This thesis investigates how the technique of Topic Modelling can give a Digital Humanities scholar an entry point to a body of literary and historical texts. In three exemplary case studies, it shows how the technique can complement a traditional close reading approach, and how a corpus can be analysed diachronically. For this, it applies the technique to the Dutch general cultural magazine De Gids (1837-), which in its early days has proven to be highly influential in the field of literature, culture and science, and nowadays still functions as an archive for the evaluation of culture and science in the recent history of The Netherlands.
In the first case study, the technique is applied to a single volume (72 articles) of De Gids to find information on the Dutch constitutional reform of 1848. The corpus size is scaled up in the second case study (1,642 articles), in which writings on the Dreyfus affair (1894-1906) are found and analysed with help of the topic models outcome. The third and last case study analyses all articles from the first one hundred volumes (10,624 articles) of De Gids on a macro scale to investigate the propagation of disciplinary trends in the magazine by identifying and interpreting them, and by connecting changes in proportion to the magazine’s internal and external factors.
This study shows that a topic model is able to show De Gids’s involvement in politics and liberalism in 1848 and that it is able to highlight other themes and text types in articles from the period of the Dreyfus affair. It thereby shows the applicability and validity of a topic models outcome for literary historical research. The real benefit of the technique exists in applying it to largescale corpora as is shown in the third case study in which fluctuations in the distribution of (academic) disciplines and fields in the magazine over time are visualised.