OPEN THREAD: Interactive Map Tracks Hate Speech On Twitter -- How Does Your Region Stand Up?

A new interactive map tracks "hate speech" on Twitter over the past year, and the results look pretty much how you'd expect. Which is exactly why I'm suspicious of them.

May 14, 2013 at 3:30pm | Leave a comment

You may remember Floating Sheep, masters of geotagging analysis nerdery, from their project last November, in which they mapped out racist tweets flooding Twitter in the wake of President Obama’s reelection -- a project for which they caught no small amount of criticism. 
 
They heard the grievances and are back at work depressing everyone again, this time mapping all geotagged “hateful” tweets (specifically homophobic, racist and disableist) posted between June 2012 and April 2013, with a slightly different process:
 
…[S]tudents at Humboldt State manually read and coded the sentiment of each tweet to determine if the given word was used in a positive, negative or neutral manner. This allowed us to avoid using any algorithmic sentiment analysis or natural language processing, as many algorithms would have simply classified a tweet as ‘negative’ when the word was used in a neutral or positive way. For example the phrase ‘dyke’, while often negative when referring to an individual person, was also used in positive ways (e.g. “dykes on bikes #SFPride”). The students were able to discern which were negative, neutral, or positive. Only those tweets used in an explicitly negative way are included in the map.
 
The final collection of 150,000 tweets was then aggregated by county and normalized according to the overall number of tweets in said county -- basically, for the map to be useful, raw data is not real helpful as it’s natural that the areas with the highest overall Twitter activity would also have higher instances of hate speech, which could give the wrong impression that simply using Twitter makes you more likely to be vocally racist. The result is thus “a comparison of places with disproportionately high amounts of a particular hate word relative to all tweeting activity.”
 
 
The overall findings are that hate speech happens all over! Big surprise. Unfortunately, the concentration of red over certain parts of the country has led many to take this as gospel truth that the American South is SUPER WAY MORE RACIST/HOMOPHOBIC/EVIL than the rest of the country, which is a misleading assumption to say the least.
 
All this map really measures is the frequency with which people in certain areas use certain words as negative terms -- and while language is absolutely important in how we all create the culture in which we live, it’s not the only factor. 
 
Obviously, you can be a deeply racist person and never utter the n-word in your life. Unfortunately, many people seem to be reading this map to mean that the places more rife with negative occurances of words like “wetback,” “cripple,” and “fag” are intrinsically more hateful than places where these words are considered more taboo. Racism in Massachusetts behaves differently from racism in Kentucky, but both are real. Just because New Englanders recoil in horror from a Confederate flag doesn’t make them immune from racism, and while some regional cultures are reluctant to use certain slurs and this may make them seem less racist, that doesn't mean the racism that does exist in these places is less bad or more acceptable.
 
Racism -- and homophobia and hate of all kinds -- is a culturally-distinct phenomenon that differs dramatically from place to place. So while Floating Sheep’s efforts may be interesting from a linguistic perspective, they don’t necessarily say anything substantive about hate in the United States. At least I don’t think.
 
But check out the fully interactive map for yourself. Do you live in a place with a lot of Twitter hate speech going on? Do you think this map offers some legit insight into American regional culture -- or more to the point, do you believe that the frequency of hate speech in some places means they really ARE way more racist than others? Let’s work it out in comments.