Book Description

The variety of surface forms that may be used to convey a given speech act pose a major problem in modelling task-oriented (and other) dialogues. Many such forms are so-called indirect speech acts, that is, surface form does not correspond to the (or one) intended speech act. While this topic has received attention from linguists, their concerns have not usually been computationally motivated. In this paper, I present a non-computational analysis of indirect speech act forms with an eye to computational considerations. The paper is divided into two parts. Part 1 presents categories and rules for indirect speech acts, justified where possible by traditional linguistic arguments. The second part of the paper draws a set of computational implications from the material presented in Part 1. This is done within the general framework of a process model of recognition. Part 2 contains a discussion of the basic types of mechanisms needed for the classes of indirect speech act identified in Part 1. The discussion includes an examination of the dependencies between processes and an initial categorization of the types of knowledge that must be considered in interpreting indirect speech acts. (Author).