Elamur Research⌕
☰
home / papers / 2015 / Pragmatic Rejection

Pragmatic Rejection

Paper · 2015 · Julian J. Schlöder and Raquel Fernández

Summary

The paper looks at replies in dialogue that refuse an earlier assertion even though they do not logically contradict it. The authors call these pragmatic rejections. Examples include answering "we're all mad" with "well, some of us", which is entailed by the original but implicates "not all". Standard approaches treat rejection as inconsistency with the context, so they miss such cases.

The authors make four contributions:

  • They define the class, covering three types: rejection-by-implicature, rejection-of-implicature and rejection-of-presupposition.
  • They build a small corpus of examples drawn from the AMI Meeting Corpus, Switchboard and the spoken part of the BNC.
  • They propose a formal model, with a probabilistic simulation, that derives rejection force from implicatures.
  • They run a crowdsourcing study, adapted from Potts' work on indirect answers, to gather human agree/disagree judgements.

The simulation behaves well on plain agreements and semantic disagreements and predicts rejection-by-implicature. However, people judged the pragmatic rejections as rejecting only about half the time and often chose "probably agree". The authors conclude that rejection should be modelled as graded or partial rather than binary. This matters for applications such as meeting summarisation, which must keep only content both sides accepted.

Key points

  • The working definition treats a rejection as an answer to an assertion or proposal that refuses to add its content to the common ground. Replies to polar questions are deliberately excluded.
  • Prior agreement/disagreement classifiers struggle with this phenomenon; one cited system (Germesin and Wilson 2009) reached 0% recall on disagreements despite high accuracy.
  • Crowd judgements of disagreement varied by type:
    • rejections-of-presupposition: 68%, the highest; all of these match Negative Assessment labels in AMI
    • rejections of generalising implicatures, typically marked with "but": 61%
    • or-implicatures: 58%
    • scalar implicatures: 51%
    • rejections by generalising implicature: never judged rejecting
  • Repeating part of the antecedent lowered perceived rejection. "Or three" after "Four. Yeah." was judged rejecting 86% of the time, while "Three or four days" after "Maybe three days" got 40%.
  • When annotators saw agreement, "probably agree" accounted for 78% of those judgements. The simulation gave near-identical results for or- and and-implicatures, but humans treated the two differently.

Interesting ideas

  • I1. An answer that is merely entailed by the antecedent (informationally redundant) is a signal that a Quantity implicature such as "not all" or "not necessarily X" may carry a rejection. Such an answer should not be read as automatic agreement.
  • I2. Splitting rejections into by-implicature, of-implicature and of-presupposition helps because each needs a different inference step: enriching the reply, enriching the antecedent, or computing presuppositions first.
  • I3. Acceptance can be partial: a reply may endorse one conjunct or a weaker version of a claim while refusing the rest. Committing all-or-nothing to the common ground loses this.
  • I4. Surface cues shift how strongly a reply reads as rejecting. Echoing a phrase from the antecedent pushes it toward agreement, while a contrastive "but" pushes it toward disagreement.
  • I5. The authors check a formal model against graded crowd judgements but compare only qualitative patterns, since the judgement scale and the model's probabilities do not measure the same thing.

For Mova

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

  • I1 (frames): crate prag can flag replies whose content is a strict weakening of the previous claim and compute the negated stronger alternative as a candidate implicated rejection.
  • I2 (frames): speech-act labelling can carry a rejection subtype that records which inference produced the rejection force, keeping the explanation traceable.
  • I3 (world): the text world can store per-proposition commitment status per speaker, so that "in dialogue it's fine" grounds D while leaving N contested.
  • I4 (frames): these cues can serve as explicit, inspectable weights on rejection confidence, layered on top of the logical implicature check.
  • I5 (learn): when evaluating pragmatic modules against LLM-teacher or human labels, compare ordering and contrast patterns between categories rather than forcing raw score agreement.

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

📚 Shelves 4 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works

Cite

Cite as Julian J. Schlöder and Raquel Fernández. 2015. Pragmatic Rejection. Proceedings of the 11th International Conference on Computational Semantics.

