Abstraction and adaptation: Key mechanisms listeners use to cope with speech variability

Listeners have to cope with a highly variable input in order to be able to recognize speech sounds and spoken words. Two theoretically opposite accounts have been proposed: On the abstractionist view listeners make phonological abstractions about speech and discard variable information as noise; on the episodic view listeners store fully-detailed episodes of spoken language in long-term memory. In this talk I will present evidence from experiments on speech learning which lead toward a synthesis of these two positions. Episodic traces of speech are initially stored, but with time, phonological abstractions are formed, and it is these abstract representations that then mediate language understanding. The experiments (using a variety of behavioural and neuroscientific techniques) explore perceptual learning about speech sounds, how new words are learned, and how flexible lexical processing is. These experiments thus provide evidence that listeners use not only abstraction but also adaptation – the ability to adjust to different listening contexts – to cope with speech variability.