The semantics of roots determines argument structure

Abstract

A challenge for theories of argument structure that take verb meanings to decompose into so-called roots and event templates (Rappaport Hovav and Levin, 1998; Ramchand, 2008; Alexiadou et al., 2015, i.a.) relates to successfully accounting for the distribution of roots, as not all roots seem to appear in the same syntactic contexts. In this respect, an influential approach holds that roots are indifferent to syntactic distribution, and therefore that in principle any root can appear in any context (Borer, 2005, 2013; Mateu and Acedo-Matell´an, 2012; Acedo-Matell´an and Mateu, 2014). Another influential approach on the other hand classifies roots into semantic classes constraining the syntactic contexts roots can appear in (Marantz, 1997; Harley and Noyer, 2000; Reinhart, 2002; Alexiadou et al., 2006; Ramchand, 2008). Here, I show that the two main approaches to argument structure either undergenerate (Rappaport Hovav and Levin, 1998; Alexiadou et al., 2015) or overgenerate (Borer, 2005; Acedo-Matell´an and Mateu, 2014). In particular, I provide empirical data that show that so-called result verbs enjoy a certain degree of elasticity, contra Rappaport Hovav and Levin (1998); Alexiadou et al.(2015), yet there are cases of lack of verbal elasticity, contra Borer (2005); Acedo-Matell´an and Mateu (2014), i.a. To this end, I propose a root-sensitive approach to event structure in which the semantics of distinct classes of roots can determine the syntactic contexts roots can appear in. Under the present approach, cases of ungrammaticality are argued to result from clashes between the semantics of roots and the semantics of the event structure.