Carnegie Mellon computer scientist Jacob Eisenstein and his colleagues looked for geotagged tweets – messages that were marked with their tweeter’s location. They collected one week’s worth of messages in March 2010 from people who tweeted at least 20 times during that week. That gave them a whopping 380,000 tweets from 9,500 users.
The researchers found that well-known regionalisms were thriving on Twitter. Northern Californians used “hella” to mean “very” or “a lot,” and Pittsburgh residents used “yinz” to mean “you guys.” (Southerners, naturally, tended toward “y’all.”)
“When you think about mass media it would be natural to assume regional differences would start to disintegrate as we watch all the same movies and TV,” Eisenstein said, “but there’s strong evidence that the way people talk has become more different rather than more similar.”
Take New York City tweeters' “suttin” as opposed to the more widely used “sumthin,” or the way that mid-Atlantic users tended to employ LLS (rough translation: laughing like poop) instead of LOL. Among the various spellings of “for sure” (i.e. “fo sho” or “fsho”), Angelenos often leaned toward “fasho,” Eisenstein said.
Spelling has historically not been widely examined by linguists studying conversational language, Eisenstein added. But in the Twitterverse, he said, he’s already starting to see the same linguistic processes – such as drawing out the last syllable of a word for emphasis, as in “lmaaaoooooo” – affect the way that people use language when they tweet.
Sali Tagliamonte, a sociolinguistics professor at the University of Toronto, agreed. “What the Internet offers us is variation in the way words are spelled,” she said. “And that shows us another dimension of language and how people use language to differentiate themselves from another.”
Follow me on Twitter @LAT_aminakhan.