Computational and Conversational Discourse : Burning I s s u e s - - A n Interdisciplinary Account

The array of perspectives from which discourse processing is pursued can be mindboggling to students of discourse, both new and experienced. We are bombarded with concepts from various arenas that require sorting out: topic and focus, given and new, cohesion and coherence; the list goes on. Many questions loom for the computational linguistics (CL) researcher. What do notions such as topic and focus and given and new have to say about how discourses are interpreted, as modeled by formal approaches to coherence resolution? More generally, do works from traditionally functional perspectives contain deep insights, obscured by the lack of formal tools necessary to make them concrete, or does the articulate yet complex prose in these works obscure a lack of methodological rigor in their analyses? Is it time to follow other subdisciplines within CL and eschew theoretical work in favor of purely empirical approaches, or would such a move be wildly premature? Computational and Conversational Discourse, an outgrowth of a workshop entitled "Burning Issues in Discourse" held in April 1993 in Maratea, Italy, captures a snapshot of current discourse studies at the breadth necessary to begin addressing these questions.