[lingtalks] Experimental syntax talk: Jon Sprouse 11/30 3:30

[Grant Goodall]

The final event this quarter for the Experimental Syntax Reading Group will
be a talk this Thursday by Jon Sprouse of the University of Maryland.
Everyone is welcome!  Note the 3:30 start time (by popular demand).


"Experimental Syntax: What does it get you?"

Jon Sprouse
University of Maryland

Thursday, Nov. 30
3:30
AP&M 2452


------------------
Abstract:
The question faced by the field of experimental syntax is: What does it
get you? The answer that everyone wants to give is: It gets you
[new/better/different] data. There are many ways, however, in which such
an answer is less than ideal. If the data contradicts previous findings,
or demonstrates the inadequacies of informal methods, than it becomes
fodder for those who argue that judgments are too unstable and insensitive
to form an empirical foundation for linguistics (e.g. Edelman and
Christiansen 2003). If the data corroborates previous findings,
demonstrating the judgments are robust and stable, then there will be no
justification for the added time and expense of formal experiments (e.g.
Phillips 2006). In this talk, I argue that experimental syntax can
actually satisfy both masters: the fine-grained control afforded by formal
experiments not only reveals the stability and sensitivity of judgments,
but also yields data that is beyond the reach of informal methods,
resulting in a research program for experimental syntax that is
complementary to standard theoretical syntax, rather than merely a
footnote to it.
	In the first case study, I present a series of experiments designed
to
determine the source of the syntactic satiation effects found by Snyder
2000, Hiramatsu 2000, and Goodall 2005. I hypothesize that there are
three basic models of the satiation effect, and use magnitude estimation,
yes/no tasks, and Likert scale tasks to test each model. The picture that
emerges is that judgments are much more stable than previously thought
when collected through formally controlled methods. The second case study
is a series of experiments designed to test for effects of temporarily
ungrammatical representations on the acceptability of final grammatical
representations. The results suggest that judgment tasks are incredibly
sensitive to syntactic information, so much so that they can detect the
temporarily illicit representation created by the filled-gap effect (a
low-level sentence processing effect). This overlap in the effect
inventories of syntactic theory and sentence processing suggests
experimental syntax might lead to data that has previously eluded
theoretical syntax: evidence for the correct division of properties
between grammar and parser.