Listener sensitivity to structured phonetic variation

Abstract

Memory for spoken language is not a veridical representation of experience. Instead, memory reflects integration across our interlocutors’ messages, resulting in robust memory for meaning with relatively poor memory for the specific form of the message. This is striking considering that in the process of mapping from speech to meaning, listeners show exquisite sensitivity to the acoustic-phonetic structure of speech. In this talk, I will review selected findings from work in our laboratory that examines listeners' ability to dynamically adapt to structured phonetic variation, focusing on variation associated with talkers' idiolects. These studies examine mechanisms that allow listeners to exploit structured variation for speech perception, voice recognition, and memory of spoken language. Collectively, the results (1) identify principles that govern how listeners modify the mapping to speech sounds to reflect cumulative experience with talkers’ phonetic input, (2) show that sensitivity to structured phonetic input facilitates identification of a talker’s voice in addition to the linguistic message, and (3) demonstrate that talker identity provides a critical structure for the integration of spoken language in memory. These findings help explicate a theoretical framework that accounts for tension in a linguistic architecture that uses both abstract and instance-specific representational knowledge to guide spoken language processing.