A paper presented at the annual conference of New Ways of Analyzing Variation at Indiana University found significant gender differences in the use on language on Twitter. The researchers examined more than 9 million tweets in the English language. By examining user names they were able to assign gender designations to most tweeters. The authors say that they can predict the gender of Twitter users with an accuracy rate of 88 percent.
The results showed that women were more likely than men to use pronouns, emoticons, abbreviations common to online discourse such as OMG (Oh My God) and LOL (Laugh Out Loud), and words expressing emotions such as love, sad, or glad.
The results showed that men were more likely to use numbers, proper nouns, and what may be called foul language.
The researchers also identified a large group which they identified as predominantly male tweeters but who tended to tweet using language more common to females. This group of men tended to have Twitter followers who were more likely to be women.
The research was conducted by David Bamman of Carnegie Mellon University, Jacob Eisenstein of the Georgia Institute of Technology, and Tyler Schnoebelen, a recent Ph.D. graduate of Stanford University.