Beyond repair: How non-lexical cues affect responses to clarification requests in human-robot interaction

Abstract:

Clarification requests are interpreted in relation to their interactional context, yet little controlled work has examined how non-lexical cues shape responses to them. We report an exploratory Wizard-of-Oz study in which 18 participants discussed a moral dilemma with a Furhat robot that produced clarification requests in three conditions: baseline, filled pause, and laughter. Responses following laughter-prefaced clarification requests showed higher observed proportions of reformulation and sentential responses, whereas the filled pause condition showed the lowest proportions. These preliminary observations motivate further investigation into how non-lexical cues contribute to the pragmatic interpretation of clarification requests in human–robot dialogue.

Research areas:

• Repair
• Dialogue modelling
• Laughter
• DivCon

Year:

2026

Type of Publication:

In Proceedings

Book title:

Proceedings of the 30th Workshop on the Semantics and Pragmatics of Dialogue - Poster Abstracts