Linguistic Knowledge in Multilingual Grapheme-to-Phoneme Conversion

Roger Yu-Hsiang Lo, Garrett Nicolai


Abstract
This paper documents the UBC Linguistics team’s approach to the SIGMORPHON 2021 Grapheme-to-Phoneme Shared Task, concentrating on the low-resource setting. Our systems expand the baseline model with simple modifications informed by syllable structure and error analysis. In-depth investigation of test-set predictions shows that our best model rectifies a significant number of mistakes compared to the baseline prediction, besting all other submissions. Our results validate the view that careful error analysis in conjunction with linguistic knowledge can lead to more effective computational modeling.
Anthology ID:
2021.sigmorphon-1.15
Volume:
Proceedings of the 18th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology
Month:
August
Year:
2021
Address:
Online
Venues:
ACL | IJCNLP | SIGMORPHON
SIG:
SIGMORPHON
Publisher:
Association for Computational Linguistics
Note:
Pages:
131–140
Language:
URL:
https://aclanthology.org/2021.sigmorphon-1.15
DOI:
10.18653/v1/2021.sigmorphon-1.15
Bibkey:
Cite (ACL):
Roger Yu-Hsiang Lo and Garrett Nicolai. 2021. Linguistic Knowledge in Multilingual Grapheme-to-Phoneme Conversion. In Proceedings of the 18th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 131–140, Online. Association for Computational Linguistics.
Cite (Informal):
Linguistic Knowledge in Multilingual Grapheme-to-Phoneme Conversion (Lo & Nicolai, SIGMORPHON 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.sigmorphon-1.15.pdf