Abstract

Language models for speech recognition concentrate solely on recognizing the words that were spoken. In this paper, we advocate redefining the speech recognition problem so that its goal is to find both the best sequence of words and their POS tags, and thus incorporate POS tagging. The use of POS tags allows more sophisticated generalizations than are afforded by using a class-based approach. Furthermore, if we want to incorporate speech repair and intonational phrase modeling into the language model, using POS tags rather than classes gives .bet- ter performance in this task.

Citations

3143	Classification and Regression Trees - Breiman, Friedman, et al. - 1983
1654	B.: Building a large annotated corpus of english: The Penn treebank - Marcus, Marcinkiewicz, et al. - 1993
574	Estimation of probabilities from sparse data for the language model component of a speech recognizer - Katz - 1987
540	Class-based n-gram models of natural language - Brown, Pietra, et al. - 1992
302	Self-organized language modeling for speech recognition - Jelinek - 1990
218	ToBI: A standard for labelling English prosody - Silverman, Beckman, et al. - 1992
148	Towards History-based Grammars: Using Richer Models for Probabilistic Parsing - Black, Jelinek, et al. - 1993
142	The Trains project: A case study in building a conversational planning agent - Allen, Schubert, et al. - 1995
136	Natural Language Parsing as Statistical Pattern Recognition - Magerman - 1994
106	A tree-based statistical language model for natural language speech recognition - Bahl, Brown, et al. - 1989
98	Equations for part-of-speech tagging - Charniak, Hendrickson, et al. - 1993
84	Integrating multiple knowledge sources for detection and correction of repairs in humancomputer dialog - Bear, Dowding, et al. - 1992
82	Improved clustering techniques for class-based statistical language modelling - Kneser, Ney - 1993
62	The CMU statistical language modeling toolkit and its use in the 1994 ARPA CSR evaluation - Rosenfeld - 1995
44	The N-Best Algorithm: An Efficient Procedure for Finding Top N Sentence Hypotheses - Chow, Schwartz - 1989
42	A variable-length category-based n-gram language model - Niesler, Woodland - 1996
37	The Trains spoken dialog corpus - Heeman, Allen - 1995
29	Intonational Boundaries, Speech Repairs, and Discourse Markers: Modeling Spoken Dialog - Heeman, Allen - 1997
24	Speech repairs, intonational boundaries and discourse markers: Modeling speakers’ utterances in spoken dialog. Doctoral dissertation - Heeman - 1997
17	Perplexity— A Measure of the Difficulty of Speech Recognition Tasks - Bahl, Baker, et al. - 1977
10	Incorporating POS Tagging into Language Modeling - Heeman, Allen - 1997
4	Deriving phrase-based language models - Heeman, Damnati - 1997
3	Interpolated estimation of markov source paramaters from sparse data - Jelinek, Mercer - 1980
1	Natural language parsing as statistical pattern recognition. Doctoral dissertation - Magerrnan - 1994