Visibility and argument identification : a conceptual semantic approach to Alamblak and Japanese

Abstract

This study attempts to combine the Government and Binding (GB) theory and Conceptual Semantics to provide an account for problems in the basic grammatical structures of Alamblak and some topics of Japanese. The thesis assumes Jackendovian Conceptual Semantics but aims to propose an alternative theory which establishes the relationship between syntax and semantics with maximum principles and minimum stipulations. The main concepts of the theory are argument identification and visibility. First, I introduce binary conceptual structure, whereby the hierarchical relationship among conceptual arguments is structurally defined and the distinction between inner arguments and outer arguments is represented. Second I make a distinction between identifier and identifiee by introducing the functional classification of syntactic categories. Identifiers are conceptually Functions, whereas identifiees are Basic Categories. PPs and APs are classified as Functions, i.e. identifiers. I also propose a theory of argument identification, which unifies the Conceptual Structures of an identifier and an identifiee under government and predication. It is proposed that the unification of two identifiers is carried out by argument sharing. This enables us to eliminate the specification of identifiers from the syntactic selectional information registered in the lexical specifications of verbs, e.g. locational verbs and motion verbs, for example, do not syntactically specify that they select a PP. In chapter one, it is suggested that what was formerly considered to represent syntactic selectional information (Predicte Argument Structure or Argument Structure) is radically reduced and that a (P)AS only represents the ability of a verb to identify an identifiee under government. In chapter three, the concept of syntacticisation patterns is introduced. Syntacticisation patterns derive the argumentidentifying abilities of verbs from their lexical conceptual specifications (LCS), where syntacticisation patterns are subject to parametric variations. The introduction of syntacticisation patterns completely eliminates the syntactic selectional specification in unmarked cases. The amount of the information specified in the lexical entries of verbs is minimised. Chapters one and three offer a concrete solution to the question of how syntactic structures are determined on the basis of semantic specifications of lexical items. Chapter four discusses one of the basic problems of Alamblak, agreement and possessor raising. Second or “object” agreement markers are described as incorporated pronouns. The complete complementarity of second agreement markers and overt objects is accounted for by the theory of unification. It also discusses the fact that the argument relations indicated by second agreement markers do not a have grammatical function, i.e. that they are conceptually interpreted without having a grammatical function. This fact is described only in Conceptual Semantic approaches to grammar. Possessor raising is characterised as a construction involving the modification of selectional information. The theory of argument identification gives a coherent account of the construction. Chapters five and six discuss “visibility”. The distinction of identifier and identifiee leads to a fundamental understanding of the concept of “visibility”. The basic concept of visibility proposed is that an identifiees must be visible for conceptual unification, whereas identifiers are not. Person-Number-Gender markers in Alamblak are visibility markers which are manifested only on identifiees but not on identifiers. Since they do not indicate the grammatical relations of NPs, the function of visibility marking is not to identify grammatical relations. The theory of unification proposed here provides a coherent account of the problems of visibility. Show more