“Hold on, I’ll connect you to a human agent”: Recipient design, repair, and their impact on progressivity in human-chatbot conversations.

Abstract

This study analyzes recipient design in handover conversations involving customers, customer service chatbots, and human agents, and examines how repair strategies initiated by chatbots and human agents impact conversational progressivity. These text-based conversations were analyzed using Conversation Analysis. Findings show customers formulate their request and subsequent messages concisely when communicating with a chatbot (pre-handover), but use details, full sentences, and politeness markers when communicating with a human agent (post-handover). Moreover, both chatbots and human agents frequently initiate repairs to address troubles of understanding promptly, but their approaches differ: chatbots generalize inquiries broadly, often resulting in multiple sequences of repair, while human agents use follow-up questions to specify and guide responses, ensuring progressivity.