On obligatory obligation: the composition of Italian causatives

Abstract

Since Kayne (1975), it has been recognized that Romance causatives with faire fall into two classes, the faire par (FP) and faire infinitif (FI). In the FP construction, the Causee argument is optionally expressed in an adjunct by -phrase, while in the FI construction, as argued by Guasti (1996), the dative Causee is not optional, and occurs in an argument position. We argue from Italian data that the observed syntactic and semantic differences between the two constructions do not follow simply from the argument/adjunct distinction, but rather reflect a selectional difference: FI constructions embed a particular vP, while FP constructions do not; they embed a nominalized VP. We claim that the semantic characteristics of the complement vP in the FI construction explain an 'obligation' requirement on the FI Causee, which is absent in the FP construction. Consequently, we can eliminate the need to refer to an independent level of semantic representation (theta-grids, as in Guasti) to account for the different interpretations of the two constructions. Further, we show that the theta-grid approach is misleading as it leads Guasti to conflate several distinct structures under the FI label, including passives of fare. We argue that passives of fare causatives are all FP constructions. When a dative argument is present, it is a Benefactee, not the dative Causee of the FI.