The meaning of an exception

Dustin Chacón

University of Minnesota

Monday, February 18, 2019
10:30 a.m.–Noon

Humanities Center Conference Room D, Rush Rhees Library

Abstract

Syntactic theory and psycholinguistics both share the goal of describing how grammatical form is represented in the mind. However, these disciplines have progressed independently. In this talk, I show that sentence processing data can be used to productively constrain syntactic theory, when seen through the lens of a clear linking hypothesis. I examine filler-gap (movement/A') dependencies, which have a well-described profile in sentence processing, and appear to be guided by syntactic principles (Stowe 1986; Phillips 2006; Yoshida, Kazanina, Pablos, Sturt 2014). In (3), the filler dependency appears to resolve with a "resumptive" pronoun inside of a syntactic island, which is surprising since they typically resolve with a gap. I argue that resumption is perceived to be licensed when typical filler-gap dependency processing mechanisms falter, which then enables comprehenders to violate syntactic constraints to build a coherent interpretation (Chacón 2015, 2018, under review). Second, I examine filler-gap dependencies that cross into adjuncts, (4). Adjuncts are typically thought to be islands, (2; Huang 1982; Chomsky 1986; Uriagereka 1999), but extraction from some non-finite adjuncts are perceived to better than others (Truswell 2007; 2011). In a series of studies, I show that the processing profile for these configurations do not show the same profile as well-formed filler-gap dependencies (Kohrt, Sorensen, Chacón 2018; Kohrt, Sorensen, O'Neill, Chacón 2019; Kohrt & Chacón 2019). I argue that these sentences are not syntactically well-formed, but should instead be analyzed as 'repair' of an ungrammatical sentence motivated by interpretation, similar to resumption. On my account, therefore, we do not need to complicate the theory of islands to permit resumption or adjunct extraction. Moreover, these results suggest that semantic information may intervene to rescue a syntactically ill-formed sentence in sentence processing (e.g., Kim & Osterhout 2005), and in the informal acceptability judgments that syntacticians use for data collection.

(1) This is the apple that Ernie ate ___ .

(2) *This is the apple that Chris ran [Adj while Ernie ate ___ . ]

(3) ?This is the cat that Chris said [NP its owner] is a linguist.

(4) ?The linguist bought the apple that the cat laid around [Adj eating __ ]