Paper

Modeling sentence polarity asymmetries: Fuzzy interpretations in a possibly wonky world

---

Abstract

Negation is an important aspect of human language and reasoning. Prior work has proposed that positive- and negative-polarity sentences exhibit a number of asymmetries. This paper focuses on two of them: (i) Regarding cost, marked forms like negation are known to elicit more production cost than the unmarked positive polarity, and (ii) regarding pragmatic inference, the negative polarity is said to presuppose the prominence of its positive-polarity counterpart, but not the other way around. We present novel empirical evidence regarding these two asymmetries and offer one of the first formalizations of these asymmetries within the Rational Speech Act (RSA) framework. W e show that existing extensions of the standard RSA model, e.g., soft semantics and common ground update, while not originally proposed to address sentence polarity asymmetries, can nonetheless be applicable to these phenomena.

How to Cite:

He, M., Kaiser, E. & Iskarous, K., (2025) “Modeling sentence polarity asymmetries: Fuzzy interpretations in a possibly wonky world”, Society for Computation in Linguistics 8(1): 22. doi: https://doi.org/10.7275/scil.3180