"How Social Categories and Speaker Attributes affect Phonological Development and Phonological Processing"

Spoken utterances--indeed, even single spoken words--convey multiple types of information simultaneously.  A single token of the word /kæt/ conveys first of all the asserted meaning "member of the class of objects 'felis catus'."  Subtle details of the pronunciation index additional information, such as the talker's pragmatic intent (i.e., a falling pitch glissando on the nucleus of the /æ/ would indicate that the talker is presenting this as new information, and that subsequent turns in the conversation need not reference this assertion).  Finally, speech-sound variation encodes social attributes about the talker.  For example, Smith, Munson, and Hall (2008) showed that /æ/ variation, like the kind one might get in /kæt/, indexes attributes like the talker's regional origin and age.

This talk focuses on how these types of knowledge of the sound structure of language are learned by children, and are processed by adults.  Recent research cutting across a variety of academic disciplines has shown convincingly that the phonological categories needed to express lexical contrast are imputed from the input.  The first part of the talk will review studies in our laboratory that have shown how children's sophisticated statistical learning mechanisms, prodigious perception abilities, articulatory exploration, and growing lexicon conspire to lead the child to the sound system of the languages that they are acquiring.  The second part of the talk will examine studies in our laboratory that have shown that children during this same period learn how to convey different attributes, specifically those associated with gender.  The third part of the talk will present preliminary results from a series of studies examining the extent to which these types of learning are independent from one another.