Cue-based Probability Estimation for Phonetic Reduction
General idea - We take a two-step modeling approach (cf. Jaeger, 2006-thesis, Levy & Jaeger, 2007) to investigate (a) what types of cues do speakers employ when estimating the information density of the word they are producing (which seems to influence its production (Aylett & Turk, 2004; Bell et al., 2003, in press), and (b) to what extent are cues weighted optimally -- i.e. how much do they effect determine phonetic reduction dependent on how much information they contain about the probability of the word?
For more detail continue to the lab log.