Mapping Lexical Innovation on American Social Media

In this paper, we introduce a method for mapping lexical innovation, which we then use to track the origin and spread of new words on American Twitter, based on a multi-billion-word corpus of Tweets collected between 2013 and 2014. We first extract fifty-four emerging words from the corpus by searching for words that are very uncommon at the end of 2013 but whose use rises dramatically over the course of 2014. We then map the origin and spread of each of these words. Based on these results, we identify five main regional patterns of lexical innovation on American Twitter, primarily associated with the West Coast, the Northeast, the Mid-Atlantic, the Deep South, and the Gulf Coast. We conclude by proposing explanations for these results and by discussing their significance to theories of language variation and change, including both the actuation and diffusion of lexical innovations.