That Vexed Problem of Choice

This chapter argues for a reorientation in corpus linguistics from traditional per-million-word frequency approaches toward choice-based variationist methods. While mainstream corpus linguistics typically cites exposure rates measuring how often readers encounter linguistic forms, this approach conflates the opportunity to make linguistic choices with the actual choices made. The chapter advocates for alternation-based methodology grounded in the variationist paradigm, formulating research questions in terms of speakers’ and writers’ decisions between alternative forms. This treats language production as primary, viewing exposure rates as secondary consequences of producer choices. Using examples from, inter alia , modal shall/will alternation and progressivisation of verb phrases, the author demonstrates how choice-based baselines provide more linguistically meaningful and statistically robust analyses than word-based frequencies. Three major objections to variationist approaches are addressed: feasibility of identifying alternates, claims of arbitrary baseline selection, and arguments about oversimplification of linguistic ecologies. The chapter proposes a methodological continuum from word-based frequencies through refined baselines to verified alternation sets. Central is the principle of mutual substitution, requiring alternative forms to be genuinely replaceable within contexts. This leads to sophisticated experimental designs using decision trees for multiway choices and binomial statistics. The variationist perspective promotes the identification of genuine choice points, focuses hypotheses, increases analytical robustness, and enables precise investigations of language variation by factoring out opportunity variation from choice variation.