Corpus linguistics will benefit from greater adoption of pre-registration: A novice-friendly split-corpus approach to pre-registration

Abstract

In this brief article, I contend that the field of corpus linguistics stands to gain significantly from an increased adoption of pre-registration. Pre-registration serves to constrain the almost infinite degree of analytic freedom inherent in corpus analysis, thereby enhancing the transparency, reliability, and potential impact of corpus research. While pre-registration is increasingly popular in fields such as psychology and medicine, its uptake in corpus linguistics remains notably limited. To facilitate the transition toward pre-registration, I describe a straightforward split-corpus approach, ideally suited for corpus linguists new to pre-registration and for both hypothesis-testing and exploratory research. This method involves dividing a corpus into an exploratory set (20–40 % of the corpus) and a confirmatory set (the remaining 60–80 %). The exploratory set allows researchers to freely generate hypotheses and develop analysis plans, while the confirmatory set is then used for a more structured and objective analysis according to the pre-specified protocols. By employing this approach, corpus linguists can effectively balance exploratory flexibility with the rigour of confirmatory analysis, boosting the reliability of corpus findings. An increased uptake of pre-registration may not only bolster recognition of corpus linguistics as a robust empirical field, but it may also encourage a stronger emphasis on the building of cumulative knowledge.