Towards criterion validity in classroom language analysis: methodological constraints of metadiscourse and inter-rater agreement

Resumo

This paper reports on a process to validate a revised version of a system for coding classroom discourse in foreign language lessons, a context in which the dual role of language (as content and means of communication) and the speakers' specific pedagogical aims lead to a certain degree of ambiguity in language analysis. The language used by teachers and students has been extensively studied, and a framework of concepts concerning classroom discourse well-established. Models for coding classroom language need, however, to be revised when they are applied to specific research contexts. The application and revision of an initial framework can lead to the development of earlier models, and to the re-definition of previously established categories of analysis that have to be validated. The procedures followed to validate a coding system are related here as guidelines for conducting research under similar circumstances. The advantages of using instruments that incorporate two types of data, that is, quantitative measures and qualitative information from raters' metadiscourse, are discussed, and it is suggested that such procedure can contribute to the process of validation itself, towards attaining reliability of research results, as well as indicate some constraints of the adopted research methodology.