A methodological progress report

Abstract by J. Berenike Herrmann (FU Berlin & DH Lab University of Basel), Giulia Grisot (DH Lab University of Basel), Jonas Waeber (University Library Basel)

In our presentation we will report on the first steps taken in our project “Distant Reading Topographies of Sentiment in German Swiss Novels in the early 20th Century,” funded by the SNSF and based at the DHLab at University of Basel. We will first introduce our framework of computational geography of literature and computational literary affect studies applied to the literary landscape of literature written in German around 1900. In the main part, we will report on the current state of the computational method we have been developing for detecting emotion in relation to representations of (fictional) space, addressing issues of construct validity, accuracy, and computational resources in application to literary German discourse around 1900.

To examine how geo-entities are depicted emotionally in our corpus, we develop an semi-automatic procedure using existing Sentiment Analysis and Named Entity Recognition resources (available for the programming languages R and Python). To this aim, we have been revisiting sentiment analysis tools (such as BAWL-R, SentiWS and EmoLex), and are currently compiling a comprehensive dictionary of natural entities characteristic of ‘landscape‘ and ‘urban space.’ This list is inclusive of spatial entity terms commonly used in the historical context of the corpus (Weiher, Weg, Hütte, Berg, See; Strasse, Gebäude, Dom), as well as of spatial named entities (Berlin, Rhein, Salzburg, Luzern). Implementing “corpus literary studies” (Herrmann & Lauer, 2019), the project taps into resources of digital library and information studies in order to gauge the overall population of texts and authors attributable to “Swiss” national literature.

We will report on our method for creating a descriptive account of authors and texts using manual as well as multiple library and semantic-web resources, such as the German and Swiss National Library data, including GNDs, Lobid, Wikidata, VIAF, swissbib. The paper will also discuss first findings gathered by means of a pilot analysis of the representation of emotions and sentiments in relation to so-called ‘geo-entities’. Eventually, we aim at building the grounds for understanding whether the representation of place and space can be considered as part of the construction of different national (specifically ‘Swiss’) identities in those years.