Public Discourse Semantics. A Method of Anticipating Economic Crisis

Abstract

This paper provides a proof that anticipation of an economic crisis by analysing public discourses (in particular, speeches on economic issues) is feasible. It proposes a method of text classification and semantic interpretation based on natural language processing techniques that could be used to trace, over a period of time, the print press discourses, with the aim to valuate the perspective of occurrence of crises. Classification is the task of assigning tags (words, expressions) to the texts that make up a corpus. In our case, we were interested to identify among the texts under scrutiny those belonging to classes like financial, economic, nationalism, etc. This approach is sustained by the fact that public discourses can be characterized from a rhetorical perspective, depending on the specific strategies their authors have chosen: orientation to change opinions or to determine action, ratio between rational (logos) and emotional (pathos), etc. We are sugesting an automatic analysis of the content of the public language, by using quantitative measures. Our purpose was to develop a computational tool able to offer to researchers in the economic, social or political sciences, but, not less, to the public at large, the possibility to measure the acuity of different accents of a written public discourse (financial, emotional, etc.), as mean to anticipate the threat of financial waves. Such a tool could help the processes of decision making in the analysis of crisis. Although our analysis used as data the journalistic and economic environments of Romania, it could easily be extrapolated to other languages/countries.