Abstract
In this paper, we investigate the usefulness of tonal features for unsupervised word discovery, taking Mboshi, a low-resource tonal language from the Bantu family, as our main target language. In a preliminary step, we show that tone annotation improves the performance of supervised learning when using a simplified representation of the data. To leverage this information in an unsupervised setting, we then present a probabilistic model based on a hierarchical Pitman-Yor process that incorporates tonal representations in its backoff structure. We compare our model with a tone-agnostic baseline and analyze if and how tone helps unsupervised segmentation on our small dataset.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
The distinction between high and low tones is phonological (see [14]).
- 2.
We use scikit-learn’s implementation (http://scikit-learn.org/stable/modules/tree.html).
- 3.
With padding at the beginning and end of the sentence.
- 4.
We exclude word boundaries corresponding to the beginning and end of the sentence.
- 5.
Similar results are obtained for the larger corpora.
- 6.
Following [13], we use a forward filtering-backward sampling (FFBS) algorithm to sample segmentations. As this method only approximates the posterior distribution, we also perform a Metropolis-Hastings correction step.
- 7.
where we add initial padding symbols as needed.
- 8.
References
Adda, G., et al.: Breaking the unwritten language barrier: the Bulb project. In: Proceedings of SLTU (Spoken Language Technologies for Under-Resourced Languages). Yogyakarta, Indonesia (2016)
Austin, P.K., Sallabank, J. (eds.): The Cambridge Handbook of Endangered Languages. Cambridge University Press, Cambridge (2011)
Beapami, R.P., Chatfield, R., Kouarata, G., Embengue-Waldschmidt, A.: Dictionnaire Mbochi-Français. SIL-Congo Publishers, Congo (Brazzaville) (2000)
Bird, S., Hanke, F.R., Adams, O., Lee, H.: Aikuma: a mobile app for collaborative language documentation. ACL 2014, 1 (2014)
Blachon, D., Gauthier, E., Besacier, L., Kouarata, G.N., Adda-Decker, M., Rialland, A.: Parallel speech collection for under-resourced language studies using the LIG-Aikuma mobile device app. In: Proceedings of SLTU (Spoken Language Technologies for Under-Resourced Languages). Yogyakarta, Indonesia, May 2016
Bouquiaux, L., Thomas, J.M.C. (eds.): Enquête et description des langues à tradition orale. SELAF, Paris (1976)
Godard, P., et al.: Preliminary experiments on unsupervised word discovery in Mboshi. In: Proceedings of the Interspeech (2016)
Goldwater, S., Griffiths, T.L., Johnson, M.: Contextual dependencies in unsupervised word segmentation. In: Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the Association for Computational Linguistics, Sydney, Australia, pp. 673–680. Association for Computational Linguistics, July 2006. http://www.aclweb.org/anthology/P06-1085
Goldwater, S., Griffiths, T.L., Johnson, M.: Interpolating between types and tokens by estimating power-law generators. In: Advances in Neural Information Processing Systems 18, pp. 459–466. MIT Press, Cambridge (2006)
Goldwater, S., Griffiths, T.L., Johnson, M.: A Bayesian framework for word segmentation: exploring the effects of context. Cognition 112(1), 21–54 (2009)
Johnson, M., Demuth, K.: Unsupervised phonemic Chinese word segmentation using adaptor grammars. In: 23rd International Conference on Computational Linguistics (COLING) (2010)
Ludusan, B., Synnaeve, G., Dupoux, E.: Prosodic boundary information helps unsupervised word segmentation. In: Annual Conference of the North American Chapter of the ACL, pp. 953–963. Denver, Colorado, USA (2015)
Mochihashi, D., Yamada, T., Ueda, N.: Bayesian unsupervised word segmentation with nested Pitman-Yor language modeling. In: Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 1-Volume 1, pp. 100–108. Association for Computational Linguistics (2009)
Rialland, A., Aborobongui, M.E.: How intonations interact with tones in Embosi (Bantu C25), a two-tone language without downdrift. In: Intonation in African Tone Languages, vol. 24. De Gruyter, Berlin, Boston (2016)
Stücker, S., et al.: Innovative technologies for under-resourced language documentation: the Bulb project. In: Proceedings of CCURL (Collaboration and Computing for Under-Resourced Languages : toward an Alliance for Digital Language Diversity). Portoroz̃ Slovenia (2016)
Teh, Y.W., Jordan, M.I., Beal, M.J., Blei, D.M.: Hierarchical Dirichlet processes. J. Am. Stat. Assoc. 101(476), 1566–1581 (2006)
Acknowledgments
This work was partly funded by the French ANR and the German DFG under grant ANR-14-CE35-0002. We warmly thank Martine Adda-Decker and Annie Rialland (from LPP-CNRS) for the sketch on the Mboshi language, as well as Gilles Adda (from LIMSI-CNRS), for many meaningful conversations and contributions.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 Springer Nature Switzerland AG
About this paper
Cite this paper
Godard, P., Löser, K., Allauzen, A., Besacier, L., Yvon, F. (2023). Unsupervised Learning of Word Segmentation: Does Tone Matter?. In: Gelbukh, A. (eds) Computational Linguistics and Intelligent Text Processing. CICLing 2018. Lecture Notes in Computer Science, vol 13396. Springer, Cham. https://doi.org/10.1007/978-3-031-23793-5_28
Download citation
DOI: https://doi.org/10.1007/978-3-031-23793-5_28
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-23792-8
Online ISBN: 978-3-031-23793-5
eBook Packages: Computer ScienceComputer Science (R0)