There are some things I really miss about academia, specifically engaging with people about new ideas. I also realized a couple weeks ago that I was really unaware of the current literature outside of economic educational and infection disease stuff. I’m trying to read something weekly and to share it as a way to get caught up/create discussion/get smarter. I have to say it’s surprisingly hard. I’ve been really struggling with some writing recently and have to come to miss my past academic-skillz. The below was easier for me, ironically, because it was so math-not-theory-based.
Bryden, John, Sebastian Funke, and Vincent AA Jansen. “Word usage mirrors community structure in the online social network Twitter.” 2013. EPJ Data Science, 2:3.
“This indicates how the language we use bears the signature of societal structure, and is suggestive of the enormous potential in using topological analysis to identify cultural groups.”
Summary: Bryden et al. use a dataset of 250,000 Twitter users, trying to find linguistic links inside communities, e.g. those who @ed each other. They conclude that communities use unique language patterns beyond basic subject terms, particularly word length and endings, and find a way to predict community involvement of a Twitter user based on word usage.
Analysis: I love the idea of created spaces using language. One of my big interests in college was kawaii culture and how specific terms and ways of talking created an individual/safe space for a sub-set of women in Japan. Unfortunately most of their examples don’t seem to be all that unexpected. For example, one language pattern is the use of Twilight terms in the Twilight community. Another is the use of phrases such as n**ga, poppin, and chillin together; language that is created off-line and then brought online.
That said, there were a few examples where the online community itself (if not specifically on Twitter) was creating linguistic trends, such as the interaction between the words bieber, pleasee, and <33, which they define as “lengthened endings (repeated last letter).” I’d love for Bryden et al. to present more unusual examples like this for greater analysis, maybe some qualitative to understand what those linguistic patterns mean to the community that uses them, particularly in in-group vs. out-group interactions, how people learn the language, etc. There also seems to be a lot of potential in looking at community drift, both in the language a community uses over time but also how language used changes when a Twitter user enters or exits new communities, based on changed interests, life experiences, etc.
Overall: short and sweet if math-heavy; the charts are worth checking out on their own.