My Memoji, my self: prosodic correlates of online performed code-switching via avatar

Abstract Collecting code-switching data has long been a challenge for sociolinguists, because eliciting naturalistic alternations between one language or variety and another is often pragmatically difficult. Data for the current study comes from a social media thread dedicated to “code-switching”, representing a new way of using online recordings for analysis of sociolinguistic phenomena. In October 2019, a Twitter user posted a Memoji avatar video, soliciting responses that illustrate how individuals “code-switch” between “regular” and “work” voices. Responding users’ decisions about the design of their Memoji images provide sociolinguistic information frequently absent from online data; namely the user’s performance of race, gender, and style. Analysis focuses on prosodic differences between the “regular” and “work” styles for 142 user responses to the original post. Regression models reveal differences between users’ “home voices” and “work voices”; these include different pitch range settings by context, more monotone realizations in the work voices, and rise-fall and fall-rise patterns that differ between contexts. These results provide important information about how the same individuals may employ intraspeaker prosodic differences to construct racialized professional and personal identities. Additionally, they represent a new point of departure for studying the phonetic correlates of racialized style-shifting.