Elamur Research⌕
☰
home / papers / 2022 / Conversational Analysis of Daily Dialog Data using Polite Emotional Dialogue Acts

Conversational Analysis of Daily Dialog Data using Polite Emotional Dialogue Acts

Paper · 2022 · Chandrakant Bothe and Stefan Wermter

Summary

The paper examines how politeness relates to two other conversational cues, emotion and dialogue acts, in everyday English dialogue. The authors start from DailyDialog, a multi-turn corpus that already has human labels for six emotions plus "no emotion" and for four dialogue acts (Inform, Question, Directive, Commissive). They add an automatic politeness layer to every utterance using an existing regressor from Bao et al. (2021). That regressor is a BERT-based model pre-trained on Reddit with masked language modelling and then fine-tuned on two politeness datasets. It outputs a real-valued degree of politeness (DoP) from 1 to 5, with about 3 meaning neutral. The authors then plot DoP histograms for each emotion and each dialogue act class and inspect utterances at the extremes and in the middle of the ranking. Anger and Disgust lean toward neutral or impolite, while Happiness and Sadness lean polite. Less expectedly, Inform and Commissive utterances are more often polite than Questions and Directives. The augmented corpus is released, and the authors present it as a basis for socially aware dialogue systems and human-robot interaction.

Key points

  • DailyDialog contains 102,979 utterances. "No emotion" covers 83.78% of them, and Happiness (12,885 utterances) makes up 74.02% of the emotion-bearing ones. Fear is the rarest class (174 utterances).
  • By dialogue act, the corpus has 47,089 Inform, 29,693 Question, 17,356 Directive and 8,841 Commissive utterances.
  • The politeness analyzer (Politenessr, from Bao et al. 2021) was fine-tuned on the Danescu-Niculescu-Mizil et al. (2013) and Wang & Jurgens (2018) data. Its scores agree with human judgments at a mean Pearson correlation of 0.66.
  • The extremes of the ranking are telling:
    • The highest scores (about 4.63) go to Happiness/Inform "thank you"-type turns.
    • The lowest score (1.37) goes to an Anger-labelled Directive insulting the listener.
  • Fear and Surprise are mostly neutral in politeness. The politeness-annotated data is published on GitHub as bothe/politeEDAs.

Interesting ideas

  • I1. Instead of building a new corpus, the authors add one automatic layer (politeness) to a corpus that already has gold layers, then cross-tabulate the new layer against the gold ones to find associations cheaply.
  • I2. Politeness is scored as a graded value with a neutral midpoint rather than a polite/impolite switch, which keeps a middle zone where most ordinary utterances belong.
  • I3. The automatic annotator is sanity-checked by sorting utterances by score and reading samples from both extremes and several middle bands, with near-duplicates removed. This shows whether the scale behaves as intended across its whole range.
  • I4. The finding that Directives and Questions are less polite than Inform and Commissive turns suggests that the form of the dialogue act itself shifts expected politeness, independent of emotion.
  • I5. Emotion state and dialogue act jointly predict which politeness level is natural (thanks with happiness, blunt commands with anger). This can guide what a system says next in a dialogue flow.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): run prag's speech-act and politeness judgments over a corpus with gold dialogue-act or emotion labels and compare distributions, as a cheap check that prag's pragmatic labels pattern sensibly.
  • I2 (frames): represent politeness or face-threat in a speech-act frame as a scaled attribute with explicit neutral, so that downstream implicature rules can reason about degrees.
  • I3 (learn): when inducing any scalar score from silver data, routinely dump deduplicated samples from top, bottom and middle bands as a human-readable evaluation step next to negative controls.
  • I4 (frames): use dialogue-act-conditioned politeness expectations as a prior, so that an angry-context utterance with unusually polite surface form can be flagged as a candidate for irony or sarcasm.
  • I5 (back): when generating replies from the meaning graph, pick politeness markers ("please", "thank you", softening modals) according to the speaker's emotion state and the act type being realized.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Universität HamburgHamburg · Germany79 works · 5,422 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 4Research subjects and languages the work is about.English · 2,263 worksSentiment & emotion · 972 worksDialogue systems · 624 worksSpeech act theory · 229 works

Cite

Cite as Chandrakant Bothe and Stefan Wermter. 2022. Conversational Analysis of Daily Dialog Data using Polite Emotional Dialogue Acts. Proceedings of the Thirteenth Language Resources and Evaluation Conference.

