Hypothesis-testing in corpus-assisted discourse studies

Abstract This paper chronicles a falsified hypothesis and reflects on the ways we ask questions in corpus-assisted discourse studies. It tests the expectation that representations of life before Covid, appearing in the news during the pandemic, would be fraught with nostalgic discourse. The analysis looks at discourses surrounding linguistic markers signalling time preceding the pandemic, in order to reveal which topics and moods dominated the discussion in the British press, using a large corpus of 2021 newspapers. In doing so, the paper raises questions about representativeness and cherry picking (i.e. selecting examples that suit an argument), and it explores issues of operationalisation. I demonstrate the value of falsification in corpus-assisted research through a reflection on the failure to get the expected results. The topic of nostalgia serves as a vehicle to explore ways in which pragmatic function could be studied using corpora, with pandemic-related news as the context for our experiment.