Elements of Style Change

Abstract

This thesis considers aspects of stylistic change over time with respect to a corpus of literary authors and corresponding background change for the same period of time. The primary focus of this thesis is the development of methods for the analysis of authorial style over time, in particular with respect to how ageing affects language over the lifespan, what type of linguistic features are particularly changeable and how different features interact with each other. Rather than providing an exhaustive treatment of stylistic change with respect to the examined corpora, the purpose is to extend the toolkit for analysis and interpretation of diachronic style.

As part of the exploration of ageing effects in literary language, a comprehensive model is proposed taking into account background language influence when analysing effects in individual authors. While there is evidence of underlying language change in the variables previously linked to ageing, these findings could not be replicated in the literary authors examined here.

In order to detect features that change over time, a prediction task is proposed taking as response variable the year a text originated in. Having identified interesting features in individual authors, as before a general model is proposed to take into account how the background language changed at the same time, thus separating effects into what is individual and what is general.

Finally, the last part examines how features that appear in all time instances relate and are influenced by other less regularly appearing items. While this analysis indicates that literary language may change more gradually, for the background language corpus temporal expressions in the news data emerge as interesting candidates. The results and comparisons to other genre suggest that there may indeed have been clusters of irregular words that affected the frequency of the more regularly appearing expressions.