Transformations in Linguistic Structure

Abstract

Empirically, a language is a set of discourses — things said or written. Each discourse can be shown to be a sequence of (one or more) sentences (or certain fragments of sentences), a sentence being the largest stretch of language whose composition can be described in certain compact ways. (It is possible to state additional properties of a discourse, but not — at least at present — direct rules of how it is composed.) There are several ways of analyzing the structure of sentences, and the applicability of one does not falsify the others. The most common method, both in traditional grammar and in modern linguistics, is to describe sentences as composed of certain constituents, e. g., subject and predicate, and these in turn of certain smaller constituents (say, subject as composed of noun and its modifiers; predicate as composed of verb plus object), and so on until we arrive at morphemes (morphologically indivisible words, stems, affixes). This can be stated in a compact hierarchy of rules or mappings, the rules and their hierarchy all showing some regular character. All sentences, or all of a distinguished subset of sentences, are composed in this way. The various constructions, like ‘noun-modifier’,’ subject’, are only intermediate constructs of the hierarchical operation of the rules.