Corpus-assisted discourse studies (CADS) explores discourse (i.e., language as social practice) through examining corpora (i.e., large computerised sets of textual data). CADS allows one to survey a corpus in its entirety rather than focusing only on certain texts which, by accident or design, may be those that confirm what one wanted to show all along. It thus counteracts the ‘cherry- picking’ charge that has often been levelled at discourse studies (DS). […]. Although discourse is a notoriously fuzzy term (Stubbs, 1983; Baker and Ellece, 2011; Mautner, 2016: 16–24), used in different disciplines and with various shades of meaning, it has a fairly robust semantic core consisting of three key components. Discourse refers to (i) longer stretches of language (usually complete texts and interactions rather than merely single sentences) which (ii) occur naturally in a specific social context and (iii) are analysed as performing social functions. Typical questions that a discourse analyst might ask are: How is language used to represent a particular social group? Which linguistic choices correspond with which ideological position? Do discursive representations change over time? What role do particular linguistic choices play in institutional discourses? In answering these types of questions and many more, CADS is a useful ally. What all CADS projects have in common is that they have a social question at their centre rather than a purely linguistic one. That question may involve an issue such as inequality, poverty, racism, or other social ills. But projects may equally be driven by a more general interest in the links between a social practice and its associated linguistic choices. CADS can make a useful contribution to unpacking what makes discourse tick, which is why it should be an attractive option not only for linguists but also for any researcher investigating the interplay between discourse and society, whether they work in sociology, psychology, law, management, or indeed any other discipline with an interest in discourse.

Gillings, M., Mautner, G., & Baker, P. 2023. Corpus-Assisted Discourse Studies. Cambridge: Cambridge University Press, p1. (link)

The authors adopt a definition of ‘discourse’ as ‘language as social practice’ (p.1) which accords well with Halliday’s (1985) widely-used description of discourse as ‘language that is doing some job in some context’ (p. 10). This is, however, followed by the startling assertion that ‘[w]hat all CADS [Corpus-Assisted Discourse Studies] projects have in common is that they have a social question […] such as inequality, poverty, racism, or other social ills […] at their centre rather than a purely linguistic one’ (p. 1). This definition reflects the particular interest of the three authors, namely Critical Discourse Analysis (CDA), in which ‘discourse’ is a near synonym of social or political ‘narrative’, and which focuses largely on studying power differentials in society […]  In reality CADS is not confined by topic or the stance or school of the researcher/s, whether it be ontological and descriptivist, or normative and political. CADS is an omnivore: ‘[w]hat distinguishes CADS from traditional corpus research is the integration [when necessary] of additional information outside of the corpus during and after the linguistic analysis, namely through inductive, qualitative interpretation in order to uncover “non-obvious meaning” associated with a particular discourse type’ (Skalicky, 2021, p. 591). Much CADS work also concerns itself with language research methodology, including searching for absences, for similarities, using multiple data-sets, content analysis, the consequences of different ways of dividing data, multi-modal CADS, re-evaluating the qualitative-quantitative ‘dichotomy’, rather than having any particular social question as their core emphasis (Taylor & Marchi, 2018).

Partington, A. 2023. Book Review: Gillings, M., Mautner, G. & Baker, P. 2023. Corpus-Assisted Discourse Studies. Cambridge Elements. Journal of Corpora and Discourse Studies, 6(1):53–60 (Link)