Over the course of this full-day workshop, we aim to facilitate an in-depth debate about the potentials and pitfalls of Computational Social Science-assisted text analysis (CSS) methods for research projects broadly informed by Discourse Analysis (DA). Discourse Analysis is an inclusive approach that stretches across disciplinary boundaries within the social sciences and humanities, and that can take various appearances. All its versions and sub-traditions are united, however, by a concern with how the way we talk about the world around us involves the construction of positions of power, inequalities, and social norms that privilege some standards and practices as normal, while demoting others to a deviant status.
We offer two formats for participation. Those interested in presenting a theoretical or methodological argument, a critical question, or an interesting example from your work at the intersection of DA and CSS, are asked to submit a proposal before 31 October. We will integrate these (if suitable) in the workshop format, on a rolling basis. Also non-presenting participants are invited to share their views and experiences during the workshop, and should use the Symposium’s registration form.
Discourse Analysis has developed a strong theoretical apparatus to understand how power relations follow from wordings, phrasings, denotations, practices, and framings, but it is always looking for methods to improve how it teases these crucial signifying features out of larger data units such as texts and conversations. The study of large corpora with digital assistance is becoming an ever more popular option in this regard (Baker 2006). Yet many more advanced tools such as topic modelling and semantic network analysis remain underappreciated. Furthermore, the reflection on how DA relates to CSS on a theoretical level remains superficial.
Computational Social Science bares the potential to radically transform existing methodologies and approaches. It makes answering new questions and solving hitherto unsolvable puzzles possible by enabling researchers to trace far more subtle and nuanced variations in far larger datasets. Thus, researchers working on subjects for which large collections of text are potentially available, have to decide whether and how they want to draw from the vast array of new tools made available by statisticians and computer scientists (cf. Grimmer and Stewart 2013). Discourse Analysis in particular has long faced internal debates and external criticism — concerning methodological rigour and questions of validity (e.g. Breeze 2011) — that will take new directions as scholars working with the approach embrace or reject computerised methods (Baker 2006).
This workshop aims to do more than just to demonstrate the practical potential of CSS for studying discourses through examples and case studies. Rather than discussing specific studies in detail, we invite participants to use cases or theoretical arguments as a way to raise critical questions, share methodological dilemmas, and put forward arguments about how Big Data and CSS does, could and should (not) change the way we conduct discourse-oriented text analysis. In the workshop, practitioners working at this intersection, and those curious about it, will be able to test their arguments and deepen their understanding of theoretical, methodological and practical synergies and disjunctures between DA and CSS.
We will tackle different aspects of the ongoing debate about what Computational Social Science has to offer for Discourse Analysis and vice versa: What research questions can be tackled with the help of CSS that have been difficult to address before? How do existing strategies benefit from using CSS tools, and what are their limits? What can practitioners of CSS learn from a DA-inspired reflection about their data and methods? Throughout the workshop, we organise these discussions around four different themes:
The workshop will be organised in four short sessions dedicated to each of the themes, followed by a final open debate. In each session, a selection of 2-3 speakers are invited to present a 10-15 minute input statement (including a 2-minute Q&A), putting forward an argument, presenting a problem, or raising a pertinent question, followed by a brief (20-30 minute) integrating discussion by all workshop participants. At the end of the day, a moderated plenary discussion about the main issues raised throughout the workshop will round of the event (exact timings are subject to change).
|09:30||Session 1 - Ontological and epistemological foundations|
|11:15||Session 2 - The impact of CSS on internal debates and external critiques of DA|
|13:30||Session 3 - Practical challenges of using CSS for DA|
|15:15||Session 4 - Limitations of CSS as a set of tools for DA and other approaches|
|16:30||Final Plenary - How to (not) embrace CSS and learn from DA (1 hour)|
Scholars from all fields of study and disciplines with an interest in CSS and DA are welcome to submit an extended abstract (max. 750 words) of their talk and written contribution (if applicable). The latter can be existing work (published or unpublished) or prepared for the occasion, and should serve as additional background reading for interested participants. The abstract should indicate which of the four themes the talk is intended to fall into, and how it addresses extant debates in the CSS or DA-related literature. Deadline for submissions: 31 October, 2017.
The organising committee will select contributions based on their fit with the overall topic of the workshop and the specific theme of the session. In the selection process, the committee aims, above all, to create the basis for a multi-faceted and nuanced debate, and apart from the subject matter of the contribution, the disciplinary background, area of research, and experience of the participant play a role. The aim is to draw a diverse crowd of computer scientists, data specialists, linguists, social scientists working with discourse analysis, and digital humanities scholars, in order to create a debate that does not get bogged down in subtle nuances of discourse theory or the technicalities of a particular algorithm, but which discusses the meeting of DA and CSS at an appropriately profound and abstract level. We furthermore believe it to be self-evident that a diverse crowd entails more than a trans-disciplinary membership, but also implies geographical diversity and diversity regarding e.g. gender, race, seniority, employment.
Please use firstname.lastname@example.org for all questions and inquiries concerning the workshop.
Robin Tschötschel Amsterdam School of Communication Research, University of Amsterdam, Amsterdam, The Netherlands
Department of Political Science, Ghent University, Ghent, Belgium
Julie M. Birkholz
Department of Literary Studies, Ghent University, Ghent, Belgium
Department of Sociology and Work Science, Gothenburg University, Gothenburg, Sweden
Department of Sociology, University of Amsterdam, Amsterdam, The Netherlands
Baker P. (2006). Using corpora in Discourse Analysis. London: continuum.
Breeze, R. (2011). Critical discourse analysis and its critics. Pragmatics, 21(4), 493–525. http://doi.org/10.1075/prag.21.4.01bre
Grimmer, J., & Stewart, B. M. (2013). Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts. Political Analysis, 21(3), 267–297. http://doi.org/10.1093/pan/mps028