• deutsch | english

Emotion and argument in digital information environments

A central development of the last twenty years is the movement of societal discussions, which were traditionally determined by print media, radio, television, and their structures, into the digital world. While "old style" discussions that are shaped by gate keepers such as media and experts continue to exist in the digital domain, there is an ever expanding scope for groups of laypersons, provided in particular by social media, to self-organize and communicate within the group. In addition to many positive outcomes, this development can also turn problematic, for instance by creating so-called "filter bubbles" in which every user feels surrounded by a large, seemingly representative crowd of like-minded individuals.
A question that becomes highly relevant in this context is how discourses in such digital forums are structured. In particular, we believe that it is interesting to compare the respective prominence of argumentative structures (corresponding to ‘classical’ expert discourses) and emotional components (following the patterns of tweets and other short message formats). This question brings together computational linguistics and psychology. 

* On the side of psychology, we ask which factors are involved in  the relationship between argument and emotion; which are key for convincing the reader; and what the role of the reader's state is (e.g. motivation, stance towards topic).

* On the side of computational linguistics, we ask how we can automatize the identification of these key factors for large collections of texts of different text types; and how we can automatically vary the emotional content of the text in order to produce texts that are more convincing or, conversely, easier to process for machines.

Project Team