Two hypotheses regarding text generation and analysis

Publisher Website
Google Scholar
Abstract
It is proposed here that in any language and for any discourse type within that language the verb forms/clause structures can be arranged in a rank scheme in which a mainline of discourse development is encoded by a characteristic construction (or a very limited set of constructions) while lines of subsidiary development, which represent progressive degrees of departure from the mainline, are encoded in other constructions. It is further proposed that this graded salience scheme can then provide guidelines for the analysis of local spans of text (paragraphs) so that sentences whose independent clauses have constructions which are high in the salience schemes are dominant over ancillary sentences which have constructions which are lower in the scheme. The first hypothesis has more to do with text generation, while the second has to do with text analysis. The two hypotheses are meant to yield salience schemes and constituent analyses which mutually corroborate and correct each other. These hypotheses and their reciprocity are illustrated here relative to narrative discourse in eight languages in five distinct linguistic areas.

This publication has 6 references indexed in Scilit: