---
title: "A Case Against Implicit Standards — paper by Nigatu et al. (2025)"
description: "As an alternative, the authors propose normalizing only the model outputs and references at scoring… 39 references, cited by 1 paper here."
authors: ["Hellina Hailu Nigatu", "Atnafu Lambebo Tonja", "Henok Biadglign Ademtew", "Hizkiel Mitiku Alemayehu", "Negasi Haile Abadi", "Tadesse Destaw Belay", "Seid Muhie Yimam"]
year: 2025
venue: "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing"
source: "https://aclanthology.org/2025.emnlp-main.523/"
doi: "10.18653/v1/2025.emnlp-main.523"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/nigatu-2025-case-against-implicit-standards-homophone/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/nigatu-2025-case-against-implicit-standards-homophone/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/13/) / A Case Against Implicit Standards: Homophone Normalization in Machine Translation for Languages that use the Ge'ez Script.

# A Case Against Implicit Standards: Homophone Normalization in Machine Translation for Languages that use the Ge'ez Script.

**Authors:** [Hellina Hailu Nigatu](https://md.research.elamur.ai/free/en/authors/hellina-hailu-nigatu/), [Atnafu Lambebo Tonja](https://md.research.elamur.ai/free/en/authors/atnafu-lambebo-tonja/), [Henok Biadglign Ademtew](https://md.research.elamur.ai/free/en/authors/henok-biadglign-ademtew/), [Hizkiel Mitiku Alemayehu](https://md.research.elamur.ai/free/en/authors/hizkiel-mitiku-alemayehu/), [Negasi Haile Abadi](https://md.research.elamur.ai/free/en/authors/negasi-haile-abadi/), [Tadesse Destaw Belay](https://md.research.elamur.ai/free/en/authors/tadesse-destaw-belay/) and [Seid Muhie Yimam](https://md.research.elamur.ai/free/en/authors/seid-muhie-yimam/) · **Year:** 2025 · **Venue:** [Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing](https://md.research.elamur.ai/free/en/venues/emnlp/) ([EMNLP 2025](https://md.research.elamur.ai/free/en/venues/emnlp/2025/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2025.emnlp-main.523/) · [PDF](https://aclanthology.org/2025.emnlp-main.523.pdf) · [DOI 10.18653/v1/2025.emnlp-main.523](https://doi.org/10.18653/v1/2025.emnlp-main.523) · [OpenAlex](https://openalex.org/W4416036748)

**Citations:** 1 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/nigatu-2025-case-against-implicit-standards-homophone/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Amharic NLP work often applies homophone normalization as a routine cleanup step. This step collapses Ge'ez-script characters that sound alike in Amharic into one character, which raises exact-match metrics such as BLEU. The authors argue that this habit quietly sets a standard for every language written in the script. Characters that are interchangeable in Amharic can carry distinct sounds and meanings in Tigrinya and Ge'ez.

To test this, they built an MT evaluation set from public corpora for Amharic, Tigrinya and Ge'ez, selecting sentences dense in same-sound characters. They then compared several training setups: no normalization, homophone-only normalization and broader schemes, across zero-shot, monolingual and cross-lingual transfer settings.

The findings are mixed for normalization:
- Its gains are inconsistent across languages and models.
- It harms transfer to related languages.

As an alternative, the authors propose normalizing only the model outputs and references at scoring time, while training on unaltered text. This raised BLEU by up to 1.03 and keeps the model able to handle varied spellings and dialects.

## Key points
- The study covers three languages that share the Ge'ez abugida: Amharic, Tigrinya and Ge'ez. For example, the four letters read as /ʔä/ in Amharic map to four different sounds in Tigrinya, and in Ge'ez swapping such letters can change a word's meaning.
- Prior work has no agreed scheme. Tools variously merge only same-sound characters, add labialized forms, or also merge some same-consonant characters. Variants are usually mapped to the most frequent form.
- Training data sizes are about 199.2k sentence pairs for Amharic, 75.4k for Tigrinya and 15.7k for Ge'ez. Normalization rules were defined separately per language.
- The model from Belay et al. (2022) was trained without normalization. Applying post-inference normalization to its outputs gave +1.03 BLEU on this test set. Belay et al. had reported +3.09 BLEU from fine-tuning M2M on HSL-normalized data, but on a different test set.
- Evaluation relied on BLEU and ChrF plus manual inspection of 50 outputs, with no large-scale human evaluation. Code, data and models are on GitHub (hhnigatu/geez_script_normalization).

