thecenter

Activities and Seminars

James McQueen, Radboudt Institute, NL.
 
Date: Feb 16, 2012

What: The nature of phonological representation in spoken-word recognition: Evidence from speech learning

Where: BCBL, Paseo Mikeletegi 69, Floor 2.

Who: James McQueen, Behavioural Science Institute & Donders Institute for Brain, Cognition & Behaviour, Radboud University Nijmegen & Max Planck Institute for Psycholinguistics, Nijmegen, NL.

What: The nature of phonological representation in spoken-word recognition: Evidence from speech learning


Listeners must make use of stored knowledge about the sound structures in spoken language in order to be able to convert the acoustic speech signals that they hear into interpretations of those utterances. But what is the nature of this knowledge? How are segments and words represented? Answers to these questions are offered by experiments on speech learning. In this talk, three lines of research on learning about phonological representations will be discussed. In the first, evidence will be presented on how listeners use lexical knowledge to retune their representations of individual speech segments. These representations appear to be prelexical and to be phonologically abstract, but are allophonic rather than phonemic. Concerning the second research line, it will be argued that listeners can tune in to talker-specific reduction styles (i.e., which structures talkers tend to utter in an acoustically reduced manner in conversational speech). Evidence from eye-tracking experiments suggests that there needs to be prelexical representation of abstract phonological structures larger than individual segments. The final line of research concerns novel word learning. Data will be presented which indicates that listeners have abstract phonological knowledge not only about the segmental properties of spoken words, but also about their suprasegmental properties (e.g., their lexical stress patterns), and that listeners use this knowledge as they are recognizing newly-learnt words.