Elamur Research⌕
☰
home / papers / 2024 / AGE: Amharic, Ge'ez and English Parallel Dataset

AGE: Amharic, Ge'ez and English Parallel Dataset

Summary

Ge'ez is the ancient liturgical language of Ethiopian and Eritrean Christian churches. It has almost no NLP support: commercial translators and large multilingual models such as NLLB, MT5, ByT5 and M2M-100 do not cover it. Earlier Ge'ez MT studies mostly kept their data and models private. The authors built AGE, an open Amharic–Ge'ez–English parallel corpus. Most of it was scraped from religious sources (Bible sites, Awde Mehret, The Open Siddur Project) and then cleaned and normalized. It also adds 1,000 new sentences from non-religious material such as news and novels, translated into Ge'ez by human translators and checked by reviewers. As a baseline they fine-tuned the distilled NLLB-200 600M model. In Table 3, Ge'ez→English reached 30.66 BLEU, while pairs with Ge'ez as target or with Amharic scored roughly 9–13 BLEU. The authors attribute the gap to English dominating NLLB pretraining and to Ge'ez's rich morphology. The resource matters because it gives Ge'ez a shared, reusable benchmark for the first time.

Key points

  • The corpus has about 17.6K Amharic–Ge'ez pairs and about 18.7K Ge'ez–English pairs. The prose and Table 2 give slightly different counts (17,585/18,676 vs 17,584/18,722). The pair counts differ because some sources, such as Kufale (Book of Jubilees), exist only in Ge'ez–English.

  • The 1,000 newly created Ge'ez-centred sentences were handled by three translators and three evaluators using an in-house tool.

  • Preprocessing removed URLs, hashtags and duplicates, normalized Amharic homophone characters, lowercased English, and dropped badly disordered portions.

  • The model was NLLB-200 600M, fine-tuned with HuggingFace (lr 5e-5, batch 4, max length 128, beam 10) on an 80/10/10 split using Google Colab Pro.

  • Table 3 BLEU scores:

    • Amharic→Ge'ez: 9.39
    • Ge'ez→Amharic: 12.29
    • English→Ge'ez: 12.87
    • Ge'ez→English: 30.66

    The abstract and the discussion state some of these numbers inconsistently. Of the earlier works surveyed in Table 1, only one (Abate et al., 2019) released its data, and none released models.

Interesting ideas

  • I1. The authors fold script-level homophone variants (different letters with the same sound) into one form before training, which cuts sparsity coming from spelling rather than meaning. — For Mova (tokens): A similar normalization pass for English spelling variants (British/American, archaic forms in fables) could map variants to one lemma key before tagging while keeping the original surface form for explanations.
  • I2. Every new sentence goes through separate translator and reviewer roles in a dedicated tool, so each data point has a second human check. — For Mova (learn): Silver trees agreed by witness parsers could likewise pass a distinct "reviewer" step, such as an LLM teacher tier, whose verdicts are logged next to the item.
  • I3. Results are reported separately for each direction, showing that translating into the morphologically rich language is much harder than translating out of it. — For Mova (back): The text → graph → text compressor should be scored in both directions, so that losses in generation are not hidden by strong parsing.
  • I4. The related-work table checks each prior study for whether its data and model are actually available, which makes reproducibility a measured property. — For Mova (learn): A similar availability column for every treebank, lexicon or test set Mova depends on would flag which evaluations can be rerun and independently checked.
  • I5. Aligning three languages around one central language (Ge'ez) makes it possible to train and compare several pairs from shared sentences. — For Mova (back): Translation through the meta-world could use one meaning graph as the hub and check whether outputs in different languages from the same graph agree.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages · Machine translation

Topics: Machine translation

Cited by

3 papers of our library cite it (of the 140884 whose bibliographies we read); 1 of them have a page here, 2 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2024.loresmt-1.14

