Gender attribution trouble in interaction about a gender ambiguous robot

Abstract

Gender attribution is related to the linguistic system of many languages, for instance in person reference. However, gender may also become relevant to what the participants are doing socially (action relevance). This article examines practices related to gender attribution in the context of a gender ambiguous robot. We examine how gender attribution to the robot emerges, unfolds and thus impacts the course of interaction. Our data consist of videorecorded Dutch interactions of two participants in the presence of a Pepper robot. We use Conversation Analysis as a method. Our analysis shows that gender attribution may involve interactional trouble. Sometimes, this is minimal (“or he or it”), marking uncertainty regarding the robot’s gender. But gender may surface more explicitly and even extensively in the case of gender negotiation and accounts that include gender assumptions (“in terms of figure I think it is more of a woman”). Such extended sequences are characterized by tensions: gender is constructed as an opinion versus a knowable; robot gender is deflected as irrelevant while gender relevance persists in the conversation. Overall, gender is treated as problematic and/or delicate, warranting a diversion from the ongoing activity. The recurrence of gender attribution talk in our data is striking in light of reported difficulty of capturing gender categorisation work in naturally-occurring interactions. Overall, deeply ingrained gender norms and their constitution in language and social interaction seem to surpass progressive robot design and may even create a context for the articulation of questionable gender assumptions.