Sprinkled Metonymies in the Analysis of Political Discourse with Corpus Linguistics Techniques: A Case Study.

This paper argues that a traditional corpus linguistics approach to the analysis of the political discourse of Barack Obama may fail to detect some of his discourse strategies. To better describe this phenomenon, the concept of sprinkled metonymies – considered a typology of extended semantic networks – is introduced and the adaptation of specific software to enhance data mining techniques to identify them is discussed. The realization of these networks is accounted for by introducing in the working model the Discourse Space-Time and Proximization Theories. The model is designed to detect and cognitively map gradual shifts in strategic discourse, even when considering a time span of 6 years, as in the case study considered, that could otherwise go unnoticed if other corpus linguistics approaches were used.