Milica Denic (ILLC): "Probabilistic informativeness of alternatives in implicature computation"
Monday August 9, 18:15-19:00
How do logical and probabilistic reasoning interact with language interpretation? I will discuss this question in the context of a case study on scalar and ignorance implicatures. All models of these inferences rely on the existence of a set of alternative utterances that the speaker could have said instead of the original assertion. The central question in modeling these inferences is thus: what counts as an alternative for a given sentence? I will argue that probabilistic informativeness of alternatives plays a central role in determining which alternatives feed implicature computation, but that the computation of probabilities needs to be blind to some aspects of contextual knowledge.
Christopher Potts (Stanford): "Rational choices about complex linguistic representations"
Tuesday August 10, 17:30-18:15
There is no tension between "exhaustification-based" approaches to pragmatics and "rational choice" approaches. Exhaustification-based approaches are focused on the nature of interpreted linguistic representations, with special emphasis on unpronounced semantic operators and linguistic constraints on alternatives. Rational choice approaches are focused on how people reason about those representations in context -- for example, how listeners make inferences about speakers' intentions when their utterances are underspecified, and how that inferential process drives speaker behavior. It seems certain that the representations and the reasoning mechanisms are shaping each other. Thus, theories that draw on both of these approaches are likely to be able to explain more than theories that artificially pit them against each other. In my talk, I'll try to substantiate these claims with older examples involving classical implicature patterns as well as newer work that seeks to embed pragmatic theories in large-scale natural language processing systems.
Danny Fox (MIT) and Roni Katzir (Tel Aviv): "Rational choice and/or exhaustification: an assessment"
Tuesday August 10, 18:15-19:00
The rational choice approach to scalar implicatures follows the Gricean and neo-Gricean hope that an adequate understanding of what Grice called “the conditions governing conversation” (CGC) can lead to a simpler, more natural theory of grammar. We consider this hope in view of the converse observation (Fox 2007, 2014) that a slightly more complex grammar can support a simpler, more natural, understanding of CGC. We discuss ways in which one might make progress on resolving the tension between the two perspectives but conclude that the choice is more difficult than it is often made out to be. We then turn to empirical considerations that bear on the division of labor between grammar and theories of CGC, focusing on two domains in which the rational choice approach has been able to present significant accomplishments relating to inferences that could be analyzed as implicatures: conjunctive readings of disjunction and reference games. We use these cases to probe possible motivations for common design choices within the rational choice approach and to compare the approach to the grammatical approach. We conclude that, at least at present, the rational choice approach faces nontrivial challenges in handling basic phenomena within the two domains.
Marie-Christine Meyer (ZAS): "Is implicature reducible to rational and/or cooperative behavior?"
Wednesday August 11, 17:30-18:15
Alexandre Cremers (Institut Jean-Nicod & Vilnius University): "Testing probabilistic models of exhaustivity" [based on joint work with Benjamin Spector and Ethan Wilcox]
Wednesday August 11, 18:15-19:00
The RSA has successfully captured a number of linguistic phenomena at the semantics/pragmatics interface, and offers a new formalization of Grice's theory of implicatures, in particular the fact that "some" is often understood to imply "not all". However, the model also makes the counter-intuitive prediction that when the prior for an all-situation is high, "some" could in fact convey "all" rather than "not all". Degen et al. (2015) confirmed our intuition that this prediction was not borne out, and proposed a variant of the RSA which addresses the issue by downplaying the importance of priors in the model (de facto departing from the assumption that listeners are Bayesian). In this talk I will discuss another kind of implicatures: exhaustivity inferences. If you ask Sue what she had for lunch and she answers "A sandwich", you will typically infer that she had a sandwich and nothing else (i.e., that her answer is exhaustive). This is analogous to the some/all case except for the costs of the various messages. I will present a new study which tested the effects of prior beliefs on both production and comprehension in this situation, improving on previous empirical work. The results confirm that participants' behavior is in clear contradiction with the predictions of RSA, for both production and comprehension. The alternative model proposed by Degen et al. captures the data, but it must assume unrealistic values for some of its parameters. A model built on Spector's (2017) model for homogeneity offers a better fit of the data with acceptable parameter values. The crucial feature needed to capture the data seems to be the ability to break the symmetry between stronger alternatives in the grammar, rather than relying on cost and informativity alone.
Nicole Gotzner (Potsdam): "The measurement mechanism: The role of scale structure and granularity in implicature computation"
Thursday August 12, 17:30-18:15
The meaning of gradable adjectives is standardly modeled in terms of measurement scales (e.g., Kennedy, 2007; Solt, 2015). In this talk, I argue that pragmatic inferences involving adjectival scale-mates (e.g., <large, gigantic> vs. <dirty, filthy>) are based on reasoning about positions on an underlying measurement scale. For example, large may get an upper-bounded interpretation via the exclusion of degrees that are higher on the measurement scale. I present experimental data suggesting that scale structure and granularity affect the derivation of different kinds of implicature in non-negated and negated environments.
Erica Yoon (Stanford) and Michael Henry Tessler (MIT): "Beyond information: Rational use of polite language to communicate social goals"
Thursday August 12, 18:15-19:00
Though language is a remarkably efficient tool for transmitting information, human speakers make statements that are inefficient, imprecise, or even contrary to their own beliefs, all in the service of being polite. What rational machinery underlies polite language use? We show that polite speech emerges from the competition of three communicative goals: to convey information, to be kind, and to present oneself in a good light. We formalize this goal tradeoff using a probabilistic model of utterance production, which predicts human utterance choices in socially sensitive situations with high quantitative accuracy, and we show that our full model is superior to its variants with subsets of the three goals. This utility-theoretic approach to speech acts takes a step toward explaining the richness and subtlety of social language use.
Richard Breheny (UCL): "A limited role for exhaustification in explaining scalar implicature – evidence from scalar diversity"
Friday August 13, 16:00-16:45
It is a conceptual possibility that, apart from exclusion of alternatives, many scalar expressions give rise to SI in virtue of their association with degree scales and our ability to build meanings from these scales that are ‘anti-monotone’. To date, research on only a handful of scalar expressions provide evidence that this possibility is realised – Numerical NPs, paucals, etc. We provide further evidence that, in a more general way, expressions involved in SI may associate with degree scales. Our results suggest that SI may result not only from exclusion but also via this other means. This insight leads to a better understanding of Scalar Diversity. The results have methodological relevance when it comes to issues such as the modularity problem, the symmetry problem, etc.