Psychobabble 01/15 – Bozena Pajak (UCSD/Rochester)

Bozena Pajak (UCSD/Rochester) will be presenting in Psychobabble this week.

Speaker: Bozena Pajak
Date/Time: Tuesday, January 15, 4-6pm
Location: Campbell 2122
Title: Modeling generalization in distributional learning of phonetic categories
Contact: Carson Schutze

Abstract:

Distributional learning guides the acquisition of phonetic categories, both native and non-native, but its underlying mechanisms are still relatively poorly understood. In this talk I will present results showing how adults make use of distributional cues in a novel language to make inferences about other properties of that language. Specifically, after learning a novel distinction between short and long consonants for one segmental class (e.g., sonorants), participants were also inclined to think that length distinguishes between consonants in another class (e.g., fricatives). I will present a non-parametric Bayesian model that accounts for this generalization pattern by assuming a hierarchy of abstract phonological representations, where evidence from one part of the system informs inferences about the rest of the system.

Advertisements

SynSem 01/11 – Peter Jenks (UCB)

We will have Pere Jenks from UCB this week in the first SynSem of Winter Quarter.

Speaker: Peter Jenks (UCB)
Date/Time: Friday, January 11, 2-4pm
Location: Campbell 2122
Title: Quantifier Float and Scope in Thai
Contact: Dominique Sportiche

Abstract:

In this paper I present an analysis of quantifier float (Q-float) in Thai as overt Quantifier Raising (QR). I show that the position of floated object quantifiers in Thai cannot be identified with A-traces of overt arguments, militating against an analysis in terms of stranding (e.g. Sportiche 1988, Miyagawa 1989). At the same time, Thai Q-float is subject to scope and locality restrictions of floated quantifiers, facts which I take to be problematic for a purely adverbial analysis. In light of these difficulties, I propose that Thai Q-float is QR, taken to be A-scrambling of a quantifier to a projection of VP (Johnson and Tomioka 1997). Q-float itself is argued to result from scattered deletion of the moved quantifier, where the pronunciation of the quantifier in the floated position is focus-driven, a claim which is supported by evidence from ellipsis and intervention effects.