Foreword To The Special Issue On Tense And Aspect

The phenomena of tense and aspect have long been of interest to linguists and philosophers. Linguists have tried to describe their interesting morphological, syntactic, and semantic properties in the various languages of the world, while philosophers have tried to characterize formally their truth conditions. (For some recent collections of papers, the reader is referred to Tedeschi and Zaenen 1981; Hopper 1982; Dahl 1985; and LoCasio and Vet 1985.) Recently, computational linguists have joined in the act, their interest being sparked by a desire to characterize--at the level of processing--how we understand and describe complex events in a changing world. Here, two kinds of questions converge--one concerning the problem of encoding event descriptions, the other to do with manipulating references to events. In approaching the first question, researchers of all linguistic stripes (computational linguists, philosophers of language, psycholinguists, and linguists of the "unmarked case") have begun to turn their attention from how languages convey information about individuals (or sets of individuals) and their properties to how they convey information about events and situations changing over time. In approaching the second question, computational linguists have become interested in developing systems that can converse with users about events and situations (e.g., for planning) or can process accounts of events and situations (e.g., for summarizing and/or integrating messages). Last year, following the appearance of a number of papers on this topic at the 1987 Conference of the Association for Computational Linguistics at Stanford, it was suggested that a special issue of Computational Linguistics should be devoted to the topic of tense and aspect, in order to examine what appeared to be an emerging consensus on these questions within the computational-linguistics community. This issue is the result of that suggestion, and many of the papers collected below constitute extensions of the papers presented at the Stanford meeting. The papers demonstrate both practical and theoretical advances in our understanding of tense and aspect. With respect to those approaches that have been implemented (cf. Hinrichs, Passonneau), the papers demonstrate how far we have come in developing general methods for extracting and representing event-related information from text and embedding those methods in question-answering and text-processing systems. With respect to theoretical issues involved in how we understand and describe events in a changing world (a subject of all the papers included here), the papers demonstrate the significance of ideas of processing, knowledge representation, and common-sense reasoning drawn from artificial intelligence and computational linguistics. It is these ideas that are computation's unique contribution to our understanding of tense and aspect, augmenting existing contributions from linguistics and philosophy.