Me and my colleague will be in Utrecht, Netherlands, for CAQR2009 (2nd International Conference on Computer-Aided Qualitative Research) on June 4th and 5th. Our paper is related to my project on crime victim discourse, but the presentation will focus on methodological aspects.
DISCURSIVE NETWORKS: VISUALIZING MEDIA REPRESENTATIONS OF CRIME VICTIMS USING PAJEK
This presentation will outline a method for text analysis which combines qualitative discourse analysis and quantitative network analysis. The approach has been developed as a response to the fact that the traditional variety of quantitative content analysis tends to decontextualize data to an extent that makes results potentially meaningless. On the other hand, much of qualitative discourse analysis – a field from which the authors themselves originate – is quite insensitive to things such as frequencies and correlations. It seems ideal to be able to combine the advantages of the two approaches without losing too much complexity.
One of the challenges faced by qualitative text analysis in the 21st century is related to the specific considerations that need to be made when data collection or fieldwork is carried out on the Internet. This presentation concentrates on one particular type of online data, namely print newspaper articles available in digitized form from fulltext databases. Our specific analytical example comes from a research project in which media representations of crime victims have been analyzed.
We will introduce a method that can be useful in working with such material. Even though the approach could be applied to any type of text data, its advantages become more apparent in the case of online fulltext data. This is because the sheer volume of text collected in this way – and also quite fast − easily exceeds the amount collected through traditional fieldwork or manual archival studies. In order to come to grips with these large corpora of text, some sort of quantitative strategy of analysis is called for. But since the straightforward word counting of standard content analysis is not a viable option from the perspective of cultural analysis, we want instead to sketch out an approach combining the discourse theory of Laclau & Mouffe with bibliometric and network analytical tools. In our particular case, we have used the freeware applications Bibexcel and Pajek in order to prepare the text data, analyze co-occuring concepts within in, and visualize the results in the form of vector based network maps.