@inproceedings{ademtew2024age,
  title = {{AGE: Amharic, Ge'ez and English Parallel Dataset}},
  author = {Ademtew, Henok Biadglign and Birbo, Mikiyas Girma},
  booktitle = {Proceedings of the Seventh Workshop on Technologies for Machine Translation of Low-Resource Languages (LoResMT 2024)},
  year = {2024},
  doi = {10.18653/v1/2024.loresmt-1.14},
  url = {https://doi.org/10.18653/v1/2024.loresmt-1.14}
}

References

24 references, 6 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, Martha Yifiru Tachbelie, Million Meshesha, Solomon Atinafu, Wondwossen Mulugeta, Yaregal Assabie, Hafte Abera, Biniyam Ephrem, Tewodros Gebreselassie, et al. 2019. English-ethiopian languages statistical machine translation. In Proceedings of the 2019 Workshop on Widening NLP, pages 27–30. find
  2. Biruk Abel. 2018. Geez to amharic machine translation. find
  3. Ife Adebara and Muhammad Abdul-Mageed. 2022. Towards afrocentric nlp for african languages: Where we are and where we can go. Preprint, arXiv:2203.08351. ↗
  4. David Adelani, Jesujoba Alabi, Angela Fan, Julia Kreutzer, Xiaoyu Shen, Machel Reid, Dana Ruiter, Dietrich Klakow, Peter Nabende, Ernie Chang, Tajuddeen Gwadabe, Freshia Sackey, Bonaventure F. P. Dossou, Chris Emezue, Colin Leong, Michael Beukman, Shamsuddeen Muhammad, Guyo Jarso, Oreen Yousuf, Andre Niyongabo Rubungo, Gilles Hacheme, Eric Peter Wairagala, Muhammad Umair Nasir, Benjamin Ajibade, Tunde Ajayi, Yvonne Gitau, Jade Abbott, Mohamed Ahmed, Millicent Ochieng, Anuoluwapo Aremu, Perez Ogayo, Jonathan Mukiibi, Fatoumata Ouoba Kabore, Godson Kalipe, Derguene Mbaye, Allahsera Auguste Tapo, Victoire Memdjokam Koagne, Edwin Munkoh-Buabeng, Valencia Wagner, Idris Abdulmumin, Ayodele Awokoya, Happy Buzaaba, Blessing Sibanda, Andiswa Bukula, and Sam Manthalu. 2022. A few thousand translations go a long way! leveraging pre-trained models for African news translation. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3053–3070, Seattle, United States. Association for Computational Linguistics. ↗ our card
  5. Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, and Armand Joulin. 2020. Beyond english-centric multilingual machine translation. Preprint, arXiv:2010.11125. ↗
  6. Sefineh Getachew and Yirga Yayeh. 2023. Gex’ez-english bi-directional neural machine translation using transformer. In 2023 International Conference on Information and Communication Technology for Development for Africa (ICT4DA), pages 160–164. find
  7. Barry Haddow, Rachel Bawden, Antonio Valerio Miceli Barone, Jindřich Helcl, and Alexandra Birch. 2022. Survey of low-resource machine translation. Preprint, arXiv:2109.00486. ↗ our card
  8. Franz Josef and Hermann Ney. 2001. Statistical machine translation. find
  9. Tadesse Kassa. 2018. Morpheme-based bi-directional ge’ez -amharic machine translation. find
  10. Ertiban Demewoz Molla and Debre Tabor. 2018. An analysis of ge’ez language heritage potential: traditional church schools and the practices of ethiopian orthodox tewahido churches. find
  11. Dawit Mulugeta. 2015. Geez to amharic automatic machine translation: A statistical approach. find
