editorial Register and social media Isobelle Clarke Lancaster University Social media has become an integral part of modern-day life. Nearly 57 percent of the 7.7 billion people worldwide use it (4.48 billion) with the average social media user having 8.4 social media accounts and engaging with an average of 6.6 social media platforms (Dean 2021). Each social media platform varies in terms of the medium’s characteristics and afordances (e.g., message format, privacy settings, synchronicity of message), and the situational factors (e.g., participants, goals, topics and norms). These register distinctions inevitably infuence the language used on the platforms. Despite social media hosting a variety of interactional sit- uations and therefore providing linguists with the opportunity to understand the social nature of language, social media have been underexplored from a register perspective. One of the main reasons for this is because social media posts are typically short texts. To describe patterns of register variation on any scale, linguists ofen compare the normalised frequencies of features across texts. However, the nor- malised frequencies of features in short texts are not very meaningful. Given the brevity of social media texts, the raw counts of features in these texts are infated in normalised rates to levels not seen in longer stretches of texts. For instance, a 5-word text with one adjective has adjectives occur 200 times per a thousand words. Yet even the most descriptive texts come nowhere near this rate. Likewise, short texts tend to have many features missing. Compare the sentences: (1) That man is very silly. (2) Yes, he is very silly. While there is no doubt that both sentences are functionally and situationally very similar, they do not share two features. Sentence (1) does not have the interjection yes or the third person pronoun he and sentence (2) does not have the demonstra- tive determiner that or the noun man. These features would have a normalised frequency of 0 per a thousand words in the texts where they are absent, yet they will have a normalised frequency of 200 times per a thousand words in the text where they occur. These infated disparities are not helpful in text comparisons. As a result, register studies of social media have been limited to longer texts above https://doi.org/./rs..edi Register Studies : (), pp. –. ISSN - | E‑ISSN - © John Benjamins Publishing Company