Digital Goethe

Humanities scholars in the engine room

Interdisciplinary workshop in the digital era - for texts' analysis, CRETA provides a novel, interdisciplinary approach.

What do Goethe’s The Sorrows of Young Werther and Adorno’s Aesthetic Theory have in common with the Parzival romance and speeches made at the German parliament (Bundestag)? They are the focus of a collaborative project at the University of Stuttgart that promises new insights for Humanities scholars. The project involves philosophers, political scientists, linguists and literary scholars, who aim to drive progress in the field of digital research methods. In collaboration with computational linguists and experts in visualisation technology, the researchers are developing programmes to enable computers to analyse large volumes of text rapidly and efficiently. Research in the Humanities or Sociology usually entails a lot of reading.

Yet, no matter how much one works on a given subject, the results will always be fragmentary given the plethora of available sources and literature. Computers, by contrast, can swiftly analyse huge amounts of text – but only at the structural level. A collaborative research group at the University of Stuttgart is now giving their digital colleagues a leg up: the Center for Reflected Text Analytics (CRETA) intends to develop algorithms for contents-based textual analysis. The head of the CRETA, Professor Jonas Kuhn of the Institute for Natural Language Processing (IMS), is convinced that this will lead to new developments in textual analysis.

Researchers from multiple disciplines come together in the CRETA workshop to develop digital research methods.
Researchers from multiple disciplines come together in the CRETA workshop to develop digital research methods.

The interdisciplinary field of Digital Humanities combines both the Humanities and Cultural Studies with the systematic use of computer-aided processes for research purposes. The foundations of the CRETA were laid by two projects, both focused on the use of digital resources for textual analysis. A team led by literary scholar Professor Sandra Richter collaborated with Professor Kuhn and his colleagues in the “ePoetics” project to develop tools for the computer-aided analysis of poetic texts spanning three centuries.

Political scientist Professor Cathleen Kantner and her group joined forces with computational linguists in the “e-Identity” project to design a programme to interrogate almost a million newspaper articles to discover the extent to which collective identities, such as religion or nationality, had served as the basis for specific courses of action in crisis situations in the relevant period. Both teams are members of the CRETA, which has received funding for the next three years from the German Federal Ministry of Education and Research. Others involved include representatives from the fields of Literary Studies and Linguistics, the Institute for Visualization and Interactive Systems and the IMS.

As Professor Kuhn explains: “the ministry was interested in establishing digital humanities centres at various universities to promote the digitalisation of scientific methods.” He is particularly pleased that the University of Stuttgart has established a professorial chair for the Digital Humanities at the Institute of Literature, which was one of factors that contributed to the CRETA being awarded the funding in early 2016. The current holder of the professorship, Professor Gabriel Viehhauser, is also a member of the CRETA team.

Computers Simplify Text Analysis

As Professor Kuhn elaborates: “One of the things we do is to canvas opinion among experts in Literary Studies, Political Science, and the Philosophy of Science as to which questions commonly recur in textual analysis in each of these disciplines and then to develop appropriate ways to address them. This interdisciplinary approach is something of a new departure.” Because the CRETA is focused on basic research, one of its main outputs are prototypes for machine learning models, which are then continuously optimised. Professor Kuhn uses an example to explain how research carried out on this basis might look. It may be possible, for example, to develop software tools that would parse historic texts for references to dates, locations or even certain linguistic features.

In the long-term, it will be necessary to include those with a fundamental interest in the impact of specific texts in the development of the relevant technologies.

Professor Jonas Kuhn, head of CRETA

“If I then combine three tools like this to analyse huge bodies of text”, says Kuhn, “then I may notice surprising configurations that I may never have stumbled upon through traditional reading.” The participants meet twice per year for a three-day CRETA workshop to develop tools likely to be of use in several disciplines.

The first project to crystallise out in this context addresses the need to be able to recognise specific entities, such as dates, people and locations, in texts. Whilst it is true that parsing software can already identify specific entities in texts, researchers in the Humanities need to be able to use other categories of analysis. To enable the computer to recognise the criteria compiled by the team, the sample texts are marked up with a series of annotations. In parallel, the computational linguists are developing tools designed to identify and analyse interesting linguistic features in texts at different levels, whilst the computer scientists are looking at ways to visualise the results. Every two weeks, the researchers discuss their progress and any difficulties they have encountered or which may arise, which helps them to refine their tools.

Online Task Sharing

The group has ventured into terra incognita by publishing “shared tasks” on the Internet. The published corpus includes excerpts from Werther, Parzival, and Aesthetic Theory as well as speeches, including some from Dr. Angela Merkel. Researchers around the globe are then invited to try out various analytical tools and approaches. “It is interesting per se that one can annotate the same entities in such a heterogeneous collection of texts”, says Dr. Nils Reiter, who is responsible for the internal coordination of the CRETA project.

“We’re keen to develop this shared task idea”, he continues, “because we believe that the concept is of interest within the field of Digital Humanities.” In the medium term, findings from the CRETA will also be exploited within the wider economy. Some firms, for example, want to be able to automatically record how consumers rate products on Internet sites. Professor Kuhn is certain that “it will be necessary in the long-term to include those with a fundamental interest in the impact of specific texts in the development of the relevant technologies.” That is exactly what the CRETA is about – far more than Goethe, Adorno and Merkel.
Daniel Völpel

  • Prof. Cathleen Kantner, Institute ofSocial Science,Department of International Relations and European Integration  (SOWI III), phone +49 711 685-83425, E-Mail, Website
  • Prof. Jonas Kuhn, Institute for Natural Language Processing, phone +49 711 685-81365, E-Mail, Website
  • Dr. Nils Reiter, Institute for Natural Language Processing, phone +49 711 685-81354, E-Mail, Website
  • Prof. Sandra Richter, Institute of Literary Studies, Modern German Literature I, phone +49 711 685-830 65, E-Mail, Website

Contact

 

University Communications

Keplerstraße 7, 70174 Stuttgart

To the top of the page