Reading at Scale
Open access
Date
2023-10Type
- Book Chapter
ETH Bibliography
yes
Altmetrics
Abstract
The Deutscher Novellenschatz (published in 24 volumes 1871–1876) is a collection of 86 German-language novellas edited by Paul Heyse and Hermann Kurz. It is an example of a medium-sized corpus amenable, in principle, to both, scholarly reading and automated analysis. Our point of departure was the conviction that research questions at intermediate granularity would require the combination of hermeneutic and statistical methods to achieve an appropriate level of abstraction while maintaining a sufficient amount of context. By exposing the sensitivity of text similarity measures to choices in the preparation and evaluation of bag-of-word representations, we highlight the need for consideration of contextual information even in the most distant reading approaches. Literary theory suggested more coarse-grained hypotheses that we tested both empirically in a group of non-expert readers and computationally using similarity of character constellations, respectively. Based on correspondences of the editors and comparison with other corpora, the Novellenschatz was further situated in a historiographic context. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000635998Publication status
publishedBook title
Mixing MethodsJournal / series
Digital Humanities ResearchVolume
Pages / Article No.
Publisher
Bielefeld University PressOrganisational unit
09610 - Brandes, Ulrik / Brandes, Ulrik
Related publications and datasets
More
Show all metadata
ETH Bibliography
yes
Altmetrics