Diachronic analysis is intrinsically comparative, in fact ‘a prerequisite for diachrony is that at least two different time points are compared’ (Jucker & Taavitsanien 2014: 5). The way we divide up the data, i.e. what we choose to compare, determines what we see. However obvious this may seem, there is remarkably little discussion (a notable exception being Gabrielatos et al. 2012) about the effects of time segmentation in diachronic discourse analysis using corpora, or Modern Diachronic-Corpus Assisted Discourse Studies (Partington ed. 2010).