Research Objectives
Objective 3. Develop and apply new tools and methodologies for ethnographic analysis to support large-scale, multi-lingual, multi-media data.
As a research method, ethnography is not yet fully equipped for dealing with studies of the scope and ambition needed to contribute to POPREBEL. This calls for some methodological innovation. We extend the reach of ethnography in five directions:
Scale. Anthropologists routinely study small-scale communities and base their conclusions on interactions with a limited number of people. POPREBEL researchers engage with hundreds of informants, geographically spread out across the continent.
Multilingualism. For maximum inclusivity, POPREBEL operates across several languages. Running completely separate studies would lose out on the critical dimension of comparability across different languages.
Multimedia. Populism diffuses and evolves through sharing videos, photos, GIFs etc., at least as much as through text. While examples of multimedia ethnography exist, we aim to encode all media (including text) within the same data format.
Support for conversational structure. POPREBEL takes a collective intelligence approach to ethnographic work. Its central tenet is that open conversations has, under certain conditions, a self-correcting property.
This happens by exposing one’s thoughts to his or her peers’ scrutiny, just like in science, and is similar to the self-correcting behaviour observed in Wikipedia. In terms of data collection, the raw data point is not the interview, but the contribution to an online platform (a forum). All contributions, in all four languages, are coded directly on the platform. Codes are saved in the same database as the coded material, preserving its conversational context. This is an innovative approach, tested earlier, that in this project is extended in new directions.
Computational methods. The mass of data generated by POPREBEL is likely to exceed the short-term memory capacity of even the best researcher. Accordingly, we borrow methods from network science and encode the conversation (and its ethnographic coding) as a semantic social network. This makes the data (a) reducible and (b) navigable, without any loss of information.