Elamur Research⌕
☰
home / papers / 2026 / Do I know what I want to say? Modeling meaning uncertainty in RSA

Do I know what I want to say? Modeling meaning uncertainty in RSA

Summary

Work in the Rational Speech Act (RSA) framework usually assumes that the first-level speaker knows exactly which meaning they intend to convey. The authors argue that this often fails in experiments where participants describe images, because how people perceive or interpret the stimulus can leave the intended meaning fuzzy. They propose um-RSA, an extension that converts observed meanings into a distribution over intended meanings using a weight matrix W O→I. The framework can encode any hypothesis about meaning uncertainty.

As a case study, they model two sources of uncertainty:

  • Counting-Uncertainty: approximate number perception.
  • Discounting-Uncertainty: downweighting low-probability, high-impact events.

The models predict that discounting should raise the use of more confident expressions (bare not, probably) relative to might, a contrast that standard RSA does not produce. Two within-subject human experiments, comparing gumball machines with election forecasts, matched this prediction. The effect remained when images were replaced with text, which points to discounting rather than counting as the driver. The work shows that modeling speaker-side meaning uncertainty explains production patterns that plain RSA misses.

Key points

  • um-RSA inserts a W O→I matrix that maps observed meanings onto intended meanings. It can be applied before pragmatic reasoning (committed speaker model) or after it (uncommitted speaker model). The two variants gave qualitatively similar predictions in this setup.
  • The modeling covers a pragmatic speaker S1 reasoning about a literal listener L0. Predictions are made for three utterances: bare not, might and probably.
  • Experiment 1 (images): participants chose probably and bare not more often for election predictions than for gumball machines.
  • Experiment 2 (text descriptions): removing the need for numerical approximation left the same qualitative pattern, supporting Discounting-Uncertainty as the explanation.
  • Resources: code, data and analyses are released at github.com/anziw/uncertainty-exp.
  • Future directions:
    • testing committed vs. uncommitted models by manipulating utterance cost;
    • extending the approach to listeners;
    • learning W O→I weights.

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

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.scil-main.29

@inproceedings{wang2026do,
  title = {{Do I know what I want to say? Modeling meaning uncertainty in RSA}},
  author = {Wang, Anzi and Anderson, Carolyn Jane and Prasad, Grusha},
  booktitle = {Proceedings of the Society for Computation in Linguistics 2026},
  year = {2026},
  doi = {10.18653/v1/2026.scil-main.29},
  url = {https://doi.org/10.18653/v1/2026.scil-main.29}
}

References

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

  1. Judith Degen. 2023. The rational speech act framework. Annual Review of Linguistics, 9(Volume 9, 2023):519–540. find
  2. Judith Degen, Michael Henry Tessler, and Noah D Goodman. 2015. Wonky worlds: Listeners revise world knowledge when utterances are odd. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 37. find
  3. Michael C. Frank and Noah D. Goodman. 2012. Predicting pragmatic reasoning in language games. Science, 336(6084):998–998. find our card
  4. Camilla Gilmore, Nina Attridge, and Matthew Inglis. 2011. Measuring the approximate number system. Quarterly Journal of Experimental Psychology, 64(11):2099–2109. find
  5. Noah D. Goodman and Andreas Stuhlmüller. 2013. Knowledge and implicature: Modeling language understanding as social cognition. Topics in Cognitive Science, 5(1):173–184. find
  6. Sebastian Schuster and Judith Degen. 2020. I know what you’re probably going to say: Listener adaptation to variable use of uncertainty expressions. Cognition, 203:104285. find
  7. Joakim Sundh. 2024. Human behavior in the context of low-probability high-impact events. Humanities and Social Sciences Communications, 11(1). find
  8. Thomas S Wallsten, Samuel Fillenbaum, and James A Cox. 1986. Base rate effects on the interpretations of probability and frequency expressions. Journal of Memory and Language, 25(5):571–587. find
  9. Ilker Yildirim, Judith Degen, Michael K. Tanenhaus, and T. Florian Jaeger. 2016. Talker-specificity and adaptation in quantifier interpretation. Journal of Memory and Language, 87:128–143. find
  10. Jeffrey Zehr and Florian Schwarz. 2018. Penncontroller for ibex. https://github.com/PennController/ pcibex. Accessed: 2025-04-26. ↗
  11. Justin Halberda, Michèle MM Mazzocco, and Lisa Feigenson. 2008. Individual differences in non-verbal number acuity correlate with maths achievement. Nature, 455(7213):665–668. find
  12. Michele Herbstritt and Michael Franke. 2019. Complex probability expressions and higher-order uncertainty: Compositional semantics, probabilistic pragmatics and experimental data. Cognition, 186:50–71. find
  13. Daniel Kahneman and Amos Tversky. 1979. Prospect theory: An analysis of decision under risk. Econometrica, 47(2):263–291. find
  14. Justine Kao, Leon Bergen, and Noah Goodman. 2014a. Formalizing the pragmatics of metaphor understanding. Proceedings of the Annual Meeting of the Cognitive Science Society, 36. find
  15. Justine T. Kao and Noah D. Goodman. 2015. Let’s talk (ironically) about the weather: Modeling verbal irony. Proceedings of the Annual Meeting of the Cognitive Science Society, 37. find
  16. Justine T. Kao, Jean Y. Wu, Leon Bergen, and Noah D. Goodman. 2014b. Nonliteral understanding of number words. Proceedings of the National Academy of Sciences, 111(33):12002–12007. find our card
  17. Mel W Khaw, Luminita Stevens, and Michael Woodford. 2021. Individual differences in the perception of probability. PLoS computational biology, 17(4):e1008871. find
  18. Daniel Lassiter and Noah D. Goodman. 2013. Context, scale structure, and statistics in the interpretation of positive-form adjectives. In Proceedings of Semantics and Linguistic Theory (SALT), volume 23. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wang-2026-do-i-know-what-i/