Argument Realization And Event Structure: The Verb Phrase Syntax

The study of verb phrases is at the core of the study of the syntax-semantics interface. The present dissertation proposes a theory of linguistic representation that captures the syntactic structure of verbs and their arguments realization. To be more specific, it constructs a theory of verbal argument structure which is able to explain how arguments are syntactically and semantically realized and the process by which verbal meanings are compositionally constructed from some basic (conceptual) primitives.Our framework is based on the assumption that the proper representation of argument structure is event structure. We assume that argument structure, event structure and syntactic structure are isomorphic. The goal of our theory is to develop a transparent interface between syntax and semantics, in which the verb meaning and their arguments can be directly read off their structures. Within our theory, verb meanings are compositionally derived in the syntax from some basic primitives: verbalizing heads, functional elements or event introducers that license eventive interpretations, and verbal roots, abstract concepts drawn from encyclopedic knowledge. We hold languages share the same inventory of primitive building blocks and are governed by the same set of syntactic constraints—all endowed by principles of Universal Grammar and subjected to parametric variations. Our framework draws the insight from the Distributed Morphology, the Voice hypothesis and theories concerning lexical semantic representations and is supported by empirical data from both Mandarin Chinese and English.We propose three types of ontological event types: vDO, vGO, vBE and two types of ontological roots: activity and state. These basic primitives or building blocks will be employed to compositionally build up complex event structures. Our proposal is strongly supported by Mandarin verbal system, in which there are only monomorphemic state and activity verbs, and achievements and accomplishments are derived from state and activity. We give a detailed study of Mandarin resultative verb compounds (RVCs), since they are good instances of complex events and serve as an ideal touchstone for the validity of our theory. We have shown that the complexity of argument realization patterns of RVCs can be nicely captured in our framework and the ambiguity can be well accounted for. The patterns of argument sharing in these verbal compounds can be analyzed as control, thus grounding argument structure in well-known syntactic constraints such as the Minimum Distance Principle.We also have an intensive study of extra-argumentality in Mandarin Chinese, with the focus on the non-canonical double object constructions (DOC) and transitive unaccusative constructions. We hold that extra arguments in these constructions are licensed by a high applicative head, which defines the relation between an individual and an event. To be more specific, the extra arguments in non-canonical DOCs and transitive unaccusatives are the arguments affected by an activity event and change of state event respectively.Finally, since we assume the compositionality in event structure construction, that is, complex events are composed of simplex events, and all languages share the same set of primitive building blocks and are governed by the same syntactic constraints, we have to account for the great difference in their surface form when representing events or difference in the verb behavior. We argue that different languages employ different (syntactic) devices to compose a certain event and to distinguish between structures. For example, in composing a causative change of state event, Mandarin may involve either one or two roots, while English involves only one. In representing an inchoativity, Chinese may be formed by stative roots followed by aspect marker le, whereas English may overtly realize its vGO by suffixation. We show that cross-linguistic differences in the realization of verbal arguments can be reduced to variations in the way functional elements interact with verbal roots.Overall, our theory is a representation theory concerning how arguments and events are syntactically represented, but also a transparent syntax-semantics interface theory. We think a theory of argument structure grounded in independently-motivated syntactic constraints, on the one hand, and the semantic structure of events, on the other hand, is able to account for a wide range of empirical facts with few stipulations, such as linking rules.