The paper itself, not this page. BibTeX from the library's record of the work. arXiv 2205.02921

@inproceedings{bothe2022conversational,
  title = {{Conversational Analysis of Daily Dialog Data using Polite Emotional Dialogue Acts}},
  author = {Chandrakant Bothe and Stefan Wermter},
  booktitle = {Proceedings of the Thirteenth Language Resources and Evaluation Conference},
  year = {2022},
  doi = {10.63317/222rhbh9g7yx},
  eprint = {2205.02921},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.63317/222rhbh9g7yx}
}

References

23 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bothe, C., Garcia, F., Cruz-Maya, A., Pandey, A. K., and Wermter, S. (2018a). Towards Dialogue-based Navigation with Multivariate Adaptation driven by Intention and Politeness for Social Robots. In Proceedings of the International Conference on Social Robotics, pages 230–240. Springer. find
  2. Bothe, C., Magg, S., Weber, C., and Wermter, S. (2018b). Conversational Analysis using Utterance-level Attention-based Bidirectional Recurrent Neural Networks. In Proceedings of the International Conference INTERSPEECH 2018, pages 996–1000. International Speech Communication Association (ISCA). find
  3. Bothe, C., Magg, S., Weber, C., and Wermter, S. (2018c). Discourse-Wizard: Discovering Deep Discourse Structure in your Conversation with RNNs. Computation and Language. ↗
  4. Bothe, C., Weber, C., Magg, S., and Wermter, S. (2020). EDA: Enriching Emotional Dialogue Acts using an Ensemble of Neural Annotators. In Proceedings of The 12th Language Resources and Evaluation Conference, LREC 2020, pages 620–627. European Language Resources Association (ELRA). ↗
  5. Bothe, C. (2020). Conversational Language Learning for Human-Robot Interaction. Ph.D. thesis, Staats-und Universitätsbibliothek Hamburg Carl von Ossietzky, University of Hamburg. find
  6. Brown, P. and Levinson, S. C. (1987). Politeness: Some Universals in Language Usage, volume 4. Cambridge University Press. find
  7. Castro-González, Á., Castillo, J. C., Alonso-Martı́n, F., Olortegui-Ortega, O. V., González-Pacheco, V., Malfaz, M., and Salichs, M. A. (2016). The Effects of an Impolite vs. a Polite Robot Playing Rock-Paper-Scissors. In Proceedings of the International Conference on Social Robotics, pages 306– 316. Springer. find
  8. Culpeper, J. and Tantucci, V. (2021). The Principle of (Im)politeness Reciprocity. Journal of Pragmatics, 175:146–164. ↗
  9. Culpeper, J. (2011). “It’s not what you said, it’s how you said it!”: Prosody and impoliteness. In Discursive Approaches to Politeness, pages 57–84. De Gruyter Mouton. find
  10. Danescu-Niculescu-Mizil, C., Sudhof, M., Jurafsky, D., Leskovec, J., and Potts, C. (2013). A Computational Approach to Politeness with Application to Social Factors. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 250–259. ACL. ↗
  11. Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1, pages 4171–4186. ACL.
  12. Holmes, J. and Stubbe, M. (2015). Power and Politeness in the Workplace: A Sociolinguistic Analysis of Talk at Work. Routledge. ↗
  13. Kasper, G. (1990). Linguistic Politeness: Current Research Issues. Journal of Pragmatics, 14(2):193– 218. find
  14. Kumar, S., Itzhak, E., Olatunji, S., Sarne-Fleischmann, V., Tractinsky, N., Nimrod, G., and Edan, Y. (2021). Exploratory evaluation of politeness in human-robot interaction. arXiv:2103.08441. ↗
  15. Lakomkin, E., Bothe, C., and Wermter, S. (2017). GradAscent at EmoInt-2017: Character and Word Level Recurrent Neural Network Models for Tweet Emotion Intensity Detection. In Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis at the Conference EMNLP 2017, pages 169–174. ACL. find
  16. Langlotz, A. and Locher, M. A. (2017). (Im)politeness and Emotion. In The Palgrave Handbook of Linguistic (Im)politeness, pages 287–322, London. Palgrave Macmillan UK. ↗
  17. Li, Y., Su, H., Shen, X., Li, W., Cao, Z., and Niu, S. (2017). DailyDialog: A manually labelled multi-turn dialogue dataset. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 986– 995, Taipei, Taiwan. Asian Federation of Natural Language Processing. ↗
  18. Renner, B. (2020). (In) directness as an (im) politeness strategy in the contact between german and Brazilian portuguese as additional languages. Ph.D. thesis, Universidade de Brası́lia, Brası́lia. find
  19. Russell, J. A. and Barrett, L. F. (1999). Core Affect, Prototypical Emotional Episodes, and Other Things Called Emotion: Dissecting the Elephant. Journal of Personality and Social Psychology, 76(5):805. ↗
  20. Salem, M., Ziadee, M., and Sakr, M. (2014). Marhaba, how may I help you?: Effects of Politeness and Culture on Robot Acceptance and Anthropomorphization. In Proceedings of the International Conference on Human-robot Interaction, pages 74–81. find
  21. Srinivasan, V. and Takayama, L. (2016). Help Me Please: Robot Politeness Strategies for Soliciting Help From People. In Proceedings of the Conference on Human Factors in Computing Systems, pages 4945–4955. ACM. find
  22. Steinfeld, A., Fong, T., Kaber, D., Lewis, M., Scholtz, J., Schultz, A., and Goodrich, M. (2006). Common Metrics for Human-Robot Interaction. In Proceedings of the 1st ACM SIGCHI/SIGART Conference on Human-robot Interaction, pages 33–40. ACM. find
  23. Wang, Z. and Jurgens, D. (2018). It’s going to be okay: Measuring Access to Support in Online Communities. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 33–45, Brussels, Belgium. ACL. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bothe-2022-conversational-analysis-daily-dialog-data/