Bob McMurray, PhD, University of IowaFebruary 20, 2017 - 12:00pm, Psych Bldg 219
Managing ambiguity: Changes in real-time lexical competition with development, individual differences and degraded input.
Every utterance that is at least partially ambiguous. The acoustic input is variable; word meanings are context dependent; and the fact that speech unfolds over time creates short periods where there is not sufficient information to parse the sentence, recognize a word, or even categorize a phoneme. Skilled listeners deploy complex processes that unfold over milliseconds to solve these problems and map incoming speech to meaning. These processes are now well understood in typical adults, but developmental work as well as work on disordered language often treats these real-time processes as a matter of performance, focusing on whether children have the right language "knowledge". Yet, the inherent ambiguity of language means that knowledge alone cannot be sufficient to explain skilled communication. This talk examines these issues in the context of lexical and phonological processing. I present research using the visual world paradigm to assess these issues in typically developing children, as well as children and adults who use cochlear implants and other populations with communication disorders. This work shows that even in typical children, when viewed from the lens of real-time processing, speech perception and spoken word recognition develop quite late-through adolescence. I also present research showing how Cochlear Implant users may tune these dynamics to better manage the ambiguity they face in the input. Finally, I examine several mechanisms that may support the development of real-time processing skills, and present laboratory learning studies documenting that inhibition may be plastic in short-term training experiences. Together, these studies suggest that lexical competition is not a simple by-product of an unfolding input, but rather is a process that is tuned with experience to better manage ambiguity in the input.