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

@inproceedings{schloder2015pragmatic,
  title = {{Pragmatic Rejection}},
  author = {Julian J. Schlöder and Raquel Fernández},
  booktitle = {Proceedings of the 11th International Conference on Computational Semantics},
  year = {2015},
  url = {https://aclanthology.org/W15-0129/}
}

References

27 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Burnard, L. (2000). Reference Guide for the British National Corpus (World Edition). Oxford University Computing Services. find
  2. Carletta, J. (2007). Unleashing the killer corpus: experiences in creating the multi-everything AMI Meeting Corpus. Language Resources and Evaluation 41(2), 181–190. ↗
  3. Clark, H. H. (1996). Using language. Cambridge University Press. find
  4. Galley, M., K. McKeown, J. Hirschberg, and E. Shriberg (2004). Identifying agreement and disagreement in conversational speech: Use of Bayesian networks to model pragmatic dependencies. In Proceedings of ACL’04. ↗ our card
  5. Germesin, S. and T. Wilson (2009). Agreement detection in multiparty conversation. In Proceedings of the 2009 international conference on Multimodal interfaces. find
  6. Ginzburg, J. (2012). The Interactive Stance. Oxford University Press. ↗
  7. Godfrey, J. J., E. C. Holliman, and J. McDaniel (1992). SWITCHBOARD: Telephone Speech Corpus for Research and Development. In Proceedings of ICASSP’92. find
  8. Goodman, N. D., V. K. Mansinghka, D. M. Roy, K. Bonawitz, and J. B. Tenenbaum (2008). Church: a language for generative models. In Proceedings of Uncertainty in Artificial Intelligence. find
  9. Grice, H. P. (1975). Logic and conversation. In Syntax and Semantics, Vol. 3, pp. 41–58. Acad. Press. ↗
  10. Grice, H. P. (1991). Studies in the Way of Words. Harvard University Press. find
  11. Hahn, S., R. Ladner, and M. Ostendorf (2006). Agreement/disagreement classification: Exploiting unlabeled data using contrast classifiers. In Proceedings of HLT-NAACL 2006. find
  12. Hillard, D., M. Ostendorf, and E. Shriberg (2003). Detection of agreement vs. disagreement in meetings: training with unlabeled data. In Proceedings of HLT-NAACL 2003. find
  13. Horn, L. R. (1989). A Natural History of Negation. University of Chicago Press. find
  14. Lascarides, A. and N. Asher (2009). Agreement, disputes and commitments in dialogue. Journal of Semantics 26(2), 109–158. find
  15. de Marneffe, M.-C., S. Grimm, and C. Potts (2009). Not a simple yes or no: Uncertainty in indirect answers. In Proceedings of the SIGDIAL 2009 Conference. our card
  16. Misra, A. and M. Walker (2013). Topic independent identification of agreement and disagreement in social media dialogue. In Proceedings of the SIGdial 2013 Conference. find
  17. Poesio, M. and D. Traum (1997). gence 13(3), 309–347. Conversational actions and discourse situations. find
  18. Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some features of preferred/dispreferred turn shapes. In Structures of Social Action. Cambridge University Press. ↗
  19. Potts, C. (2011). The indirect question-answer pair corpus. http://compprag.christopherpotts.net/iqap. html. Accessed: 2014-11-24. ↗
  20. Purver, M. (2001). SCoRE: A tool for searching the BNC. Technical Report TR-01-07, Department of Computer Science, King’s College London. find
  21. van der Sandt, R. A. (1994). Denial and negation. Unpublished manuscript, University of Nijmegen. find
  22. Schlöder, J. J. and R. Fernández (2014). The role of polarity in inferring acceptance and rejection in dialogue. In Proceedings of the SIGdial 2014 Conference. find
  23. Stalnaker, R. (1978). Assertion. In Syntax and Semantics, Vol. 9, pp. 315–332. Academic Press. find
  24. Stuhlmüller, A. (2014). Scalar Implicature. http://forestdb.org/models/scalar-implicature.html. Accessed: 2014-11-24. ↗
  25. Walker, M. A. (1996). Inferring acceptance and rejection in dialogue by default rules of inference. Language and Speech 39(2-3), 265–304. ↗
  26. Walker, M. A. (2012). Rejection by implicature. In Proceedings of the Annual Meeting of the Berkeley Linguistics Society. find
  27. Yin, J., P. Thomas, N. Narang, and C. Paris (2012). Unifying Local and Global Agreement and Disagreement Classification in Online Debates. In Proceedings of the 3rd Workshop in Computational Approaches to Subjectivity and Sentiment Analysis. Computational Intelli- find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/schloder-2015-pragmatic-rejection/