Manual annotations of conversational feedback used in the following publication: Ildikó Pilán, Laurent Prévot, Hendrik Buschmeier and Pierre Lison. (2024). Conversational Feedback in Scripted versus Spontaneous Dialogues: A Comparative Analysis. Proceedings of the SIGDIAL 2024 Conference. September 18-20, 2024, Kyoto, Japan. (The link to the published paper will be made available shortly.)
Each Excel file consists of several sheets corresponding to different languages and corpora. The abbreviations "hi" and "fo" indicate subtitle data for hearing impared and foreign audiences respectively. For details about the list of corpora used, see the paper above.
Each sheet consists of the following columns:
- Column A: An utterance to label in bold together with a few other utterances as context
- Columns B, C, D: manual annotations for three conversational feedback classes Agree / Accept, Acknowledgement / Backchannel and Negative Feedback.
- Column E: language code and (sub)corpus of origin
- Column F: Line number of the utterance to annotate from its source file