Elamur Research⌕
☰
home / papers / 2019 / Guess Who's Coming (and Who's Going): Bringing Perspective to the Rational Speech Acts Framework

Guess Who's Coming (and Who's Going): Bringing Perspective to the Rational Speech Acts Framework

Paper · 2019 · Carolyn Jane Anderson and Brian W. Dillon

Summary

The paper asks how a listener works out whose point of view a perspectival word is anchored to, using English come and go as the test case. Come signals that some perspective holder (speaker, listener or attitude subject) is at the destination, so the hearer must infer both what happened and from whose viewpoint it is described. The authors extend the Rational Speech Acts (RSA) framework so that the pragmatic listener infers a joint distribution over worlds and perspectives given an utterance. In their lexicon, come carries a motion meaning plus an anchoring condition, while go carries motion only. A cost applies to non-speaker perspectives. The model yields three predictions:

  • go's "away from the anchor" reading can arise as a Gricean inference rather than from its lexical meaning;
  • the non-speaker cost parameter shifts which perspective holder is inferred;
  • weighing several perspectives at once changes the marginal distribution over worlds.

The work is theoretical and sets out experiments to test these predictions. It matters because it turns perspective resolution into explicit, testable probabilistic inference.

Key points

  • Come has two parts: a motion implication (x moves to d) and an anchoring implication (a salient perspective holder is located at d). Go has only the motion implication.
  • The model has three levels: literal listener, literal speaker (softmax over log L0 minus cost) and pragmatic listener. The pragmatic listener reasons over (world, perspective) pairs, with a uniform prior over perspectives.
  • Cost is charged only for non-speaker perspectives, so a non-speaker reading wins only when the semantics rule out the speaker. For "I am coming to Northampton", the listener (Lydia) perspective gets probability 1.0. For "You are coming…", the speaker perspective gets 1.0.
  • For "Thera is coming to Northampton" with speaker cost 0.5, the marginal over worlds favours the world where both speaker and listener are at the destination. This holds whenever the listener has a non-zero chance of being the perspective holder.
  • The planned experiment is a picture-choice task. It contrasts coming with a non-perspectival control (driving) to separate the effect of perspective from a general preference for co-located interlocutors.

Interesting ideas

  • I1. Treat the perspective holder as a latent variable that is inferred jointly with the world, rather than fixing it to the speaker by default.
  • I2. Derive the "away from anchor" sense of go pragmatically, from the speaker not choosing come, so the lexicon stays minimal and the inference stays traceable.
  • I3. Use lexical semantics to prune impossible readings, since a mover cannot also be the anchor at the destination.
  • I4. Marginalise over all live perspectives, so the preferred world is the one that satisfies every plausible reading.
  • I5. Build each test item with a non-perspectival twin (coming vs driving) to separate the target effect from background plausibility.

For Mova

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

  • I1 (frames): Deictic items like come/go, here, tomorrow and my right could carry an explicit, scored set of candidate anchors (speaker, addressee, attitude subject) instead of a single hard-coded one.
  • I2 (frames): The implicature module could record such contrast-based inferences ("come was available but not used") as defeasible, explainable conclusions rather than lexical facts.
  • I3 (world): When placing entities in the text world, come can assert the anchor's location only for candidates not already moving in that event, which narrows locations cleanly.
  • I4 (answer): For questions like "where is the narrator?", answers could be ranked by support summed across perspective readings rather than by one chosen reading.
  • I5 (learn): Paired negative-control items like this would show whether a deixis-handling improvement truly comes from the perspective machinery.

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 3The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksPhilosophy · 211 works

🧭 Topics 1Research subjects and languages the work is about.English · 1,373 works

Cite

Cite as Carolyn Jane Anderson and Brian W. Dillon. 2019. Guess Who's Coming (and Who's Going): Bringing Perspective to the Rational Speech Acts Framework. Proceedings of the Society for Computation in Linguistics (SCiL) 2019.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.7275/9bn3-8x38

@inproceedings{anderson2019guess,
  title = {{Guess Who's Coming (and Who's Going): Bringing Perspective to the Rational Speech Acts Framework}},
  author = {Carolyn Jane Anderson and Brian W. Dillon},
  booktitle = {Proceedings of the Society for Computation in Linguistics (SCiL) 2019},
  year = {2019},
  doi = {10.7275/9bn3-8x38},
  url = {https://doi.org/10.7275/9bn3-8x38}
}

