Functional Discourse Units for Second Language Research

Abstract

DISCOURSE ANALYSIS has been an accepted part of the methodology of second language (SL) research for some time (see, e.g., Chaudron, 1988, pp. 40-45; Hatch, 1978; Hatch & Long, 1980; Larsen-Freeman, 1980), and its procedures are becoming increasingly familiar and even to some extent standardized. As is well-known, an important preliminary stage in the discourse analysis of speech is the the identification of units relevant to the investigation within the body of text to be analyzed, or the complete separation of the corpus into those basic units. In carrying out this stage of the analysis, the varying needs of second language discourse analysis have caused investigators to make use of all of the traditional grammatical units of analysis (morpheme, word, clause, etc.), as well as other structural or interactional features of the discourse (e.g., turns), in addition to a variety of other units defined in terms of their functions (e.g., moves). These units have formed the bases of a number of different analytic systems, which can mainly be classified as either structural or functional (Chaudron, 1988), developed to address differing research objectives. In carrying out structural discourse analyses of oral text, researchers have been confronted with the fact that most grammars are based on a unit that is not defined for speech, but is based on the written mode of language that is, the sentence. Various different units have been applied to replace the sentence, but there has been little comparative discussion concerning basic units of analysis, considered as options from which the discourse analyst must select, nor of desirable criteria for s_election. In the following survey, the most common of these items will be defined, and their origins and function discussed: the T-unit (and related variants), the turn, the utterance, and (because of its prominence in first language analysis), the tone unit. Then, the relationships which exist between them will be described as far as possible. The final section of the paper discusses these units from the viewpoint of basic criteria for the selection and evaluation of discourse analysis systems, and argues for the preferability of the utterance on these grounds.