Article Submitted to Computer Speech and Language Stochastic Natural Language Generation for Spoken Dialog Systems

We describe a corpus-based approach to natural language generation (NLG). The approach has been implemented as a component of a spoken dialog system and a series of evaluations were carried out. Our system uses n-gram language models, which have been found useful in other lan-guage technology applications, in a generative mode. It is not yet clear whether the simple n-grams can adequately model human language gen-eration in general, but we show that we can successfully apply this ubiq-uitous modeling technique to the task of natural language generation for spoken dialog systems. In this paper, we discuss applying corpus-based stochastic language generation at two levels: content selection and sen-tence planning/realization. At the content selection level, output utter-ances are modeled by bigrams, and the appropriate attributes are chosen using bigram statistics. In sentence planning and realization, corpus ut-terances are modeled by n-grams of varying length, and new utterances are generated stochastically. Through this work, we show that a simple statistical model alone can generate appropriate language for a spoken di-alog system. The results describe a promising avenue for using a statistical approach in future NLG systems