Creating a Translation Matrix of the Bible's Names Across 591 Languages

What it is about

The paper builds a matrix of 1,129 English person and place names from the Bible, each with its translation or transliteration in up to 591 languages, taken from a parallel-Bible corpus. The problem is that a Bible is too small for a standard word aligner: rare names get noisy alignments. The authors refine the alignments in steps. They score candidate pairs with a weighted edit distance over spellings, use a machine-translation model trained on transliterations as a second opinion, and induce affix rules (name endings and beginnings that transfer between languages) to link forms that differ by a suffix. Their alignments beat a widely used word aligner on 97 per cent of the test words. The matrix covers about 52 per cent of the languages for an average name, and more than Wikipedia does for several names. The paper is a resource paper with one main measure, agreement on the test words; it does not test the matrix in a downstream translation system.

Ideas

• I1. Use name pairs as anchors for the Hebrew reader: person and place names are low-noise, so align MACULA's proper nouns to meta-tokens first, with a weighted edit distance between the Hebrew transliteration and the concept's English form, and add it as a prior in ibm.rs. Measure alignment accuracy on the names subset before and after, twice.
• I2. A names table as the standard: the encoder's form for a name concept (Yahweh, Jehovah, the LORD) should come from the name matrix, and author deltas should be the exceptions. Test whether the table lowers encoder bits on Genesis.
• I3. Count-based affix induction for morphology: take the affixes that recur across name pairs and feed them to en/src/morph.rs as candidate rules, each with its count and its error.
• I4. Run the paper's own control on Mova's merges: the shuffled-lexicon control is not red for close authors, so apply a name-only shuffle and count the merges that still survive.

For Mova

The paper does what open problem 7 asks for in the Hebrew reader: a better alignment for words that the frequency-based IBM model cannot place. The names are the easy part, and a good first test. The code is tools/translate/hebrew.rs and ibm.rs, the lexicon is meta.rs, and the morphology is en/src/morph.rs. The matrix itself is under a non-commercial licence, so its data may be used only in tests and measurements, never shipped in what Mova builds; the method is ours to write.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{wu2018creating,
  title = {{Creating a Translation Matrix of the Bible's Names Across 591 Languages}},
  author = {Wu, Winston and Vyas, Nidhi and Yarowsky, David},
  booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
  year = {2018},
  url = {https://aclanthology.org/L18-1263/}
}

References

25 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Dasigi, P. and Diab, M., (2011). Proceedings of the 3rd Named Entities Workshop (NEWS 2011), chapter Named Entity Transliteration Generation Leveraging Statistical Machine Translation Technology, pages 106–111. Asian Federation of Natural Language Processing. find
2. Fraser, A. and Marcu, D. (2006). Semi-supervised training for statistical word alignment. In Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the Association for Computational Linguistics, pages 769–776. Association for Computational Linguistics. find
3. Heafield, K. (2011). Kenlm: Faster and smaller language model queries. In Proceedings of the Sixth Workshop on Statistical Machine Translation, pages 187–197. Association for Computational Linguistics. find
4. Jia, Y., Zhu, D., and Yu, S., (2009). Proceedings of the 2009 Named Entities Workshop: Shared Task on Transliteration (NEWS 2009), chapter A Noisy Channel Model for Grapheme-based Machine Transliteration, pages 88–91. Association for Computational Linguistics. find
5. Koehn, P., Hoang, H., Birch, A., Callison-Burch, C., Federico, M., Bertoldi, N., Cowan, B., Shen, W., Moran, C., Zens, R., Dyer, C., Bojar, O., Constantin, A., and Herbst, E. (2007). Moses: Open source toolkit for statistical machine translation. In Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics Companion Volume Proceedings of the Demo and Poster Sessions, pages 177–180. Association for Computational Linguistics. find
6. Kumar, S., J. Och, F., and Macherey, W. (2007). Improving word alignment with bridge languages. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL). find
7. Lee, J., Lee, D., and Lee, G. G. (2006). Improving phrase-based korean-english statistical machine translation. In INTERSPEECH. find
8. Levenshtein, V. I. (1966). Binary Codes Capable of Correcting Deletions, Insertions and Reversals. Soviet Physics Doklady, 10:707–710, February. find
9. Liang, P., Taskar, B., and Klein, D. (2006). Alignment by agreement. In Proceedings of the Human Language Technology Conference of the NAACL, Main Conference. find
10. Mayer, T. and Cysouw, M. (2014). Creating a massively parallel bible corpus. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014). European Language Resources Association (ELRA). find
11. Mermer, C. and Saraclar, M. (2011). Bayesian word alignment for statistical machine translation. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 182–187. Association for Computational Linguistics. find
12. Moore, R. C. (2005). A discriminative framework for bilingual word alignment. In Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing. find
13. Moravcsik, E. (1978). Language contact. Universals of human language, 1:93–122. find
14. Myers-Scotton, C. (2002). Contact linguistics: Bilingual encounters and grammatical outcomes. Oxford University Press on Demand. find
15. Och, F. J. and Ney, H. (2003). A systematic comparison of various statistical alignment models. Computational Linguistics, Volume 29, Number 1, March 2003. find
16. Och, F. J. (2003). Minimum error rate training in statistical machine translation. In Proceedings of the 41st Annual Meeting on Association for Computational Linguistics-Volume 1, pages 160–167. Association for Computational Linguistics. find
17. Riesa, J. and Marcu, D. (2010). Hierarchical search for word alignment. In Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 157–166. Association for Computational Linguistics. find
18. Sanchis, G. and Sánchez, J. A. (2008). Vocabulary extension via POS information for SMT. Mixing Approaches to Machine Translation. find
19. Song, Y., Kit, C., and Chen, X., (2009). Proceedings of the 2009 Named Entities Workshop: Shared Task on Transliteration (NEWS 2009), chapter Transliteration of Name Entity via Improved Statistical Translation on Character Sequences, pages 57–60. Association for Computational Linguistics. find
20. Songyot, T. and Chiang, D. (2014). Improving word alignment using word similarity. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1840–1845. Association for Computational Linguistics. find
21. Taskar, B., Simon, L.-J., and Dan, K. (2005). A discriminative matching approach to word alignment. In Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing. find
22. Tsvetkov, Y. and Dyer, C. (2016). Cross-lingual bridges with models of lexical borrowing. J. Artif. Intell. Res.(JAIR), 55:63–93. find
23. Vaswani, A., Huang, L., and Chiang, D. (2012). Smaller alignment models for better translations: Unsupervised word alignment with the l0-norm. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 311– 319. Association for Computational Linguistics. find
24. Whitney, W. D. (1881). On mixture in language. Transactions of the American Philological Association (1869-1896), 12:5–26. find
25. Wu, W. and Yarowsky, D. (2018). A comparative study of extremely low-resource transliteration of the world’s languages. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC-2018). European Language Resources Association (ELRA). find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/wu-2018-creating-translation-matrix-bible-s/