STRUCTURE BUILDING OPERATIONS AND WORD ORDER

• Main Author: FLYNN, MICHAEL JAMES
• Language: ENG
• Published: ScholarWorks@UMass Amherst 1981
• Subjects: Linguistics
• Online Access: https://scholarworks.umass.edu/dissertations/AAI8117994

One of the goals of linguistic theory is to discover generalizations about the syntax and semantics of natural languages, and to construct theories of human cognitive capacity and development that explain these generalizations. Each statement about a language is theory-laden, that is, the characterizations of a possible generalization are determined by a theory (regardless of whether or not it is explicit) about the nature of human language. Further, statements about a particular language carry varying degrees of theoretical commitment. Compare the statements in (1) and (2): (1) The basic word order in English is such that the object noun phrase follows the verb, while in Hopi, the object noun phrase precedes the verb. (2) English has a rule to expand the VP node which has as a special case: VP (→) V NP; whereas the corresponding rule in Hopi has as a special case: VP (→) NP V. Though, in a sense, the method for checking both (1) and (2) is the same, (2) presupposes a claim that (1) does not: the incorporation of context-free phrase structures rules (PS rules) will lead to a revealing theory about how humans acquire the languages they do. However, if PS grammars are adopted as a component of the representation of the knowledge (or belief system) humans acquire in this domain, as was a natural assumption in the early days of generative grammar, we require some theory about these rules from which generalizations stated in terms of them follow. For example, alongside (1) and (2) consider (3) and (4): (3) English is a prepositional language; Hopi is postpositional. (4) English has (a subcase of) a rule: PP (→) P NP; for Hopi: PP (→) NP P. As Greenberg (1963) noticed, statements like (1) and (3) are not unrelated. In fact, with much greater frequency than chance, (5) and (6) hold: (5) If a language has VO order, it will be prepositional. (6) If a language has OV order, it will be postpositional. In Montague grammar, there has been little attempt to account for such generalizations. This dissertation proposes a theory of syntax which shares some of the features of Montague's work, yet attempts to give a principled explanation of low-level generalizations. The first two chapters are an introduction to the theory. Montague's methodology of stating a tight connection between the syntax and semantics is embraced and some of his technical apparatus is borrowed. Some ways in which the theory differs from common practice in Montague grammar are in (7). (7) Phrase structure rules are discarded entirely. Given category assignments to lexical items, hierarchical organization of phrases is defined universally. For languages with strict word order, left-to-right ordering of constituents of phrases is specified by a language-particular word order convention, which by its very nature is cross categorial. Chapters three and four extend and elaborate on the proposal in the first two chapters. A "wrap" convention is introduced for discontinuous constituents, as are category changing and relating rules which account for nominalizations and the double role played by adjectives in English. In the fifth and final chapter, word order conventions for other languages (Hopi among them) are adduced, and the first steps toward a theory of ordering conventions are taken. The goal of the dissertation is to offer support for the view that while (1) and (3) are generalizations about the languages in question, (2) and (4) are not. Thus the focus of interest shifts away from a theory which explains how children acquire a set of PS rules to one which must explain how they acquire word order conventions.