Comparing Language Communities: Characterizing Collaboration in the English, French and Spanish Language Editions of Wikipedia

Abstract

Is Wikipedia a standardized platform with a common model of collaboration or is it a set of 312 active language editions with distinct collaborative models? In the last 20 years, researchers have extensively analyzed the complexities of group work that enable the creation of quality articles in the English Wikipedia, but most of our intellectual assumptions about collaborative practices on Wikipedia remain solely based on an Anglo-centric perspective. In my dissertation work, I aim to understand how collaboration models on Wikipedia generalize across online language communities. This dissertation extends the current Anglo-centric body of literature in human-computer interaction (HCI) and computer-supported cooperative work (CSCW) through three studies that mutually help build an understanding of collaboration models in the English (EN), French (FR) and Spanish (ES) editions of Wikipedia. In the first study, I replicated a model by Viégas et al. (2007) based on editors' behaviors in the English Wikipedia. This model was used as a lens to examine collaborative activity in EN, FR and ES. In study two, I leveraged a collaboration model by Kriplean et al. (2007) that suggested editors used “power plays” – how groups of editors claim control over article content through the discourse of policy – in their talk page debates to justify their edits made on articles. In study three, I interviewed editors from the English, French and Spanish language editions to build a typology of collaborative behavior and understand editor’s perceptions of power and authority on Wikipedia. In the first two studies, two well-known collaborative models discovered in EN were replicated in EN, FR and ES. I show that these models manifest differently across Wikipedia language editions. In Study 1, across all languages, the editor had similar behaviors, but they exist in different quantities. In Study 2, the qualitative coding of a dataset in EN, FR and ES show that these “power plays” still exist and no new ones were discovered in any of these language platforms across all three languages. Through the participant interviews, I find that across language editions, editors have different perceptions on the factors that drive consensus on Wikipedia talk pages. I use all of the findings to build a more comprehensive model of collaboration in all three languages. In each of these studies, I demonstrate that Wikipedia is a sociotechnical system – an instantiation of both social and technical processes. To better compare each language edition, I specifically draw out the social and technical processes within each of my studies’ findings to understand the differences in collaboration. Further, in this dissertation work, I introduce the idea that language processes also help play a role on Wikipedia talk pages. The fact that Wikipedia has multiple language editions shows that language may have a relationship with broader social issues. Essentially, language and language use manifests into new social processes across different language editions of Wikipedia. The differences that might exist in the language processes between editions is an accessible way of understanding the social processes - this has not been adequately represented other research on Wikipedia language editions. The implications of this study are threefold: (1) empirical contribution, (2) methodological contribution and (3) validation contribution. First, this dissertation presents an empirical understanding of peer production collaboration models in three different language editions. Secondly, I demonstrate a clear example of how to conduct research across multiple sites or contexts. Lastly, from a more intellectual standpoint, this work provides a complete replication or validation of a prior study, the synthesis of two models that already exist in EN and the existence of collaboration models in three languages.