Abstract: Availability of large corpora for training language models to develop dialogue systems is rare. Fortunately, for specific dialogue application, many sentences follow a limited number of typical patterns. In a language like French, frequent errors are due to homophones.Three paradigms are proposed in this paper to rescore a trellis of hypothesized words. They are based on sentence patterns detected in the most likely sentence hypothesized in a first recognition phase.

This paperintroduces a simple mixtare language model that attempts to capture long distance conslraints in a sentence or paragraph. The model is an m-component mixture of Irigram models. The models were constructed using a 5K vocabulary and trained using a 76 mil-lion word Wail Street Journal text corpus. Using the BU recognition system, experiments show a 7 % improvement in recognition accu-racy with the mixture trigram models as compared to using a Irigram model. 1.

This paper focuses on the learning of multi-word lexical units, or phrases, and how to model them within the variable n-gram framework. We introduce the notion of contextdependent phrases and suggest an algorithm for unsupervised learning of phrases. Also, we propose an approach to integrate a phrase grammar and a variable n-gram without the need of explicitly handling multi-word lexical items. The combined variable n-gram phrase grammar improves recognition accuracy on the Switchboard corpus over both the baseline trigram and using a variable n-gram alone. 1. INTRODUCTION Although words in English are reasonable lexical units for language modeling, there are many cases that longer lexical units may be more appropriate. Frequently used word sequences, such as I mean or you know, are so common in conversational speech that they may be effectively used by the speaker as a single lexical item. We call these multiword units "phrases". There are several ways of treating a multi-word sequ...