Exploring emojis through corpus linguistic approaches

Abstract Businesses increasingly interact with stakeholders through various digital media channels: a process known as webcare. Although emojis are widely used in webcare, they have not been explored extensively in this context and, as most corpus software cannot process visual modes of expression, their use has not been studied at scale using a corpus linguistic methodology. In this article, we present the results of a large-scale data-driven analysis of emoji use in the US Corporate Twitter Corpus (UCTC), comprising 4.4 million tweets posted by and addressed to US companies across three industries. Through frequency and collocational analyses, we explore the different functions of emojis in the UCTC, with a focus on disambiguating their use in tweets serving different webcare goals. While our study thus offers new insights into the use of emojis in webcare communication, it also demonstrates how they can be studied by adapting well established corpus linguistic methods.