We Tweet Like We Talk and Other Interesting Observations: An Analysis of English Communication Modalities

Modalities of communication for human beings are gradually increasing in number with the advent of new forms of technology. Many human beings can readily transition between these different forms of communication with little or no effort, which brings about the question: How similar are these different communication modalities? To understand technology’s influence on English communication, four different corpora were analyzed and compared: Writing from Books using the 1-grams database from the Google Books project, Twitter, IRC Chat, and transcribed Talking. Multiword confusion matrices revealed that Talking has the most similarity when compared to the other modes of communication, while 1-grams were the least similar form of communication analyzed. Based on the analysis of word usage, word usage frequency distributions, and word class usage, among other things, Talking is also the most similar to Twitter and IRC Chat. This suggests that communicating using Twitter and IRC Chat evolved from Talking rather than Writing. When we communicate online, even though we are writing, we do not Tweet or Chat how we write books; we Tweet and Chat how we Speak. Nonfiction and Fiction writing were clearly differentiable from our analysis with Twitter and Chat being much more similar to Fiction than Nonfiction writing. These hypotheses were then tested using author and journalists Cory Doctorow. Mr. Doctorow’s Writing, Twitter usage, and Talking were all found to have very similar vocabulary usage patterns as the amalgamized populations, as long as the writing was Fiction. However, Mr. Doctorow’s Nonfiction writing is different from 1-grams and other collected Nonfiction writings. This data could perhaps be used to create more entertaining works of Nonfiction. INTRODUCTION The evolution of new dialects of spoken languages has often been studied and modeled with respect to the contribution of idiolects (1) and the impact of spatial pressures (2). With technology allowing ease of communication between spatially distant individuals, there is the possibility that modern dialects may form between members of social networks without similar geographical locations. This raises a somewhat semantic question: What constitutes a dialect? It is clear that graphemic communication can be very different from phonemic communication, as these have been suggested to be different dialects in English (3). However, overall this question is complicated and overlooks a simpler question: How similar or different are communication corpora from each other? In the past 25-30 years modes of online communication, such as Chat, appear as a combination of phonemic and graphemic dialects, but it is unclear how much these modes differ. Modern modalities of communication were developed on the basis of previous forms of communication, but it is unclear if the language usage is more similar to written or spoken “dialects”. In this study, I have analyzed the vocabulary from four different bodies of work: books from the Google Books project (4), chat logs from IRC, transcribed verbal communication (i.e., Talking), and over one million unique messages sent on Twitter. What I find is unexpected. Though Chat and Twitter could be considered forms of written communication, the word usage patterns are more similar to Talking. Comparisons between Fiction and Nonfiction writing show that Fiction more closely resembles Chat and Twitter. Using a case study of Author, Speaker and Twitter user Cory Doctorow I find that the way someone speaks can be as similar to Nonfiction writing as it is to Fiction writing but we incorrigibly Tweet like we speak.