## Interesting ideas
- Canonicalize spelling inside the scorer rather than in the training data, so the model keeps seeing real variation while the metric stops penalizing harmless surface differences.
- Report both the raw score and the normalized score, because the gap between them shows how much apparent error is orthographic rather than a real model mistake.
- A normalization that is safe for one language variety can destroy meaning in another variety that shares the same writing system, so each rule should be scoped to the variety it was designed for.
- Build a targeted evaluation set by selecting sentences that are dense in the phenomenon under study, so its effect is measurable rather than diluted.
- Normalization choices spread downstream when machine output is reused as training data for other systems.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (learn): Treebank and silver-data scorers could fold variants such as British/American spellings, curly versus straight quotes, and hyphenation only at comparison time, leaving the parser's input untouched.
- **I2** (answer): Graders for school tests and fable QA could log exact-match and variant-tolerant results side by side, separating spelling slips from wrong reasoning.
- **I3** (tokens): Normalizers in the en crate could carry an explicit tag for the variety or register they apply to, instead of acting as silent global rules.
- **I4** (learn): Stress sets rich in homographs or spelling variants could serve as controls showing whether a preprocessing change actually helps or harms.
- **I5** (back): Generation and translation through the meta-world should preserve the source's spelling choices rather than canonicalizing them, since generated text may later feed training.

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

## Institutions

- [University of California, Berkeley](https://md.research.elamur.ai/free/en/universities/university-of-california-berkeley/) — University, Berkeley, United States · 175 works · 227,440 citations
- [Instituto Politécnico Nacional](https://md.research.elamur.ai/free/en/universities/instituto-politecnico-nacional/) — University, Mexico City, Mexico · 18 works · 4,307 citations
- [Paderborn University](https://md.research.elamur.ai/free/en/universities/paderborn-university/) — University, Paderborn, Germany · 10 works · 3,560 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

🧭 Topics (2): [🔁 Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/) · 1,087 works · [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Hellina Hailu Nigatu, Atnafu Lambebo Tonja, Henok Biadglign Ademtew, Hizkiel Mitiku Alemayehu, Negasi Haile Abadi, Tadesse Destaw Belay and Seid Muhie Yimam. 2025. A Case Against Implicit Standards: Homophone Normalization in Machine Translation for Languages that use the Ge'ez Script. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.

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

```bibtex
@inproceedings{nigatu2025case,
  title = {{A Case Against Implicit Standards: Homophone Normalization in Machine Translation for Languages that use the Ge'ez Script.}},
  author = {Hellina Hailu Nigatu, Atnafu Lambebo Tonja, Henok Biadglign Ademtew, Hizkiel Mitiku Alemayehu, Negasi Haile Abadi, Tadesse Destaw Belay and Seid Muhie Yimam},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  year = {2025},
  doi = {10.18653/v1/2025.emnlp-main.523},
  url = {https://doi.org/10.18653/v1/2025.emnlp-main.523}
}
```

- [← previous You Shall Know a Construction by the Company it Keeps: Computational Construction Grammar with Embeddings](https://md.research.elamur.ai/free/en/papers/verheyen-2025-you-shall-know-construction-company/)
- [next → A Platform for Generating Educational Activities to Teach English as a Second Language](https://md.research.elamur.ai/free/en/papers/rosa-2025-platform-generating-educational-activities-teach/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/13/)