References

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

  1. Jefferson Barlew. 2017. The semantics and pragmatics of perspectival expressions in English and Bulu: The case of deictic motion verbs. Dissertation, Ohio State University. find
  2. Leon Bergen, Noah D. Goodman, and Roger Levy. 2012. That’s what she (could have) said: how alternative utterances affect language use. Proceedings of the Annual Conference of the Cognitive Science Society, 34. find
  3. Leon Bergen, Roger Levy, and Noah D. Goodman. 2016. Pragmatic reasoning through semantic inference. Semantics and Pragmatics, 9. ↗
  4. Edit Doron. 1991. Point of view as a factor of content. Proceedings of tSemantics and Linguistic Theory Conference (SALT). ↗
  5. Charles J. Fillmore. 1975. Deixis. find
  6. Michael C. Frank and Noah D. Goodman. 2012. Predicting pragmatic reasoning in language games. Science, 336:998–998. ↗
  7. Virgina Gathercole. 1987. Towards a universal for deictic verbs of motion. Kansas Working Papers in Linguistics, 3. find
  8. Noah D. Goodman and Andreas Stuhlmüller. 2013. Knowledge and implicature: Modeling language understanding as social cognition. Topics in Cognitive Science, 5:173–184. ↗
  9. Noah D. Goodman and Andreas Stuhlmüller. 2014. The Design and Implementation of Probabilistic Programming Languages. http://dippl.org. ↗
  10. Jesse Harris. 2012. Processing Perspectives. Dissertation, University of Massachusetts, Amherst. find
  11. C. T. James Huang and C. S. Luther Liu. 2001. Logophoricity, attitudes, and ziji at the interface. Long-distance reflexives, 33:141–195. find
  12. Elsi Kaiser. 2015. Perspective-shifting and free indirect discourse: Experimental investigations. ↗
  13. Justine T. Kao, Jean Y. Wu, Leon Bergen, and Noah D. Goodman. 2014. Nonliteral understanding of number words. Proceedings of the National Academy of Sciences, 111. ↗
  14. David Lewis. 1979. Score-keeping in a language game. Semantics from a Different Point of View. find
  15. Katherine A. Loveland. 1984. Learning about points of view: spatial perspective and the acquisition of ‘i/you’. Journal of Child Language, 11:535–556. ↗
  16. Tsuneko Nakazawa. 2007. A typology of the ground of deictic motion verbs as path-conflating verbs: the speaker, addressee, and beyond. Poznan Studies in Contemporary Linguistics, 43. find
  17. Tsuneko Nakazawa. 2009. A typology of the ground of deictic motion verbs as path-conflating verbs: the entailment of arrival and the deictic center. Poznan Studies in Contemporary Linguistics, 45. find
  18. David Y. Oshima. 2006. Motion deixis, indexicality, and presupposition. Proceedings of Semantics and Linguistic Theory (SALT), 16. ↗
  19. Yangsook Park. 2017. Attitudes dese and logophoricity. Dissertation, University of Massachusetts, Amherst. find
  20. Christopher Potts, Daniel Lassiter, Roger Levy, and Michael C. Frank. 2016. Embedded implicatures as pragmatic inferences under compositional lexical uncertainty. Journal of Semantics, 33. ↗
  21. Ciyang Qing, Noah Goodman, and Daniel Lassiter. 2016. A rational speech-act model of projective content. Proceedings of the Cognitive Science Society, 38. find
  22. Craige Roberts. 2015. Indexicality: de se semantics and pragmatics. find
  23. Nathaniel J. Smith, Noah D. Goodman, and Michael C. Frank. 2013. Learning and using language via recursive pragmatic reasoning about other agents. Advances in Neural Information Processing Systems (NIPS). find
  24. Margaret Speas. 2000. Person and point of view in navajo: direct discourse complements. Proceedings of West Coast Conference on Formal Linguistics. find
  25. Robert C. Stalnaker. 2014. Context. Oxford University Press. find
  26. Yasutada Sudo. 2018. Come vs. go and perspectival shift. Ms. find
  27. Stephen Wechsler. 2010. What ‘you’ and ‘i’ mean to each other: person indexicals, self-ascription, and theory of mind. Language, 86:332–365. ↗
  28. David P. Wilkins and Deborah Hill. 1995. When ‘go’ means ‘come’: Questioning the basicness of basic motion verbs. Cognitive Linguistics, 6:209–259. ↗
  29. Erica J. Yoon, Michael Henry Tessler, Noah D. Goodman, and Michael C. Frank. 2016. Talking with tact: Polite language as a balance between kindness and informativity. Proceedings of the Cognitive Science Society. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/anderson-2019-guess-who-s-coming-who/