Context Modeling for Language and Speech Generation

This paper discusses the various ways in which the generation of an utterance requires modeling of the linguistic context of the utterance. To illustrate the role of context modeling in monologue generation, the so-called Dial-Your-Disc (DYD) system is presented, which supports browsing through a large database of musical information and generates a spoken monologue once a musical composition has been selected. The paper devotes special attention to the differences between the 'practical' models of context used in language generation system, on the one hand, and a number of 'theoretical' models of context that have been proposed in theoretical linguistics and (declarative) artificial intelligence, on the other.