## References

39 references, 10 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 Assibie, Hafte Abera, Binyam Ephrem, Tewodros Abebe, Wondimagegnhue Tsegaye, Amanuel Lemma, Tsegaye Andargie, and Seifedin Shifaw. 2018. Parallel Corpora for bi-lingual English-Ethiopian Languages Statistical Machine Translation. [find](https://scholar.google.com/scholar?q=Solomon+Teferra+Abate%2C+Michael+Melese%2C+Martha+Yifiru+Tachbelie%2C+Million+Meshesha%2C+Solomon+Atinafu%2C+Wondwossen+Mulugeta%2C+Yaregal+Assibie%2C+Hafte+Abera%2C+Binyam+Ephrem%2C+Tewodros+Abebe%2C+Wondimagegnhue+Tseg)
2. Solomon Teferra Abate, Martha Yifiru Tachbelie, and Tanja Schultz. 2020. Multilingual Acoustic and Language Modeling for Ethio-Semitic Languages. In Interspeech 2020, pages 1047–1051. ISCA. [find](https://scholar.google.com/scholar?q=Solomon+Teferra+Abate%2C+Martha+Yifiru+Tachbelie%2C+and+Tanja+Schultz%2E+2020%2E+Multilingual+Acoustic+and+Language+Modeling+for+Ethio%2DSemitic+Languages%2E+In+Interspeech+2020%2C+pages+1047%E2%80%931051%2E+ISCA%2E)
3. Ife Adebara and Muhammad Abdul-Mageed. 2022. Towards afrocentric NLP for African languages: Where we are and where we can go. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3814–3841, Dublin, Ireland. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Ife+Adebara+and+Muhammad+Abdul%2DMageed%2E+2022%2E+Towards+afrocentric+NLP+for+African+languages%3A+Where+we+are+and+where+we+can+go%2E+In+Proceedings+of+the+60th+Annual+Meeting+of+the+Association+for+Computati)
4. Henok Ademtew and Mikiyas Birbo. 2024. [AGE: Amharic, Geez and English Parallel Dataset](https://md.research.elamur.ai/free/en/papers/ademtew-2024-age-amharic-ge-ez-english/). In Proceedings of the Seventh Workshop on Technologies for Machine Translation of Low-Resource Languages (LoResMT 2024), pages 139–145, Bangkok, Thailand. Association for Computational Linguistics. Gabe Adugna. Research: Amharic: Home. Language Learning - Orevaoghene Ahia, Sachin Kumar, Hila Gonen, Jungo Kasai, David Mortensen, Noah Smith, and Yulia Tsvetkov. 2023. Do All Languages Cost the Same? Tokenization in the Era of Commercial Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 9904–9923, Singapore. Association for Computational Linguistics. [↗](https://doi.org/10.48448/dv35-y042)
5. Amsalu Aklilu. 2010. Problems of Writing Homophones without care and its Solution [Title translated from Amhairc]. [find](https://scholar.google.com/scholar?q=Amsalu+Aklilu%2E+2010%2E+Problems+of+Writing+Homophones+without+care+and+its+Solution+%5BTitle+translated+from+Amhairc%5D%2E)
6. Shaik Johny Basha, Duggineni Veeraiah, Boddu Venkat Charan, Wiltrud Sahithi Joyce Yeddu, and Devalla Ganesh Babu. 2023. Detection and Comparative Analysis of Handwritten Words of Amharic Language to English using CNN-Based Frameworks. In 2023 International Conference on Inventive Computation Technologies (ICICT), pages 422–427. ISSN: 2767-7788. [find](https://scholar.google.com/scholar?q=Shaik+Johny+Basha%2C+Duggineni+Veeraiah%2C+Boddu+Venkat+Charan%2C+Wiltrud+Sahithi+Joyce+Yeddu%2C+and+Devalla+Ganesh+Babu%2E+2023%2E+Detection+and+Comparative+Analysis+of+Handwritten+Words+of+Amharic+Language+to+E)
7. Tadesse Destaw Belay, Abinew Ali Ayele, Getie Gelaye, Seid Muhie Yimam, and Chris Biemann. 2021. Impacts of Homophone Normalization on Semantic Models for Amharic. In 2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA), pages 101–106. [find](https://scholar.google.com/scholar?q=Tadesse+Destaw+Belay%2C+Abinew+Ali+Ayele%2C+Getie+Gelaye%2C+Seid+Muhie+Yimam%2C+and+Chris+Biemann%2E+2021%2E+Impacts+of+Homophone+Normalization+on+Semantic+Models+for+Amharic%2E+In+2021+International+Conference+on+)
