Competing Analyses of Complement Coercion: New Evidence from Behavioral and Electropsychophysiological Methods

The term “coercion ” has been used in linguistics to define a wide range of phenomena in which semantic mismatches are resolved through extra-syntactic processing (Jackendoff, 1997; Klein & Sag, 1985). The present study investigates one particular manifestation: complement coer-cion. In complement coercion, a coercion verb is combined with an entity-denoting complement, result-ing in a semantically underspecified meaning (Piñango, 2014). To explain the semantic processes underlying these types of sentences, the literature has traditionally proposed the Type-Shifting Hy-pothesis. This hypothesis is grounded in the observation that coercion verbs (such as begin and enjoy) and entity-denoting complements combine to result in an eventive interpretation of the com-plement (Jackendoff, 1997; Pustejovsky, 1991, 1995). Previous work (Baggio et al., 2010; Ku-perberg et al., 2010) presents evidence consistent with this hypothesis, demonstrating that coercion verbs elicited higher processing costs, attributed to type-shifting, in real-time comprehension. However, recent real-time processing investigations have challenged this traditional definition: converging evidence from Katsika et al. (2012) and Lai et al. (2014) suggest that a reexaminatio