“Write another one more emotional”: Emoji as a tenor resource in chatbot responses to requests for linguistic services

This paper investigates how emoji function as tenor resources in chatbot-generated responses to user requests for linguistic services, such as composing emails or social media posts. Drawing on Systemic Functional Linguistics, we analyse emoji as semiotic resources that modulate interpersonal meaning using the discourse semantic system of emoji-text convergence and the tenor system of tuning . Undertaking qualitative discourse analysis of 8,397 emoji-containing conversations from the LMSYS-CHAT-1 M dataset of naturally-occurring human-chatbot interactions, we explore how emoji contribute to interpersonal alignment with user-specified genres, styles, and audiences. Rather than treating emoji as isolated symbols, we demonstrate how their interpersonal function emerges through their integration with the co-text and genre expectations. Our findings show that emoji are frequently used to [warm] spirit , [lower] stakes , and [collectivise] scope , with these tuning configurations realised through emoji-text convergence relations, such as interpersonal resonance and textual synchronicity . This study contributes to emerging research on artificial communication, suggesting how emoji help simulate human-like interpersonal positioning in LLM-mediated discourse.