An Event Structure Analysis Of The Resultative/Lexical Causative And The Double Object Constructions

The idea that event structure plays an important role in the argument structure of a sentence has gained great attraction in recent years. Baker (1997) states that one central task for any theory of the semantics-syntax interface is to solve what is called the "linking problem or mapping problem": the problem of discovering how the elements of an event, mainly the participants and their relationships, are expressed in surface grammatical forms. To address the linking problem, a considerable number of theories of argument linking from a wide range of theoretical perspectives have been proposed in recent years, including Hale and Keyser (1991, 2002); Jackendoff (1990); Pinker (1989); Rappaport and Levin (1988, 1995, 1998, 2001); Van Valin and Lapolla (1997); Dowty (1991); Wang (2003). Most current syntactic theories (e.g., Government-Binding theory, Lexical Functional Grammar, Head-driven Phrase Structure Grammar) posit a verb's argument structure as a subset of the verb's lexical entry. The licensing of an argument is based on its individual relationship to a certain verb in the syntax---a feature that has been referred to as 'predicate-centeredness' (Geuder and Butt 1998). Argument structure, it follows, is determined by properties of lexical items. One problem encountered by the predicate-centered approach involes the resultative construction, in which the complex predicate can be said to have two predicates. An intransitive verb, by definition, does not have an object, yet when appearing in the resultative construction, it must have an object. This fact leads Simpson (1983) to propose the Direct Object Restriction (DOR) to account for the resultative construction. However, we present evidence that the DOR is not a valid generalization. To characterize the resultative construction, it is necessary to adopt an event structure approach. In going beyond the 'predicate-centered' view, event semantics allows a far more flexible approach to argument projection. In their various forms, the approaches proposed by Krifka (1992), Tenny (1994), Verkuyl (1993), Hale and Keyser (1993, 2002), Van Valin and Lapolla (1997), Rappaport and Levin (1998, 2001) and Croft (1991, 1998) explore the interaction between aspectual properties and syntactic argument realization. However, little research has been done in questions related with the constitution and composition of the event structure: How many basic event structures underlie the resultative/causative constructions? What exactly are the basic event structures made up of? What kinds of relationships are involved in an event structure? How should the event structure be characterized and represented? These questions form the concerns of this study. Since sentences express events and situations that happen in the world or a imagined world, representing the relationships between the participants that are involved in those events and situations is the major task facing the event approach. The participants and their relationships are what we call the event structure. The main purpose of this study is to investigate how the event structure of the resultative and the lexical causative constructions and the double object construction is realized as the syntactic structure. We identify four basic event types. They are the action event, the motion event, the change of state event and the relation event. The combination of the action event and one of the other three events form a complex event structure that underlies the resultative, the lexical causative and the double object constructions. Based on Van Valin and Lapolla (1997), we develop a system for representing the four events and the combination of the events. The representation of events takes the form of [ACT (Agent, Patient)], [MOVE (Theme, Goal)], [BECOME (Theme, state)], [HAVE (Possessor, Theme)]. That is, the event structure is composed of an atomic predicate like ACT, MOVE and thematic roles like the agent, the patient. The representation of the complex event takes the form of [ACT (Agent, Patient)] CAUSE [BECOME (Theme, state)] or [ACT (Agent,Patient)] CAUSE [MOVE (Theme, Goal)] etc. We also provide means of structurally representing these event structures, based on Hale and Keyser (2002) and Wang (2003). The event structure approach outlined above provides a unified and satisfactory account for many sentence constructions, especially the resultative and the lexical causative constructions and the double object constructions. The possible conflation patterns of the predicates and the goal or state characterize the syntactic structures of the sentence expressing the event structure. The explanatory power of the present event structure approach manifest itself in its application to the analyses of the resultative, the lexical causative construction and the double object constructions in English and Chinese. In addition, the present approach predicts that the Direct Object Restriction or DOR, first proposed by Simpson (1983), is not a valid generalization about the resultative constructions, thus, in contrast to Wang's (2003) claim that the DOR is a cross-linguistic condition on resultatives in Chinese as well as in English. We present evidence from Chinese and from English that our predictions are indeed valid. Based on the event structure approach, we also illustrate that the much-discussed double object construction and the to-dative construction are in fact special types of the resultative constructions, conforming to the same principles that govern other resultative constructions. Finally, we give a detailed discussion of the special Ba construction in Chinese. We propose a Ba constraint as a condition for the Ba construction, which states that only the theme of a change of state event or a motion event that is in the object position, e.g. combined with an action event, can be introduced by ba to the preverbal position.