Humboldt State University professor Monica Stephens and her undergraduate students just released an interactive map showing where hate speech pops up on Twitter around the U.S.
Students searched a year’s worth of geocoded tweets — more than 150,000 — that contained certain “hate words” and individually ranked them positive, negative or neutral, depending on context. Counties with more negative comments than the national average were marked on the map with a color gradient — pale blue for fewer tweets to bright red for higher concentrations.
“We’re really looking at these ways that the Internet relates to material space, how data online reflects material conditions,” Stephens said. “There have been a lot of conversations about the impacts of online bullying and the use of these words on the Internet to target specific groups.”
The British newspaper The Guardian, the io9 Gawker blog and other news outlets have picked up the map. Stephens and her students previously made a map tracking tweets directing racial slurs at President Obama following the 2012 presidential election.
Using her map, people can see the prevalence tweets by homophobic, racist or ableist (derogatory against the disabled) language, and select individual hate words to see where they’re concentrated around the U.S. The data is proportional based on the amount of Twitter activity for each county.
The map isn’t a complete reflection of hate language on Twitter. Geolocated tweets, an account option that can be turned on or off, represent a small portion of Twitter users — mostly smart phone users.
The next phase of the project, Stephens said, is viewing the data against other county demographics like education and unemployment.
Read more at Floating Sheep, a collaboration between Stephens and professors from Oxford, the University of Kentucky and Clark University.