Henk Zeevat

The asymmetry of optimality theoretic syntax and semantics

This paper argues for a combination of semantics and syntax in an optimality theoretic framework that avoids the rat/rad problem and provides simultaneously a certain amount of bidirectionality, in the spirit of Blutner, for an approach to ineffability. It can be succinctly described as taking the program of optimality theoretic syntax as basic, also as a theory of interpretation, and extending it with a bidirectional pragmatic component that is closely related to existing ideas about natural language interpretation. The paper argues for the priority of the direction from content to form, develops the pragmatic component, and argues for the bidirectionality of the pragmatic component on the basis of Grices principle of cooperation. It applies the resulting theory to a small set of relevant examples. The asymmetry in the title is consistent with, but goes beyond, the asymmetry between syntax and semantics used in Smolensky (1996).

Demonstratives in Discourse

There are two influential theories that deal with the role of the context in determining the meaning of sentences: Kaplans logic of demonstratives and Kamps discourse representation theory. How Kaplan would deal with the donkey sentences must remain a matter of speculation, but there is an obvious and reasonable answer to the question of how demonstratives should be handled within discourse representation theory. The latter question is addressed in the first part of this paper. The account proposed here makes demonstratives and indexicals a special case of the treatment of definite NPs in terms of presupposition, like the account of names in Geurts (1997) or treatments of definite descriptions like Van der Sandt (1992), Asher & Lascarides (1998) and others. The treatment turns out to be rather different from Kaplans account of demonstratives and indexicals in that it appears to lack direct referentiality and in that direct referentiality does not entail rigid designation. These problems have been noted before by Kamp and have led to his controversial introduction of external anchors. This paper develops the notion of intensional anchors as an alternative and shows that they - for normal indexicals and demonstratives - allow a proper reconstruction of Kaplans theory that avoids the problems which come with external anchors. Additional evidence for intensional anchors is provided by an application to the problems of intentional identity discussed by Edelberg.

Editors’ Introduction: Pragmatics in Optimality Theory

Based on the tenets of the so-called ‘radical pragmatics’ school (see, for instance, Cole, 1981), this book takes a particular view with regard to the relationship between content and linguistically encoded meaning. The traditional view embodied in the work of Montague and Kaplan (e.g., Kaplan, 1979; Montague, 1970) sees content being fully determined by linguistic meaning relative to a contextual index. In contrast, the radical view takes it that, although linguistic meaning is clearly important to content, it does not determine it, as pragmatic principles also play a role. The central issue of this book is how to give a principled account of the determination of content. Seeing linguistic meanings as underdetermining the content (proposition) expressed, there must be a pragmatic mechanism of completion which can be best represented as an optimization procedure. It is demonstrated that the general framework of Optimality Theory (OT) makes it possible to formulate the desired explanatory principles.

Perspectives on Bayesian Natural Language Semantics and Pragmatics

Bayesian interpretation is a technique in signal processing and its application to natural language semantics and pragmatics (BNLSP from here on and BNLI if there is no particular emphasis on semantics and pragmatics) is basically an engineering decision. It is a cognitive science hypothesis that humans emulate BNLSP. That hypothesis offers a new perspective on the logic of interpretation and the recognition of other people’s intentions in inter-human communication. The hypothesis also has the potential of changing linguistic theory, because the mapping from meaning to form becomes the central one to capture in accounts of phonology, morphology and syntax. Semantics is essentially read off from this mapping and pragmatics is essentially reduced to probability maximation within Grice’s intention recognition. Finally, the stochastic models used can be causal, thus incorporating new ideas on the analysis of causality using Bayesian nets. The paper explores and connects these different ways of being committed to BNLSP.

Integrating Pragmatics into Update Semantics

Instead of two different systems, semantics and pragmatics, with a dedicated interface, we advocate the merging of semantics and pragmatics into the integrating framework of update semantics. In this approach the meaning of an utterance is not characterised by its truth conditions, but by means of the effect it is supposed to have on the information state of its addressee. The first part of the paper gives a full introduction to update semantics and to its application to the problem that triggered its development: the problem of presupposition. In the second part, we investigate its application to the further areas of topic-focus, anaphora and quality and quantity implicatures ([Grice, 1975]), by distinguishing a number of different ways in which utterances or parts of utterances can change a given information state. Updates will be classified into (proper) assertions, corrections, presuppositions and accommodations. Proper assertions are intended to add new information which does not conflict with the information which is already there. Corrections add new information which conflicts with old information and simultaneously remove the source of the conflict. Presuppositions identify old material in the information state and fail if it cannot be found. Accommodations finally, add new material to earlier stages of the information state. The full analysis of an utterance generally involves a combination of these different types of update.

Bayesian NL interpretation and learning

Everyday natural language communication is normally successful, even though contemporary computational linguistics has shown that NL is characterised by very high degree of ambiguity and the results of stochastic methods are not good enough to explain the high success rate. Bayesian natural language interpretation and the combination with speaker self-monitoring are proposed as an explanation of the high success rates. The consequences of the model for language learning are briefly explored (inhibitory effects of production in understanding can only emerge when production is good enough, and inhibitory effects of comprehension in production only when comprehension is good enough) and applied to production-comprehension asymmetries.