A Mixed-Effects Analysis of Addressee Honorifics in Japanese Voice Actor Livestreams

Abstract

This study examines the distribution of addressee honorific (AH) and plain forms in Japanese voice actor livestreams to determine whether honorific usage can be predicted by contextual factors. Using mixed-effects modeling combined with an analysis of within-section progression, we demonstrate that AH usage is neither random nor uniform, but systematically conditioned by section type, speaker role, and the unfolding interactional structure of discourse. Across most interactional contexts, speakers predominantly use the plain form, which functions as the default speech style. AH forms occur most frequently in introduction and conclusion sections, where presenters explicitly orient to the audience. In sections where the plain form is dominant, speakers dynamically alternate between AH and plain forms to index shifts in stance, participation framework, and audience orientation. These findings provide quantitative evidence that defaultness in speech style is locally and interactionally established rather than fixed by social norms or role relations. By situating honorific usage within moment-by-moment interaction, this study supports interactional and indexical accounts of Japanese honorifics (e.g., Cook 2011; Okamoto 2010) and demonstrates the value of quantitative modeling for analyzing context-sensitive linguistic resources.