Vol. 28:2 (2023) ► pp.144–171
Annotating dialogue acts in speech data
Problematic issues and basic dialogue act categories
The aims of this paper are to detect the most problematic issues related to dialogue act annotation in speech corpora and to define basic categories of dialogue acts. I critically examine and test generic schemes that represent different lines of dialogue act annotation: AMI, DART, ISO 24617–2 and SWBD-DAMSL. It is found that the most problematic issues regarding dialogue act annotation are related to the distinction between the semantic and pragmatic meanings of utterances, the annotation of metadiscourse, and the adequacy and informativeness of the tagset. The identified basic dialogue act categories are information providing, information seeking, actions, social acts and metadiscourse. The findings help improve dialogue act annotation.
Article outline
- 1.Introduction
- 2.Dialogue act annotation schemes
- 3.Methodology
- 3.1Selecting dialogue act annotation schemes
- 3.2Test data
- 3.3Annotation process
- 3.4Analytical procedure
- 4.Dialogue act annotation
- 4.1Applicability to a new language
- 4.2Utterance meaning
- 4.3Ambiguity
- 4.3.1Basic unit
- 4.3.2Tags
- 4.4Adequacy
- 4.5Informativeness
- 5.Dialogue act categories
- 5.1Information-providing acts
- 5.2Information-seeking acts
- 5.3Action acts
- 5.4Social acts
- 5.5Metadiscourse acts
- 6.Conclusions
-
References
https://doi.org/10.1075/ijcl.20165.ver