The Structure Of The Causative Resultative Construction In Mandarin Chinese

The causative resultative construction formed by a causer, a resultative verb compounds ?RVC?, and a causee is widely used as a way to express causation. Influential studies on this topic include Huang's ?1992,2006? control analysis, Sybesma's ?1999? and Peng's ?2011? small clause analysis, Y. Li's ?1990,1995? lexical-semantic account and Tang T.-C.'s ?1991? incorporation analysis, which can be distinguished based on their approach to the analysis of causation. Researchers that adopt the syntactic approach ?Sybesma 1999; Huang 2006, etc.? maintain that causation is expressed in syntax by combining the unaccusative-like RVCs with a syntactic light verb CAUS, which forms the familiar causative alternation ?Dowty 1979; Levin 1993, etc?; in other words, the syntactic approach considers the properties of the RVCs irrelevant to the causative meaning. For researchers who adopt the alternative lexical approach ?Y. Li 1990,1995; C. Li 2008,2013?, causation comes from the RVCs, which may be introduced in syntax via a causative thematic hierarchy. After a comprehensive review of the previous literature, I propose that the properties of RVCs are closely related to the syntactic structure of the causative construction, and that different types of RVCs exhibit distinct causative properties. Based on this assumption, this thesis revisits the syntactic structure of the causative construction.By examining the event types of V₁, V₂ and the properties of RVCs, I observe that the event types of V₁ determine the causative properties of the construction, namely, whether causation comes lexically or syntactically:??? if V₁ is an activity verb, the RVCs are unergative-like, encoding a causal relation between the causing event denoted by V₁ and the resultative state denoted by V₂; ??? if V₁ is a stative verb, RVCs are unaccusative-like, encoding a change-of-state relation between V₁ and V₂ and can be causativized by CAUS; ??? if the activity V₁ is unexpressed, RVCs behave like unaccusatives ?ergativized RVCs?, with no causal relation encoded. It is also found that the lexical causation encoded by unergative RVCs is different from the syntactic causation expressed by CAUS, the former expressing a bi-eventive relation ?Dowty 1979?, while the latter relating an individual to a resultative state ?Rappaport Hovav & Levin 1998?.Therefore, it is generalized that the structure of the lexical verb Vi affects the structure of the causative construction. Within the framework of the light verb theory ?Hale & Keyser 1993; Chomsky,1995; Lin 2001?, I propose that the event-type-denoting light verbs in L-syntax, such as DO, BECOME, combine with the verb root V₁ to express the causing event or the resultative state; in line with the theory of lexical structure proposed by Huang, Li & Li ?2009?, V₁ may also choose not to combine with any light verbs, i.e. the unexpressed V₁. The three types of Vi then incorporate with a stative V₂ and derive unergative RVCs, unaccusative RVCs and ergativized RVCs, respectively. Based on the semantic relations between Vi and V₂ in unergative and unaccusative RVCs, I propose a functional projection FP to express the causal or change-of-state relation. The null head is phonetically realized by verb movement or de-insertion. The proposal of FP can account for the argument sharing in theta-role assignment.The proposed analysis offers an account for the ambiguities found in some causative sentences and can be extended to the structure of the V-de construction. Furthermore, it explores the nature of causation in the causative construction and supports the view that lexical verbs are involved in the syntactic computation by examining the event types of RVCs and other syntactically relevant lexical information.