Writers’ uncertainty in a corpus of scientific biomedical articles with a diachronic perspective

The communication of the uncertainty of a scientific finding largely determines whether that information will be translated to practice. Unfortunately, however, our ability to study these phenomena is restrained since existing uncertainty corpora are limited in their number of full text articles and in their provision of a diachronic perspective. We analysed a historical corpus through a random sample of 167 years (1840–2007) of articles published in the British Medical Journal. Randomization was stratified from four distinct time periods. The Uncertainty Markers (UMs) and their linguistic scope were tagged in each full-text article in order to answer the following main questions: (1) Which and how many lexical and morphosyntactic UMs are used by writers in order to communicate their own uncertainty? (2) How much uncertainty (UMs + their scope) is present in each article, in each period and in the whole corpus? (3) Is there any significant variation in the use of UMs and their scope along the 167-year span? Although the analysis revealed significant differences in two of the six categories of UMs (non-verbs and modal verbs in the conditional mood), the amount of certainty and uncertainty along the four periods revealed no significant variation.