Blending research methods: Qualitative and quantitative approaches to researching computer corpora for language learning.
Résumé
This paper outlines how corpora (in printed, electronic or multi-modal form) can be used in language learning, an area often referred to as "data-driven learning" or DDL (Johns 1991). The alleged advantages are numerous, but are in need of empirical support which is frequently claimed to be lacking in the field. However, over 80 studies have so far attempted to evaluate some aspect of corpus use by non-native speakers (Boulton 2010): these are briefly reviewed as a whole, highlighting the general findings without ignoring problems. Empirical research is essential for the credibility of any aspect of language teaching and learning, and traditionally divides between qualitative and quantitative approaches (e.g. Richards 2009; Lazaraton 2005). Neither however is without its difficulties, as DDL research shows: quantitative studies tend to smooth over individual differences, and experimental design often leads to small or non-significant results; qualitative research can be difficult to interpret and extend beyond the initial context. The results of empirical DDL research are then reanalysed in the light of complex / dynamic systems theory (e.g. de Bot 2008; Larsen-Freeman & Cameron 2008), which allows for greater blending of qualitative and quantitative research methods, and better integration and evaluation of real classroom use.
Domaines
Linguistique
Origine : Fichiers produits par l'(les) auteur(s)
Loading...