Skip to content

MIT Linguistics: Department of Linguistics & Philosophy

Massachusetts Institute of Technology


Cross-Disciplinary Perspectives on Scalar Implicatures

Saturday, May 12, 2012; MIT Stata Center Room 32-D461

(Please RSVP if you are thinking of attending the workshop. It will make it easier for us to plan food and drinks.)

8:30 - 9:00
9:00 - 10:00
Gennaro Chierchia (Harvard University): "Three -ities: Unity of Scalarity and Polarity"
The basic idea I will outline is that Scalar and Free Choice phenomena have a unitary character across ordinary scalar items (or, some, or the indefinite article a) and various so called 'affective' items (ever, any, irgendedein, all kinds of 'epistemic' indefinites). I will first discuss some of the empirical evidence on which this claim is based. Then I will propose an account for this phenomenon. The basic idea is that ordinary scalar items (say, the indefinite article a/ein) have the same truth-conditional meaning and activate the same alternatives as affective elements (say, any/irgendein). Such alternatives, when factored into meaning (via a process of exhaustification), yield scalar and FC implicatures. However, the alternatives associated with ordinary scalar items are subject to relevance: they are factored into meaning only when the result is relevant to the conversational goals. The alternatives associated to affective items are not subject to relevance: they must always be factored into meaning. This entails that they always bring about scalar and FC implicatures, which are responsible for their narrower distribution with respect to their ordinary scalar cousins. Under this view, the role of dedicated affective morphology (e.g. German irgend- in irgendein, English -y in an-y from O.E. ænig lit. "one-y,") is to mark the inapplicability of relevance to the target items. This view of the phenomenon bears interesting connections to and/or invites a retooling of notions like 'conventional implicature' and 'short-circuited implicature'.
10:00 - 11:00
Roni Katzir (Tel Aviv University): "A note on implicatures and focus"
As noted by Rooth (1992), focus interpretation and implicatures are closely related, both involving alternatives that depend on an interaction of grammatical and contextual factors. While the alternatives for scalar implicatures and focus have often been defined separately, Fox & Katzir (2011) propose that the alternatives are in fact the same, and that they are derived via substitutions within focused constituents. By contrast, Wagner (2007,2011) has argued that focus relies on a set of alternatives derived via reference to the sister of the focused constituent. I will re-examine both arguments and try to show that, contra Wagner, focus alternatives do not make reference to sister nodes.
11:00 - 12:00
Gerhard Jäger (University of Tübingen): "Trust is good, strategic thinking is better."
The talk will give an overview of the Iterated Best Response model of game theoretic pragmatics. After presenting the model and going through basic examples, I will discuss scenarios where the interests of the interlocutors are not aligned, or if they are boundedly rational.
Lunch (8th floor lounge)
1:30 - 2:30
Noah Goodman (Stanford University): "Modeling language understanding as social cognition"
Is language understanding a special case of social cognition? To help evaluate this view, we formalize it as the rational speech-act theory: listeners assume that speakers choose their utterances approximately optimally, and listeners interpret an utterance by using Bayesian inference to ``invert'' this model of the speaker. This theory enables the construction of quantitative models of pragmatic inference that can be compared to quantitative experimental data. I will apply this framework to both scalar implicature and ad-hoc implicature. In the former case, this model predicts an interaction between the speaker's knowledge state and the listener's interpretation. These predictions have been tested in several behavioral experiments, revealing good fit between model predictions and human judgements.
2:30 - 3:30
Richard Breheny (University College London): Integration or Diversity - Local adjustments to meaning in different cognitive frameworks
In recent times, there have been debates among theoretical linguists about the status of scalar or quantity implicatures and debates among psycholinguists about the place of implicatures in the cognitive architecture underpinning language use. On the theoretical side, interest has been focussed on the question whether scalar or quantity implicatures should be explained in terms of operators represented in syntactic structure rather than in terms of certain more standard accounts based on general all-purpose ‘global’ pragmatic reasoning. In psycholinguistics, there is a question about how conversational implicatures are accessed and integrated during on-line processing. The aim of this paper is to make a link between these two debates and present experimental evidence for a non-syntacticised version of localism underpinned by pragmatic principles.
3:30 - 4:00
Coffee Break
4:00 - 5:00
David Barner (University of California, San Diego): "Grammatical alternatives and conceptual change"
Circa 1950, a common view was that natural language could not be understood using known logical models, since language is filled with vagueness, ambiguity, and context sensitivity. The emergence of pragmatics as a branch of linguistic theory paved the way for the use of logical models to describe natural language, by narrowing and simplifying the explananda of semantic theory. In this talk, I discuss a similar problem facing the current study of conceptual development, and in particular the acquisition of number words. Current approaches notice that prelinguistic numerical representations lack the content of the positive integers, and also argue that natural language quantifiers do not have exact meanings, unlike words like one, two, and three. On this basis, these approaches argue that acquiring this number words must involve a conceptual change in childhood, wherein the child creates new logical resources that are incommensurable with early linguistic and non-linguistic concepts. On analogy with the invocation of pragmatics in the 1960s, I argue that the same semantic models that we use to describe natural language quantifiers can be used to describe early number word meanings, and that the apparent discrepancy between the content of these forms is attributable to pragmatic differences between them (plus later procedural learning related to counting). In support of this, I argue that children strengthen numeral meanings via scalar implicature from the beginning of acquisition, and that core meanings are similar to singular, dual, etc. I show that children's failures with implicature in other domains (e.g., quantifiers and disjunction) can be explained by problems accessing alternatives. To do this, I present evidence regarding children's interpretation of sentences involving "only", as well as children's relatively strong ability to compute ignorance implicatures, which are formally similar to scalar implicatures, at which they fail. I also show that children use pragmatic inference to reason about number words even before they have acquired their specific meanings, giving the illusion that even these words are exact.
5:00 - 6:00
Jesse Snedeker (Harvard University): "Good things come to those who wait"
I will review two lines of research from our lab which find that the implicaure for "some" is calculated quite slowly (consistent with the much of the prior research). Next, I'll review the counter evidence, propose an explanation for the discrepancy, and provide some evidence for this alternative. Then I'll show you some nifty ERP data that may help us understand what is happening during this delay. Along the way, I'll explain why I believe that the mechanism by which numerically quantified nouns receive an upper bound is different. If things are going really well, I may say a bit about autism. These projects were conducted in collaboration with Yi Ting Huang, Josh Hartshorne, Daniele Pannizza, Manizeh Khan, Al Kim & Noemi Hahn
Dinner (TBD)