Symposium: The Role of Alternatives in Pragmatic Inference

When interpreting a speaker’s utterance, listeners routinely go beyond the information that is linguistically encoded and draw pragmatic inferences about what the speaker intended to convey. A core aspect of pragmatic inference is that it requires listeners to take into account alternative utterances that a speaker could have produced, but didn’t. For example, a listener who believes that the more informative Alex ate all of the cookies was an alternative a speaker could have used instead of her actual utterance Alex ate some of the cookies will likely expect there to be cookies left over. Similarly, if a speaker says of Alex that he caused the car to stop, a listener will likely infer that he did so in a non-stereotypical way, since she could have instead uttered the simpler, more frequent, he stopped the car. These and many other types of inferences have been studied extensively in the linguistic, philosophical, and psychological literature. While there is consensus that taking into account alternative utterances is important, developing a fullfledged theory of the role of alternatives in pragmatic inference has so far proved elusive due to the complex interactions of various aspects of alternative utterances top-down information about their relative informativeness, structural complexity, and contextual relevance, as well as bottom-up information like frequency and contextual salience. The symposium aims to bring together researchers from a wide range of perspectives and disciplines who have contributed to fundamental questions about the role that alternatives play in pragmatic inference and communication more generally: what are the formal constraints on alternatives how does an utterance become an alternative to a target utterance? How does the naturalness and availability of alternatives affect online processing of pragmatic inferences? How early are children capable of evaluating a target utterance against alternatives? How do speakers choose between alternatives in the first place and can machines be taught to make these choices as humans would? Our plan is to have four 15-minute talks headed by an overview of the many ways in which alternatives are relevant to language use and followed by a 20-30 minute discussion. The overview will be provided by N. Goodman. R. Katzir will then speak about the formal constraints on alternatives in terms of structural complexity, and the reflexes of these constraints in various pragmatic inferences like scalar implicature and association with focus. J. Degen will present work showing that the time course of processing of scalar implicatures is affected by the naturalness of the target utterance, as well as the naturalness and availability of alternatives. D. Barner will address the role that different types of alternatives play in acquisition, in particular on children’s ability to compute various types of scalar and exhaustivity implicatures. A. Gatt will then shift the perspective to how speakers choose between different utterance alternatives. In particular, he will explore the way in which informativeness and contextual salience interact in the production of referring expressions, and developments in modeling this choice computationally. This will set the stage for our discussion, to be facilitated by Goodman and Degen, with the active involvement of audience participants as well as all our speakers.