Fakespeak in 280 characters: using a corpus-based approach to study the language of disinformation on Twitter

This chapter explores a corpus of fake news spread on Twitter in 2016 and 2017 by the Russian Internet Research Agency (IRA) also known as the Russian troll factory. The corpus of fake news is compared with verified news tweets produced by the international news agency Associated Press (AP) and more sensationalist news tweets disseminated by the British middle-range tabloid the Daily Mail. The comparison is performed employing the techniques of a corpus-based approach to discourse, specifically the tool of keywords, which has been successfully used in past research to identify distinctive lexico-grammatical features of various genres or registers (e.g. Xiao and McEnery 2005; Friginal 2009; Breeze 2019). Research has shown that the difference between, for example, telling lies vs telling truth sits not so much in the content but precisely in the style or the how of telling (Galasiński 2000). Identifying distinctive lexico-grammatical features of fake news can therefore shed light on the particularities and peculiarities of its style allowing researchers to better understand how disinformation is linguistically constructed and how to better recognise misleading contents. Therefore, this chapter pursues two major goals; first, it showcases how a keyword analysis can be effectively used to identify distinctive language features of disinformation online; secondly, it endeavours to raise awareness of the particularities and peculiarities of the language of fake news that might help foster critical (social) media literacy