Abstract
Statistical language models (SLMs) for speech recognition have the advantage of robustness, and grammar-based models (GLMs) the advantage that they can be built even when little corpus data is available. A known way to attempt to combine these two methodologies is first to create a GLM, and then use that GLM to generate training data for an SLM. It has however been difficult to evaluate the true utility of the idea, since the corpus data used to create the GLM has not in general been explicitly available. We exploit the Open Source Regulus platform, which supports corpus-based construction of linguistically motivated GLMs, to perform a methodologically sound comparison: the same data is used both to create an SLM directly, and also to create a GLM, which is then used to generate data to train an SLM. An evaluation on a medium-vocabulary task showed that the indirect method of constructing the SLM is in fact only marginally better than the direct one. The method used to create the training data is critical, with PCFG generation heavily outscoring CFG generation.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Rabiner, L.: A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE 77(2), 257–286 (1989)
Moore, R.: Using natural language knowledge sources in speech recognition. In: Proceedings of the NATO Advanced Studies Institute, pp. 115–129 (1998)
Dowding, J., Hockey, B., Gawron, J., Culy, C.: Practical issues in compiling typed unification grammars for speech recognition. In: Proceedings of the 39th Annual Meeting of the Association for Computational Linguistics, Toulouse, France, pp. 164–171 (2001)
Rayner, M., Dowding, J., Hockey, B.: A baseline method for compiling typed unification grammars into context free language models. In: Proceedings of Eurospeech 2001, Aalborg, Denmark, pp. 729–732 (2001)
Bos, J.: Compilation of unification grammars with compositional semantics to speech recognition packages. In: Proceedings of the 19th International Conference on Computational Linguistics, Taipei, Taiwan (2002)
Stent, A., Dowding, J., Gawron, J., Bratt, E., Moore, R.: The CommandTalk spoken dialogue system. In: Proceedings of the Thirty-Seventh Annual Meeting of the Association for Computational Linguistics, pp. 183–190 (1999)
Knight, S., Gorrell, G., Rayner, M., Milward, D., Koeling, R., Lewin, I.: Comparing grammar-based and robust approaches to speech understanding: a case study. In: Proceedings of Eurospeech 2001, Aalborg, Denmark, pp. 1779–1782 (2001)
Rayner, M., Hockey, B., Renders, J., Chatzichrisafis, N., Farrell, K.: A voice enabled procedure browser for the International Space Station. In: Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (interactive poster and demo track), Ann Arbor, MI (2005)
Chatzichrisafis, N., Bouillon, P., Rayner, M., Santaholma, M., Starlander, M., Hockey, B.: Evaluating task performance for a unidirectional controlled language medical speech translation system. In: Proceedings of the HLT-NAACL International Workshop on Medical Speech Translation, New York, pp. 9–16 (2006)
Wang, Y.-Y., Acero, A., Chelba, C., Frey, B., Wong, L.: Combination of statistical and rule-based approaches for spoken language understanding. In: Proceedings of the 7th International Conference on Spoken Language Processing (ICSLP), Denver, CO, pp. 609–612 (2002)
Jurafsky, A., Wooters, C., Segal, J., Stolcke, A., Fosler, E., Tajchman, G., Morgan, N.: Using a stochastic context-free grammar as a language model for speech recognition. In: Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 189–192 (1995)
Jonson, R.: Generating statistical language models from interpretation grammars in dialogue systems. In: Proceedings of the 11th EACL, Trento, Italy (2006)
Rayner, M., Hockey, B., Bouillon, P.: Putting Linguistics into Speech Recognition: The Regulus Grammar Compiler. CSLI Press, Chicago (2006)
Rayner, M., Bouillon, P., Chatzichrisafis, N., Hockey, B., Santaholma, M., Starlander, M., Isahara, H., Kanzaki, K., Nakao, Y.: A methodology for comparing grammar-based and robust approaches to speech understanding. In: Proceedings of the 9th International Conference on Spoken Language Processing (ICSLP), Lisboa, Portugal, pp. 1103–1107 (2005)
Wang, Y.Y., Acero, A., Chelba, C.: Is Word Error Rate a good indicator for spoken language understanding accuracy. In: Proceedings of Eurospeech 2003, Geneva, Switzerland, pp. 609–612 (2003)
Bouillon, P., Rayner, M., Chatzichrisafis, N., Hockey, B., Santaholma, M., Starlander, M., Nakao, Y., Kanzaki, K., Isahara, H.: A generic multi-lingual open source platform for limited-domain medical speech translation. In: Proceedings of the 10th Conference of the European Association for Machine Translation (EAMT), Budapest, Hungary, pp. 50–58 (2005)
Bouillon, P., Halimi, S., Nakao, Y., Kanzaki, K., Isahara, H., Tsourakis, N., Starlander, M., Hockey, B., Rayner, M.: Developing non-European translation pairs in a medium-vocabulary medical speech translation system. In: Proceedings of LREC 2008, Marrakesh, Morocco (2008)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2008 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Hockey, B.A., Rayner, M., Christian, G. (2008). Training Statistical Language Models from Grammar-Generated Data: A Comparative Case-Study. In: Nordström, B., Ranta, A. (eds) Advances in Natural Language Processing. GoTAL 2008. Lecture Notes in Computer Science(), vol 5221. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-85287-2_19
Download citation
DOI: https://doi.org/10.1007/978-3-540-85287-2_19
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-85286-5
Online ISBN: 978-3-540-85287-2
eBook Packages: Computer ScienceComputer Science (R0)