Algorithmic Discourse Analysis (aDA): A Semi-Supervised Approach to the Study of Intertextual Meaning in Political Discourse

• In Event: Methodology: New Thinking About Methods and Measures

Abstract

To date, political discourse analysis has largely been regarded as a qualitative exercise, chiefly due to its heavy reliance on contextual knowledge and inference formation. In this paper, I propose a perspective upon discourse analysis that explicates and formalizes the major steps and resources involved in the contextualization and interpretation of discourse texts. Drawing upon techniques in computational text analysis, corpus linguistics, and semantic network analysis, I propose an algorithmic strategy for analyzing discourse. Extracting key semantic and relational information expressed in political discourse, it uses intertextual linkages to reconstruct relevant context knowledge and identify sets of information capable of sustaining coherent interpretations. Its detailed representation of both manifest and inferred contextual information enables a systematic appraisal of the contingency and multiplicity of discursive meanings. By explicating tacit knowledge, formalizing intuitive inferencing strategies and modeling their implications for discourse interpretation, algorithmic Discourse Analysis offers important new avenues for research.

Author

• Christian Baden, Hebrew U of Jerusalem