From a linguistic perspective, the rapid growth of online social media, and Twitter in particular, offers a new body of data with several tantalizing characteristics that set it apart from the forms of language previously studied.

• It is a multi-person interchange, a form of discourse that has received almost no linguistic analysis, compared with monolog and dialog.
• Utterances are of limited length, placing a premium on the development of linguistic conventions that participants can quickly recognize.
• Utterances are interactive (like speech) but persist (unlike speech).
• The medium includes conventions (for example, retweet tags, usernames, and hashtags) that facilitate linking multiple utterances together.
• Utterances are stamped with the time, and often with the location, at which they originate.
• The collection of both utterances and metadata is straightforward, reducing the cost of data acquisition.

Zappavigna’s book is one of the first extended academic studies of this kind of language. She follows Michael Halliday’s systemic functional linguistics (SFL) framework, which emphasizes the options open to the speaker in each rhetorical move, and how they are influenced by the speaker’s experiences and social relations. Her main dataset, HERMES, is a collection of 7 million Tweets collected using Twitter’s application programming interface (API). Because Twitter randomly samples the stream of all tweets, it is not possible to reconstruct individual conversations from the data, but the data is sufficient to demonstrate a number of mechanisms that Twitter users exploit to create and maintain a sense of affiliation across the community.

Chapter 1 introduces the basic nature of computer-mediated conversation (CMC) as “searchable speech,” and describes the linguistic approach that Zappavigna applies to the data.

Chapter 2 characterizes the distinctive nature of CMC, outlines the basic mechanisms used by its participants, and describes how the HERMES corpus was assembled. The interchanges studied are variously described as CMC and microblogging, but specifically are tweets, messages exchanged via Twitter. It is natural to extrapolate some of Zappavigna’s results to other forms of CMC, such as Facebook comments and mobile phone short message service (SMS) exchanges, but the reader should keep in mind that the analysis is strictly restricted to Twitter.

Chapter 3 goes into more detail on the language used in microblogging and its relation to more familiar forms of conversation. For readers without experience using Twitter, the discussion is of interest not only linguistically, but also as an introduction to the medium itself.

The main body of the book analyzes the corpus from different linguistic perspectives, including the language of evaluation (chapter 4), the function of hashtags in organizing the discourse (chapter 5), the generation and persistence of memes (chapter 6), the historical origins and cultural sensitivities associated with slang (chapter 7), and the manifestation of humor, particularly in the use of the #fail hashtag (chapter 8).

Chapter 9 analyzes a more focused corpus collected during and after the 2008 US presidential campaign, the Obama win corpus (OWC), to understand how political discourse manifests itself in CMC.

Chapter 10 briefly summarizes the discussion and points the way toward future research.

Throughout its history, linguistics has been constrained to the same forms of data that have existed for thousands of years. CMC introduces radically new forms of exchange that are evolving before our eyes, forms that test the generalizability of our linguistic tools. This volume is a solid contribution to that agenda, yielding new insights into both the shape of linguistic analysis and the underlying nature of computer-mediated social discourse.