Computational Pragmatics

Abstract

Pragmatics as a branch of linguistics can be characterized as the study of the relations between linguistic properties of utterances on the one hand, and aspects of the context in which a given utterance is used on the other. Computational pragmatics is pragmatics with computational means, which include models of dialogue management processes, collections of language use data, annotation schemes and standards, software tools for corpus creation, annotation and exploration, process models of language generation and interpretation, context representations, and inference methods for context-dependent utterance generation and interpretation processes. The linguistic side of the relations that are studied in pragmatics is formed primarily by utterances in a conversation or sentences in a written text. In the case of written text the context side consists of the surrounding text and the setting in which the text is meant to function. In spoken or multimodal dialogue, the context of an utterance is formed by what has been said before and the interactive setting, but additionally by other perceptual, social, and mutual epistemic information (see Context Modeling). Much of this information is dynamic, as it changes during a dialogue and, more importantly, as a result of the dialogue, since the participants in a conversation influence each other’s state of information when they understand each other. Dialogue contexts are thus updated continuously as an effect of communication. Central to computational pragmatics is the development and use of computational tools and models for studying the relations between utterances and their context of use. Essential for understanding these relations are the use of inference and the description of language in terms of actions that are inspired by the context and that are intended to change the context. This bibliography therefore focuses on publications concerned with the computational modeling of dialogue in terms of communicative actions including the use of inference for utterance interpretation. It also considers the more static analysis of discourse coherence and semantic relations in text, and concludes with references to recent activities concerning the construction and use of resources in computational pragmatics, in particular annotation schemes, annotated corpora, and tools for corpus construction and use. The popularity of probabilistic approaches to natural language processing can also be seen in studies of pragmatic aspects of language use, although these approaches are so far not as important as in some other areas of language processing. The so-called rational speech acts (RSA) model treats language use as a recursive process in which probabilistic speaker and listener agents reason about each other’s intentions to enrich the literal semantics of their language along broadly Gricean lines. The core references for this approach are also included in this biography under Inference in Language Processing.