Examining Talker and Phoneme Generalization of Dimension-Based Statistical Learning in Speech Perception

Speech perception flexibly adapts to short-term regularities of the ambient speech input. Recent research demonstrates that the function of an acoustic dimension for speech categorization at a given time is relative to its relationship to the evolving distribution of dimensional regularity across time, and not simply to its fixed value along the dimension. Two studies examine the nature of this dimension-based statistical learning in online word recognition, testing generalization of learning across talkers and across phonetic categories. The results indicate that dimension-based statistical learning is specific to the experienced regularities, resisting transfer across talkers or phonetic categories.