Do You Tweet With an Accent?

It’s amazing how much information can be conveyed in a mere 140 characters, even without your knowledge. For example, did you know that analyzing your tweets can show where you come from?

In fact, in a study last year by Carnegie Mellon University, researchers found that they could narrow down almost any individual Twitter user’s location to within 300 miles based solely on the language they used in their tweets.

Now,  a new study from Ohio State lingistics grad student Brice Russ has demonstrated that Twitter can successfully be used to map linguistic variations across the United States. He examined three markers of regional American dialects: the use of Coke, pop and soda to describe a sweetened carbonated beverage; the use of “hella” to mean “very,” and the use of phrases like “the car needs washed” or “the computer needs fixed.”

He found that Twitter data could be used to map out which parts of the US these variations were used in. His paper may even have a revealed a trend: the “needs X-ed” construction seems to be moving southward compared to previous analyses.

Can analyzing posts on social networks replace traditional linguistic fieldwork? No. As Mr. Russ told the New York Times, “The ‘bobbasheelys’ and ‘crawdads’ of English don’t always show up on Twitter often enough to be mapped on a large scale.” (“Bobbasheely” means “close friend,” while “crawdad” means “Crayfish.”)

Still, it does have some advantages as a way to collect supplemental data.  In his paper [PDF], Russ notes that “Twitter is a very promising source for studying regional variation” because “data can be collected easily and effectively without interviews or supervision.”

Interestingly, earlier work by British linguist David Crystal found that aside from differences in British and American spelling, tweets as a whole “aren’t very regional distinctive,” and that people were more likely to use local dialect in their Facebook posts. It’s enough to make you wonder if the amount of dialect a person uses in their tweets varies depending on whether they’re American or British?

Image CreditAttribution Some rights reserved by topgold