to other sounds. Studies of the structure and nature of speech representations have documented language-specific differences in the perception of speech sounds and in the ability to learn nonnative sounds. New models of speech learning underscore the lifelong capacity to learn as well as the diminished efficiency with which such learning can take place, due to the fine-tuned perceptual and attentional constraints resulting from native language use. It has been established that variability of the speech training stimuli is necessary for achieving generalization to new stimuli and for in-context use of the learned sounds. More recently, adaptive procedures and exaggeration of speech distinctions have been applied to counter the effects of continuous relearning of the native speech sounds during normal speech perception. Thus, future language training can be fruitfully targeted to specific difficulties in speech perception by controlling stimulus discriminability and variability to achieve efficient learning of speech sounds and thus to improve vocabulary development, comprehension processes, and ultimately better communication in the nonnative language.