In this chapter, applications of corpora to the practice of language teaching (mainly English language teaching) will be considered. There are sections devoted to data-driven learning and reciprocal learning, and to issues relating to methodology and syllabus design. The final section is a discussion of recent challenges to the use of corpora in language teaching.

Data-driven learning

Introduction

As Leech (1997c: 3) comments, the use of corpora in language teaching situations owes much to the work of Tim Johns, who developed data-driven learning (DDL) for use with international students at the University of Birmingham. An often-quoted comment by Johns is that ‘Research is too important to be left to the researchers’ (1991: 2). The theory behind DDL is that students act as ‘language detectives’ (Johns 1997a: 101), discovering facts about the language they are learning for themselves, from authentic examples. This supports learning, partly because students are motivated to remember what they have worked to find out. In addition, because corpus data can reveal previously unnoticed patterns, a student may well notice something that a teacher has overlooked, or that no textbook covers. As well as being beneficial in teaching specific items, DDL is hypothesised to improve general skills of using context to deduce meaning. DDL involves setting up situations in which students can answer questions about language themselves by studying corpus data in the form of concordance lines or sentences. The questions may arise out of something the student is writing, and may be formulated as ‘Is it better to say x or y?’ or ‘What is the difference between saying x and saying y?’ In this case, the questions are the student's own.