Analysing intra-textual patterns in corpus-assisted discourse studies

Many corpus-based discourse analysts are interested in linguistic patterns that hold across the texts in their corpus. In such corpus-based studies, relatively little attention is given to ‘intra-textual’ patterns or discourse organisation/text structure. For example, a standard practice in corpus-assisted discourse studies is to generate lists of word frequencies, keywords, collocates, and n-grams on the basis of total word frequencies in the corpus, without analysing how these words are used to structure texts or conversations. In this commentary, I reflect on this issue, and introduce selected techniques for integrating analysis of text structure/discourse organisation in corpus-assisted discourse studies, including the analysis of dispersion/concordance plots in media data, the use of clusters/n-grams across sentence breaks in conversational data, and the repurposing of parallel concordances for analysing Twitter posts and responses to them.