  12. Wilhelmina Nekoto, Vukosi Marivate, Tshinondiwa Matsila, Timi Fasubaa, Taiwo Fagbohungbe, Solomon Oluwole Akinola, Shamsuddeen Muhammad, Salomon Kabongo Kabenamualu, Salomey Osei, Freshia Sackey, Rubungo Andre Niyongabo, Ricky Macharm, Perez Ogayo, Orevaoghene Ahia, Musie Meressa Berhe, Mofetoluwa Adeyemi, Masabata Mokgesi-Selinga, Lawrence Okegbemi, Laura Martinus, Kolawole Tajudeen, Kevin Degila, Kelechi Ogueji, Kathleen Siminyu, Julia Kreutzer, Jason Webster, Jamiil Toure Ali, Jade Abbott, Iroro Orife, Ignatius Ezeani, Idris Abdulkadir Dangana, Herman Kamper, Hady Elsahar, Goodness Duru, Ghollah Kioko, Murhabazi Espoir, Elan van Biljon, Daniel Whitenack, Christopher Onyefuluchi, Chris Chinenye Emezue, Bonaventure F. P. Dossou, Blessing Sibanda, Blessing Bassey, Ayodele Olabiyi, Arshath Ramkilowan, Alp Öktem, Adewale Akinfaderin, and Abdallah Bashir. 2020. Participatory research for low-resourced machine translation: A case study in African languages. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2144–2160, Online. Association for Computational Linguistics. find our card
  13. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational Linguistics. ↗ our card
  14. Surangika Ranathunga, En-Shiun Annie Lee, Marjana Prifti Skenduli, Ravi Shekhar, Mehreen Alam, and Rishemjit Kaur. 2021. Neural machine translation for low-resource languages: A survey. ACM Computing Surveys. find
  15. Gebru Tareke, Bahru Zewde, and David Pool. 2002. A history of modern ethiopia 1855-1991. The International Journal of African Historical Studies, 35:587. find
  16. NLLB Team, Marta R. Costa-jussà, James Cross, Onur Çelebi, Maha Elbayad, Kenneth Heafield, Kevin Heffernan, Elahe Kalbassi, Janice Lam, Daniel Licht, Jean Maillard, Anna Sun, Skyler Wang, Guillaume Wenzek, Al Youngblood, Bapi Akula, Loic Barrault, Gabriel Mejia Gonzalez, Prangthip Hansanti, John Hoffman, Semarley Jarrett, Kaushik Ram Sadagopan, Dirk Rowe, Shannon Spruit, Chau Tran, Pierre Andrews, Necip Fazil Ayan, Shruti Bhosale, Sergey Edunov, Angela Fan, Cynthia Gao, Vedanuj Goswami, Francisco Guzmán, Philipp Koehn, Alexandre Mourachko, Christophe Ropers, Safiyyah Saleem, Holger Schwenk, and Jeff Wang. 2022. No language left behind: Scaling human-centered machine translation. Preprint, arXiv:2207.04672. ↗
  17. Ermias Tegenaw, Kris Calpotura, and Ashebir Dereje. 2023. Ge’ez to amharic translation with neural network-based technique. find
  18. Atnafu Lambebo Tonja, Tadesse Destaw Belay, Olga Kolesnikova, Seid Muhie Yimam, Abinew Ali Ayele, Grigori Sidorov, and Alexander Gelbukh. 2023. Amhen: Amharic-english large parallel corpus for machine translation. find
  19. Ke M. Tran, Arianna Bisazza, and Christof Monz. 2014. Word translation prediction for morphologically rich languages with bilingual neural networks. In Conference on Empirical Methods in Natural Language Processing. find
  20. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc. find
  21. Aman Kassahun Wassie. 2023. Machine translation for ge’ez language. Preprint, arXiv:2311.14530. ↗ our card
  22. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoest, and Alexander M. Rush. 2020. Huggingface’s transformers: State-of-the-art natural language processing. Preprint, arXiv:1910.03771. ↗ our card
  23. Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, and Colin Raffel. 2022. ByT5: Towards a token-free future with pre-trained byte-to-byte models. Transactions of the Association for Computational Linguistics, 10:291–306. ↗
  24. Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, and Colin Raffel. 2021. mT5: A massively multilingual pre-trained text-to-text transformer. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 483–498, Online. Association for Computational Linguistics. ↗

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ademtew-2024-age-amharic-ge-ez-english/