Swearing on Twitter could be sign of coming protests
Increased swearing on Twitter could be a sign of societal discontent, says a study from the Rand Corp that analyzed an archive of Tweets from 2009 with the hash tag #IranElection.
In a report posted online Jan. 17, Rand researchers detail their use of Linguistic Inquiry and Word Count, an text-analysis application that calculates the degree to which people use 80 different categories of words, in order to retrospectively analyze Iranian public opinion during 2009 mass protests.
LIWC (pronounced "luke") analysis doesn't examine the explicit subject matter of texts but rather focuses on how they say it implicitly, the Rand report notes.
"How people communicate and express themselves largely depends on the function words they use: pronouns, prepositions, articles, conjunctions and auxiliary verbs," it adds. Rand researchers looked for indicators such as positive- and negative-emotion words, swear and curse words, and singular and second-person pronouns.
They found that people's use of swear words on Twitter correlated more closely than any other indicator with events and protest on the ground, "and it did the best job of forecasting when protests would occur."
Twitter played an visible role during the 2009 Iranian protests, since the service offers anonymous handles and can be posted to without an Internet connection via SMS. The Tweets analyzed by researchers were all in English, meaning that they included messages from outside Iran and in general reflected a population segment "younger, more educated, and more urban" than the overall Iranian population, the report acknowledges.
But, the predominant use of the hash tag #IranElection strongly suggests that it encompasses the vast majority of communications about the contested re-election of Iranian President Mahmoud Ahmadinejad that led to the protests--and "consequently, that it captures the broadest account of public opinion available," the report adds.
By February 2010, profanity use on the hash tag decreased and the protest movement had petered out. In fact, further analysis shows that swearing against opposition figure Mir-Hossein Mousavi Khameneh in the winter of 2009-2010 reached levels initially directed to Ahmadinejad at the start of the protests in summer 2009.
Twitters' use of first person singular pronouns ("I," "me," "mine") and second person pronouns ("your") also correlated to periods of mass protests, the study says, although not as effectively as swear words did.
Ultimately, automated linguistic analysis of Tweets holds the potential to do real-time assessments of public opinion and forecast events, the report concludes.
- download the report, "Using Social Media to Gauge Iranian Public Opinion and Mood After the 2009 Election" (.pdf)