“You might think about slightly revising the title”: Identifying Hedges in Peer-tutoring Interactions

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 2160 - 2174 May 22-27, 2022 c⃝2022 Association for Computational Linguistics "You might think about slightly revising the title": Identifying Hedges in Peer-tutoring Interactions Yann Raphalen1, Chloé Clavel2, Justine Cassell1,3 1Inria Paris 2LTCI, Institut Polytechnique de Paris, Telecom-Paris 3Carnegie Mellon University yann.raphalen.pro@gmail.com, justine@cs.cmu.edu, chloe.clavel@telecom-paris.fr Abstract Hedges play an important role in the manage- ment of conversational interaction. In peer- tutoring, they are notably used by tutors in dyads (pairs of interlocutors) experiencing low rapport to tone down the impact of instructions and negative feedback. Pursuing the objective of building a tutoring agent that manages rap- port with students in order to improve learning, we used a multimodal peer-tutoring dataset to construct a computational framework for iden- tifying hedges. We compared approaches re- lying on pre-trained resources with others that integrate insights from the social science litera- ture. Our best performance involved a hybrid approach that outperforms the existing base- line while being easier to interpret. We employ a model explainability tool to explore the fea- tures that characterize hedges in peer-tutoring conversations, and we identify some novel fea- tures, and the benefits of such a hybrid model approach. 1 Introduction Rapport, most simply defined as the “... relative harmony and smoothness of relations between peo- ple . . . ” (Spencer-Oatey, 2005), has been shown to play a role in the success of activities as varied as psychotherapy (Leach, 2005) and survey interview- ing (Lune and Berg, 2017). In peer-tutoring, rap- port, as measured by the annotation of thin slices of video, has been shown to be beneficial for learning outcomes (Zhao et al., 2014; Sinha and Cassell, 2015). The level of rapport rises and falls with conversational strategies deployed by tutors and tutees at appropriate times, and as a function of the content of prior turns. These strategies include self- disclosure, referring to shared experience, and, on the part of tutors, giving instructions in an indirect manner. Some work has attempted to automatically detect these strategies in the service of intelligent tutors (Zhao et al., 2016a), but only a few strate- gies have been attempted. Other work has con- centrated on a "social reasoning module" (Romero et al., 2017) to decide which strategies should be generated in a given context, but indirectness was not among the strategies targeted. In this paper, we focus on the automatic classification of one spe- cific strategy that is particularly important for the tutoring domain, and therefore important for intel- ligent tutors: hedging, a sub-part of indirectness that "softens" what we say. This work is part of a larger research program with the long-term goal of automatically generating indirectness behaviors for a tutoring agent. Figure 1: A mock conversation displaying each type of hedged formulation. According to Brown and Levinson (1987), hedges are part of the linguistic tools that interlocu- tors use to produce politeness, by limiting the face threat to the interlocutor (basically by limiting the …

Content truncated to 500 words