[lingtalks] Talk by Daniel Grodner today

[Catherine Alioto]

UCSD Department of Linguistics presents

A talk by Daniel Grodner, University of Rochester


Monday January 29, 2007
2:00-4:00pm
AP & M 4301


Title: Discourse Contrast and Interactivity in Language Comprehension

Abstract: Multiple types of information must be rapidly coordinated 
to support real time language comprehension. In order to explain the 
efficiency of this process, investigators often posit limits on the 
types and quantities of information to which processing subsystems 
have access (e.g., Fodor, 1983). I will present two lines of inquiry 
addressing processing limits across levels of representation. 
Specifically, this work explores a dependency between linguistic 
labels and the referential environment. If two objects of the same 
type are copresent in the discourse (e.g., two apples) and a speaker 
wishes to single one out, a modified label is required (e.g., "The 
apple with the stem"). This dependency is bidirectional; upon 
encountering certain modified labels, perceivers infer the existence 
of a contextual contrast --another entity of the same type that 
differs along the dimension indicated by the modifier (e.g., an apple 
without a stem).
The first line compares restrictive and non-restrictive modifiers. 
Restrictive modifiers induce the expectation for a contrast whereas 
non-restrictive modifiers do not. I demonstrate that restrictive 
modifiers are more difficult to process in a null context, but 
non-restrictive modifiers are more difficult to process in a 
supportive context. The former finding indicates that adding a 
contrastive referent to the discourse model is costly. The latter 
finding indicates that discourse-based referential information can 
guide syntactic processing, contra the modularity hypothesis (Fodor, 1983).
The second line examines the mechanism that gives rise to contrastive 
inferences. I will argue that the projection of contrast derives from 
a rapid inference based on Gricean communicative conventions, and 
present eye-tracking evidence that perceivers consider the 
reliability of the speaker when fixing reference for a modified 
label. This has implications for whether perceivers engage in Gricean 
reasoning on-line (cf. Sperber & Wilson, 1985) or compute contrast as 
a reflex of more low-level information (cf. Levinson, 2000).



Catherine Alioto
Assistant to the Chair
Academic Affairs Coordinator
University of California, San Diego
Department of Linguistics
McGill Hall
9500 Gilman Drive
La Jolla, California 92093-0108
(858) 534-3601