Modeling dialect variation and change

old_uid12151
titleModeling dialect variation and change
start_date2013/02/28
schedule10h
onlineno
location_infosalle 407
summaryDialects often respond differently to phonetic bias. For instance, in many languages of mainland Southeast Asia, a historical contrast between voicing of initial obstruents may come to be signaled primarily by aspiration in one dialect, voice quality in another, and pitch in a third. Is a predictive model of such divergences possible? In this talk, I address the nature of phonetic change from a computational perspective. I suggest an error-driven bias to drive the differential shift in cue distributions with reference to an ongoing sound change in two dialects of Khmer (Cambodian), and present a statistical learning framework that can be used to predict such qualitative shifts. I will also discuss how this framework might be applied to the study of large-scale dialect corpora such as the PFC and PAC.
responsiblesParisse