A diachronic corpus-assisted semantic domain analysis of US presidential debates

Abstract

This corpus-assisted discourse study investigates diachronic change in a specialised corpus of seventeen American presidential debates from 2000 to 2020. The texts were tagged using the ucrel Semantic Annotation System (usas) ( Rayson, 2008 ) to facilitate the investigation of emergent and decreasing semantic trends over the period; the strength of trends was empirically evaluated through application of Kendall's Tau correlation coefficient. The analysis revealed that domains reflecting truth evaluations and matters of credibility increased alongside a more people-orientated discourse, as evidenced by increases in personal pronouns. Furthermore, instances invoking warfare and defence decreased, paralleled by a decrease in the representations of toughness. These results may reflect a shift in US political discourse generally, and American presidential discourse specifically, while also reflecting evolving contemporary social and political interests over the twenty-year span of the corpus. This study concludes with interpretations of these discursive shifts in the context of the current era of so-called ‘fake news’, intense partisanship, and social and political divisiveness. Findings indicate that the current US political climate cannot simply be attributed to an anomalous Trump administration but, rather, the discursive features contributing to and reflecting the current political environment have been present and increasing since at least the year 2000 election cycle.