Choosing and building corpora to investigate representation

Abstract

This chapter discusses the main principles that should be taken into account when choosing an existing corpus or building a new corpus for the purpose of analysing representation. It argues that the suitability of the design of a corpus should be determined by the research question(s) of the user. The chapter presents a range of case studies from research in corpus-assisted discourse studies (CADS), demonstrating the use of publicly-accessible general corpora as well as the creation of bespoke specialised corpora. Based on the themes of the case studies, which investigate text-types that are popular in CADS research (including news discourse, political discourse and social media), the chapter provides detailed and practical advice about accessing or creating corpora for the study of representation in discourse. This is informed by critical awareness of the centrality of corpus design as a crucial component of the research process, and this principle is reinforced by practical activities in manual and semi-automated text retrieval.