The Embodied Emotions project aims to map the bodily expression of emotions in seventeenth and eighteenth century Dutch plays. Our hypothesis is that historical conceptions of the functioning of the body and the emotions may have been the basis for an altogether different classification of (primary) emotions and for a different bodily experience and expression of emotions. Are fear, grief, rage or joy felt in head, heart or belly? What were actors supposed to say and do when they were acting overwhelming sadness or hate? In cooperation with the Netherlands e-Science Centre, and with Nederlab, we explore the ways in which digital tools can be employed in the analysis of the bodily expression of emotions in early modern plays. A digital corpus of about 800 Dutch plays is provided by DBNL and Nederlab.
The Embodied Emotions Project aims to do three things:
– Problematize the claim that bodily experiences and expressions of emotions are culturally universal.
– Trace historical changes in the bodily expression of emotions on stage in the Netherlands 1600-1830.
– Develop a methodology for tracing these changes by digital means in a sizeable corpus of texts.
As a test case, we have selected a corpus of circa 800 Dutch theatre plays written between 1600 and 1830. This particular genre lends itself well for this type of research, as plays usually contain many emotive expressions and in some cases stage directions that indicate what actors should do. On the one hand theatre displays a fictive world with imagined and very outspoken characters and emotions. On the other, the popularity of theatre and the size and heterogeneity of the audience suggests that the language and expression of emotions met the expectations of contemporaries.
In order to study such a large corpus, we aim to ‘teach’ computers to recognise emotions and the accompanying parts and workings of the body. To do so, we have made a smaller selection of 30 texts and annotated them by hand. Annotation means assigning meta-data to certain elements (words or phrases) in the texts. After that, the computer will be able to learn from these examples, and decide on its own what to do with the other texts, a process that is called ‘machine learning’.
In 2014 we developed an annotation scheme and made the time-consuming manual annotations. Also, ‘agreement tests’ had to be carried out in order to check whether all ‘taggers’ annotated in a consistent way and according to the guidelines. In 2015 we will first analyse the results of the manual annotation and subsequently those of the machine learning process. Simultaneously computer-engineers will work on tools for visualizing the results. We are looking forward to presenting you with the upcoming results.
The project is carried out by:
Director: Inger Leemans, Cultural History, VU
– Janneke van de Zwaan, Escience engineer
– Erika Kuijpers, Postdoc VU, History
– Erik Tjong Kim Sang, Nederlab engineer
– Wouter de Vries student assistent VU
– Alinda den Hoed, Intern Meertens Instituut
– Fieke Smitskamp PhD student VU
Active and affective advisors:
– Isa Maks, Computation Linguistics, VU
– Herman Roodenburg, Historical Anthropology, VU & Meertens Instituut
– Kristine Steenbergh, Literature and Culture, VU
– Nicolien van der Sijs & René van Stipriaan, Nederlab
For further information, please contact Inger Leemans (firstname.lastname@example.org) or Erika Kuijpers (email@example.com)