EXPRESS: The Humanizing Power of Accents in Voicebots: When Accent-Based Humanization Increases User Engagement

Given that consumers often exhibit resistance toward algorithmic systems, firms face challenges in encouraging engagement with voicebots. In this paper, we find that users’ engagement with voicebots may be enhanced by endowing these tools with non-standard accents that deviate from “standard” pronunciation. Across a range of accents, languages, genders, and contexts, our data suggest that voicebots communicating with non-standard accents make the concept of human diversity salient in the users’ minds. As a result, voicebots communicating with non-standard accents are more readily anthropomorphized. This effect translates into greater preference for the non-standard accented voicebot in customer service settings, stronger support for startups leveraging accented vs. unaccented voicebots, higher willingness to pay for monthly subscriptions, and a greater likelihood of opting in to receive future updates and communications. Consistent with our theory, however, these effects do not hold when the anthropomorphism effect is either overridden by other cues or is undesirable for the consumer. Together, this work highlights the benefits of accented speech in improving human-AI interactions. We also hope that it sparks further investigation into factors that make technology more readily accepted by consumers.