Using Digital Archives in Quantitative Discourse Studies: Methodological Reflections

Abstract

This methodological essay discusses the possibilities of using digital archives in quantitative discourse studies. I illustrate these possibilities by discussing a study in which the digital archive Delpher was used to build a relatively large corpus of newspaper narratives (N=300) in order to test hypotheses about the historical development of linguistic features associated with objective and subjective reporting. The large amount of data collected in digital archives like Delpher facilitates the construction of corpora for such hypothesis-driven studies. However, the collection of newspaper articles on Delpher in fact constitutes only a small, non-random and continuously changing selection of all available data. Due to these characteristics, the use of Delpher jeopardizes two core values of quantitative empirical research: the generalizability and the replicability of findings. Although these issues cannot be easily overcome, I argue that digital archives have the potential to broaden the methodological scope of discourse studies and increase the overall significance of the field.