A Data-Driven Analysis of Reader Viewpoints: Reconstructing the Historical Reader Using Keyword Analysis

Abstract

This study uses corpus-linguistic methods to examine the relationship between language usage patterns and divergence in text interpretation. Our target of analysis is a set of texts (Czechoslovak presidential New Year’s addresses from 1975 to 1989), which contemporary readers consider repetitious and devoid of content. These texts were statistically contrasted with corpora from two different periods: one from the totalitarian period and the other from the contemporary (post-totalitarian) period. The comparison was based on the Difference Index, the most recent effect-size estimator, which was used to enhance the interpretation of keyword analysis outcomes. The two analyses yield significantly different results: the data from the analysis using the contemporary corpus were commensurate with contemporary readers’ impressions; those from the analysis using the totalitarian corpus fluctuated in tandem with (and sometimes in anticipation of) political and social changes during the 15-year period and suggested an interpretation of the texts by a reader more familiar with totalitarian texts.