Towards a semi-automatic extraction of conceptual metaphors in a corpus of economics from natural language processing

Abstract

This article presents the linguistic phenomenon of conceptual metaphor applied to economics from the computational and corpus linguistics perspectives, approached through the analysis of n-grams and collocations. To this end, a corpus of articles on economics has been created, divided into two sub-corpora represented by academic journals in that area. The textual analysis of the automatic recognition of metaphors was carried out using computer tools and software for natural language processing, including some specialized Python libraries and the LancsBox tool. The aim is to explore the initial hypothesis, i.e., the discourse used by economists involves a tendency to use, to a large extent, conceptual metaphors that help to understand phenomena related to economics and that clearly underlie academic publications. Finally, the results from this research work offer an approach to visualizing the most commonly used conceptual metaphors in economics, which would validate the hypothesis mentioned above.