A large-scale computational analysis of metaphor density in academic discourse (Elsevier OA CC-BY corpus)

This study presents a large-scale computational analysis of metaphor density across academic discourse, in an attempt to check the traditional perception of the literal “plain style” associated with scientific writing. With its novel methodology, the study analyzed metaphor density across the Introduction, Results, and Discussion sections of research articles drawn from the Elsevier OA CC-BY corpus, across five broad disciplinary domains. The analysis revealed a significant interaction between disciplinary domain and the different sections of research articles. Furthermore, the results of the study show that metaphor density varies considerably more by discipline than by section, lending support to the view that disciplinary epistemic practices are associated with different patterns of metaphor use. In particular, Social Sciences were found to make extensive use of metaphors whereas the Sciences displayed the lowest density among the fields analyzed. Possible interpretations of these differences, such as the abstract nature of fields within social sciences as opposed to a preference for literal, standard terminology in the natural science, are discussed as hypotheses for future research. Thus, the current investigation establishes the prevalence of metaphors in academic writing and challenges the view that academic writing is (or should be) purely literal. These findings also have important pedagogical implications for English for Specific Purposes (ESP), suggesting that advanced academic writing may benefit from discipline-specific metaphoric competence tailored to the unique epistemic nature of disciplines.