ABSTRACT
This chapter discusses the difficulties involved in determining representativeness in language data, and illustrates the issues raised by the bias of corpus-assisted discourse studies (CADS) towards particular genres – typically official, public and factual ones. It also discusses how fiction and imagination are central to understanding of the real world and sketches some of the complex ways in which readers' affect and attention are engaged by imaginative discourses. The chapter addresses disciplinary divides about fiction and outlines some of the particular interpretive caveats required to deal with such materials. It offers an example of how these limitations can be addressed by exploring the role of erotic fiction in the circulation of discourses about gender and sexuality. Instead of 'considering the whole network of genres to understand the functioning' of a specific discourse domain, much CADS work is limited to snapshots of particularly salient junctures. Fragmentation has been observed in factual genres such as news and advertising, as well as literature.
