You are here

Multimodal Analysis of Interactions in Social Networks

The next speaker on this AoIR 2012 panel is Frauke Zeller, who continues the focus on mixed-methods approaches to online social network analysis. This methodological paradigm is still quite new, and there is considerable uncertainty about best practices in research; network analysis alone cannot be enough.

Frauke specifically highlights the multimodality of social network content, which complicates analysis, and notes issues with the availability and verifiability of content as well. Further, are social networks a tool for research, or an object of study?

Facebook research themes include the questions of who is using Facebook, what users do on Facebook, how people present themselves, what the role of social interaction is, and how privacy and information disclosure are managed. Some of these can be analysed through reactive tools (focus groups, surveys), some through network analysis, some through content analysis, especially where there is a focus on the language used.

Social interactions are manifested in language; our words reflect who we are and what social relationships we are in. This analyses what and how it is said for any form of communicative interaction, including visual communication including images, videos, and profile pictures. Qualitative analyses of images are problematic, however, because images are ambiguous and polysemic; contextual information such as the text-based communication around images should be included here, therefore.

Frauke's study uses Wordsmith Tools and Atlas.ti for computer-assisted qualitative data analysis (CAQDAS), in the first place to find the relevant keywords in a corpus of social media data (drawn from the Greenpeace Facebook page). For specific terms of interests, the word neighbourhood can then be calculated, and the language used around specific images on the page can also be investigated. The specific images used can also be evaluated further, of course. Subcorpora may then be constructed (e.g. based on the comments threads associated with such images).