Annotation of Communicative Functions of Short Feedback Tokens in Switchboard

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1849–1859 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1849 Annotation of Communicative Functions of Short Feedback Tokens in Switchboard Carol Figueroa, Adaeze Adigwe, Magalie Ochs, Gabriel Skantze Furhat Robotics, ReadSpeaker, Aix-Marseille University, KTH Royal Institute of Technology carol@furhatrobotics.com, adaeze.adigwe@readspeaker.com, magalie.ochs@lis-lab.fr, skantze@kth.se Abstract There has been a lot of work on predicting the timing of feedback in conversational systems. However, there has been less focus on predicting the prosody and lexical form of feedback given their communicative function. Therefore, in this paper we present our preliminary annotations of the communicative functions of 1627 short feedback tokens from the Switchboard corpus and an analysis of their lexical realizations and prosodic characteristics. Since there is no standard scheme for annotating the communicative function of feedback we propose our own annotation scheme. Although our work is ongoing, our preliminary analysis revealed lexical tokens such as yeah are ambiguous and therefore lexical forms alone are not indicative of the function. Both the lexical form and prosodic characteristics need to be taken into account in order to predict the communicative function. We also found that feedback functions have distinguishable prosodic characteristics in terms of duration, mean pitch, pitch slope, and pitch range. Keywords: feedback, prosody, function 1. Introduction According to Clark (1996), conversations are a joint activity. When one person is speaking the listener may produce vocalizations such as mhmm, yeah, wow which help establish common ground. Clark (1996) defines common ground as “the sum of mutual, common, joint knowledge, beliefs, and suppositions”. These vocal- izations serve as linguistic feedback to “give informa- tion about the basic communicative functions, i.e., con- tinued contact, perception, understanding, and emo- tional/attitudinal reaction, in a sufficiently unobtrusive way to allow communication to serve as an instrument for pursuing various human activities” (Allwood et al., 1992). Short vocalizations such as mhmm, yeah, wow have been described in different terms. Yngve (1970) re- ferred to these short messages, either vocal or gestural signals, as backchannels. The term was used to dis- tinguish the main channel (where the current speaker is sending messages) from the backchannel (where the interlocutor is sending unobtrusive messages). Schegloff (1982) described these vocalizations as con- tinuers, which function as a way of telling the inter- locutor that they would like them to continue speak- ing and are passing the opportunity to interrupt the ongoing utterance. Following the work of Schegloff (1982), Goodwin (1986) further categorizes these vo- calizations into continuers and assessments. Good- win (1986) distinguished continuers and assessments by their position, whether they occur between or within talking units of the interlocutor, as well as their func- tion. Unlike continuers, assessments are a response to something specific that the interlocutor just said. The terms generic and specific listener responses, in- troduced by Bavelas et al. (2000), describe how these listening responses contribute to the narration of a nar- rator. These listener responses have the same function as continuers …

Content truncated to 500 words