Character Alignment in Morphologically Complex Translation Sets for Related Languages

What it is about

The task is to align the characters of a word with its translation in a closely related language, for the Ethio-Eritrean Semitic languages (Amharic, Tigrinya, Tigre). These are written in the Ge'ez abugida, where one character often covers a consonant and a vowel, so the morpheme boundaries are hidden. The method is an adaptation of IBM Model 1 with EM, using characters instead of words. Three changes are made: an identity prior (identical characters start with probability 0.5), a monotonic one-to-one constraint, and a left-justification parameter that depends on the difference in word length. Data come from HornMorpho generators over 42 verb roots; similar-root sets are used for training, and 149 hand-annotated word pairs per language are used for testing. Recall and F-score rise with training, and aligning in both directions and taking the union raises recall. Alignments that come out of training recover some regular correspondences, such as a Tigrinya consonant mapping to zero in Amharic. The paper leaves the step from alignment to morpheme merging for future work.

Ideas

• I1. Align the consonant skeletons of Hebrew and Aramaic cognates monotonically, with one-to-one character links, and read off the regular sound correspondences. Use the cognate sets in The lexicon of Biblical Aramaic and The lexicon, and check the result against hand-annotated pairs.
• I2. Start the alignment with an identity prior, and measure how much training adds on top of it. For Mova, the prior can be a fixed table, not EM, so the result stays deterministic (BTreeMap order, no random start).
• I3. Use a left-justification parameter for affix alignment in Aramaic verbs, where the prefixes and suffixes of the imperfect differ in length between dialects. Put it in The verb in Biblical Aramaic as a rule to test against the annotated pairs.

For Mova

I1 and I2 are a test and a method for the lexicon of languages/arc/ (lexicon.md, texts.md) and The lexicon; they can be run with a small tool next to tools/wikt-*. I3 is a check for The verb in Biblical Aramaic. The paper also notes that an orthography that hides vowels hides the morpheme boundaries; this is a reason to keep the pointed and unpointed modes apart in 1. Script and pointing.

Cite

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

@inproceedings{gasser2020character,
  title = {{Character Alignment in Morphologically Complex Translation Sets for Related Languages}},
  author = {Gasser, Michael and Seyoum, Binyam Ephrem and Kifle, Nazareth Amlesom},
  booktitle = {Proceedings of the 7th Workshop on NLP for Similar Languages, Varieties and Dialects},
  year = {2020},
  url = {https://aclanthology.org/2020.vardial-1.5/}
}

References

16 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Solomon Teferra Abate, Michael Melese Woldeyohannis, Martha Yifiru Tachbelie, Million Meshesha, Solomon Atinafu, Wondwossen Mulugeta, Yaregal Assabie, Hafte Abera, Biniyam Ephrem, Tewodros Abebe, Wondim-agegnehu Tsegaye, Amanuel Lemma, Tsegaye Andargie, and Seifedin Shifaw. 2018. Parallel corpora for bi-directional statistical machine translation for seven Ethiopian language pairs. In Proceedings of the First Workshop on Linguistic Resources for Natural Language Processing, pages 83–90, Santa Fe, NM, USA. find
2. Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2015. Neural machine translation by jointly learning to align and translate. In Yoshua Bengio and Yann LeCun, editors, 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings. find
3. Peter F. Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and Robert L. Mercer. 1993. The mathematics of statistical machine translation: Parameter estimation. Computational Linguistics, 19(2):263–311. find
4. Ryan Cotterell, Christo Kirov, John Sylak-Glassman, Géraldine Walther, Ekaterina Vylomova, Arya D. McCarthy, Katharina Kann, Sabrina J. Mielke, Garrett Nicolai, Miikka Silfverberg, David Yarowsky, Jason Eisner, and Mans Hulden. 2018. The CoNLL–SIGMORPHON 2018 shared task: Universal morphological reinflection. In Proceedings of the CoNLL–SIGMORPHON 2018 Shared Task: Universal Morphological Reinflection, pages 1–27, Brussels, October. Association for Computational Linguistics. find
5. Girma A. Demeke. 2001. The Ethio-Semitic languages (re-examining the classification). Journal of Ethiopian Studies, 34(2):57–93. find
6. A. P. Dempster, N. M. Laird, and D. B. Rubin. 1977. Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological), 39(1):1–38. find
7. Greg Durrett and John DeNero. 2013. Supervised learning of complete morphological paradigms. In Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1185–1195, Atlanta, Georgia, June. Association for Computational Linguistics. find our card
8. Ilknur Durgar El-Kahlout, Emre Bektaş, Naime Şeyma Erdem, and Hamza Kaya. 2019. Translating between morphologically rich languages: An Arabic-to-Turkish machine translation system. In Proceedings of the Fourth Arabic Natural Language Processing Workshop, pages 158–166. find
9. Alexander Erdmann, Micha Elsner, Shijie Wu, Ryan Cotterell, and Nizar Habash. 2020. The paradigm discovery problem. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7778–7790. find
10. Michael Gasser. 2011. HornMorpho: a system for morphological processing of Amharic, Oromo, and Tigrinya. In Proceedings of the Conference on Human Language Technology for Development, Alexandria, Egypt. find
11. Huimin Jin, Liwei Cai, Yihui Peng, and Xia. 2020. Unsupervised morphological paradigm completion. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6696–6707. find
12. Arya D. McCarthy, Ekaterina Vylomova, Shijie Wu, Chaitanya Malaviya, Lawrence Wolf-Sonkin, Garrett Nicolai, Christo Kirov, Miikka Silfverberg, Sabrina J. Mielke, Jeffrey Heinz, Ryan Cotterell, and Mans Hulden. 2019. find
13. The SIGMORPHON 2019 shared task: Morphological analysis in context and cross-lingual transfer for inflection. In Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 229–244, Florence, Italy, August. Association for Computational Linguistics. find
14. Franz Josef Och and Hermann Ney. 2003. A systematic comparison of various statistical alignment models. Computational Linguistics, 29(1):19–51. find
15. Gregory Stump. 2017. The nature and dimensions of complexity in morphology. Annual Review of Linguistics, 3:65–83. find
16. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. CoRR, abs/1706.03762. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lingbulk-2020-gasser-character-alignment-morphologically-complex-translation/