Events and verb classification*

The prevailing approach to understanding the syntactic mapping of arguments and argument alternations is to assume that verbs are organized into semantically distinct verb classes and that class membership determines the grammaticality of argument alternations. This paper presents syntactic evidence that verb-class membership cannot explain the causative and locative alternations, and that alternate mappings are based on the characteristics of the particular event the predicate denotes rather than on the verb's semantics. Once argument mapping is understood properly as implicating event structure, verb classes become superfluous, and one need not posit alternation operations that create a derived form from a more basic lexical item. I further argue against universal mapping principles based on semantic roles, such as Baker's (1988) UTAH, andf'or universal event-based mapping principles. Finally, I show that a theory of event-structure mapping has broad implications for acquisition. 1. The argument alignment problem Our knowledge of language specifies how to use verbs and their arguments in the syntax, including the relation among the verb, its arguments, and the syntactic position that arguments appear in. This paper addresses the form of this knowledge and the types of computations that support verb use. I will argue that characteristics of the event drive the lexicon-tosyntax mapping, and that the mapping principles use contextual information about the event and are therefore computational. The event-structure approach to mapping relations obviates the need to posit verb classes and verb alternations in linguistic representation. The prevailing approach to the lexical-syntactic relations posits a combination of semanticaUy defined universal napping principles and semanLinguistics 34 (1996), 191-223 0024-3949/96/0034-0191 ID Wrfter de Gruyter