[lingtalks] Philip Hofmeister 11/16 (Experimental Syntax Group)

[Klinton Bicknell]

The upcoming meeting of the Experimental Syntax Group will feature Philip Hofmeister, presenting a talk entitled "Wh-Dependency Processing & the Feature Depth Hypothesis." Philip is coming to us from Stanford. For more information, check his website at http://www.stanford.edu/~philiph/ .

For further information about events in the Linguistics department, including the schedule of future events, please visit http://ling.ucsd.edu/events/ .


:: Details ::

Thursday 16 November
4pm
AP&M 2452


:: Abstract ::

Philip Hofmeister, Stanford

Wh-Dependency Processing & the Feature Depth Hypothesis

The ease and success of retrieving information from memory depends upon an array of factors: the amount of time the information was studied, recency, context, what occurred between the initial exposure and the memory retrieval, etc. Failure to retrieve the right information in sentence processing (or even just a delay) can lead to processing difficulty, incomprehension, or perceptions of ungrammaticality.

This talk addresses how the amount of information and length in a linguistic expression affects the recall of that expression. On the basis of experimental data, I propose the Feature Depth Hypothesis: linguistic elements that encode more lexical-semantic features can help processing by facilitating memory retrieval of those elements later, along the lines of Anderson & Reder (1979). I also consider various cognitive explanations for the memory facilitation, i.e. reduced interference, distinctiveness, increased study time, and activation boosting.

Evidence from the self-paced reading studies supports the view that the amount of lexical-semantic features encoded in linguistic entities directly affects the processing times at retrieval points in a variety of complex wh-dependency constructions, e.g. multiple wh- questions (Hofmeister et al. 2007), weak island violations, CNPC violations, and nested dependencies. In fact, the use of semantically richer wh-phrases enables processing of so-called syntactic island violations to take place at a rate comparable to minimally different variants that do not violate any constraints on extraction. These results call into question the status of islands as universal boundaries on extraction, suggesting instead that they represent an aggregate of processing difficulties contributed by a variety of sources.