8. Tadesse Destaw Belay, Atnafu Lambebo Tonja, Olga Kolesnikova, Seid Muhie Yimam, Abinew Ali Ayele, Silesh Bogale Haile, Grigori Sidorov, and Alexander Gelbukh. 2022. The Effect of Normalization for Bidirectional Amharic-English Neural Machine Translation. arXiv preprint. ArXiv:2210.15224 [cs]. [↗](https://arxiv.org/abs/2210.15224)
9. Yohanens Biadgligne and Kamel Smaïli. 2021. Parallel Corpora Preparation for English-Amharic Machine Translation. In Advances in Computational Intelligence, pages 443–455. Springer, Cham. ISSN: 1611-3349. [find](https://scholar.google.com/scholar?q=Yohanens+Biadgligne+and+Kamel+Sma%C3%AFli%2E+2021%2E+Parallel+Corpora+Preparation+for+English%2DAmharic+Machine+Translation%2E+In+Advances+in+Computational+Intelligence%2C+pages+443%E2%80%93455%2E+Springer%2C+Cham%2E+ISSN%3A+1611%2D3)
10. Yohannes Biadgligne and Kamel Smaili. 2022. Offline Corpus Augmentation for English-Amharic Machine Translation. In 2022 5th International Conference on Information and Computer Technologies (ICICT), pages 128–135. [find](https://scholar.google.com/scholar?q=Yohannes+Biadgligne+and+Kamel+Smaili%2E+2022%2E+Offline+Corpus+Augmentation+for+English%2DAmharic+Machine+Translation%2E+In+2022+5th+International+Conference+on+Information+and+Computer+Technologies+%28ICICT%29%2C+)
11. Sidsel Boldsen and Patrizia Paggio. 2022. Letters from the past: Modeling historical sound change through diachronic character embeddings. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6713–6722, Dublin, Ireland. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Sidsel+Boldsen+and+Patrizia+Paggio%2E+2022%2E+Letters+from+the+past%3A+Modeling+historical+sound+change+through+diachronic+character+embeddings%2E+In+Proceedings+of+the+60th+Annual+Meeting+of+the+Association+)
12. Adane Kasie Chekole, Tesfa Tegegne Asfaw, Tesfahun Nurrie Mengestie, Belayneh Teshome Ke-bie, Mengistu Kinfe Negia, and Yohannes Abinet Worku. 2024. Effect of Parallel Data Processing Model on Bi-Directional English-Khimtagne Machine Translation Using Deep Learning. In 2024 International Conference on Information and Communication Technology for Development for Africa (ICT4DA), pages 189–193. [find](https://scholar.google.com/scholar?q=Adane+Kasie+Chekole%2C+Tesfa+Tegegne+Asfaw%2C+Tesfahun+Nurrie+Mengestie%2C+Belayneh+Teshome+Ke%2Dbie%2C+Mengistu+Kinfe+Negia%2C+and+Yohannes+Abinet+Worku%2E+2024%2E+Effect+of+Parallel+Data+Processing+Model+on+Bi%2DDire)
13. Fitehalew Ashagrie Demilew. 2019. ANCIENT GEEZ SCRIPT RECOGNITION USING DEEP CONVO-LUTIONAL NEURAL NETWORK. Software Engineering. [find](https://scholar.google.com/scholar?q=Fitehalew+Ashagrie+Demilew%2E+2019%2E+ANCIENT+GEEZ+SCRIPT+RECOGNITION+USING+DEEP+CONVO%2DLUTIONAL+NEURAL+NETWORK%2E+Software+Engineering%2E)
14. Chris Emezue, Hellina Nigatu, Cynthia Thinwa, Helper Zhou, Shamsuddeen Muhammad, Lerato Louis, Idris Abdulmumin, Samuel Oyerinde, Benjamin Ajibade, Olanrewaju Samuel, Oviawe Joshua, Emeka Onwuegbuzia, Handel Emezue, Ifeoluwatayo A. Ige, Atnafu Lambebo Tonja, Chiamaka Chukwuneke, Bonaventure F. P. Dossou, Naome A. Etori, Mbonu Chinedu Emmanuel, Oreen Yousuf, Kaosarat Aina, and Davis David. 2023. The African Stopwords project: curating stopwords for African languages. arXiv preprint. ArXiv:2304.12155 [cs]. [↗](https://arxiv.org/abs/2304.12155)
15. Abebawu Eshetu. 2022. Amharic-Simple-Text-Preprocessing-Usin-Python. Original-date: 2019-08-05T09:30:04Z. [find](https://scholar.google.com/scholar?q=Abebawu+Eshetu%2E+2022%2E+Amharic%2DSimple%2DText%2DPreprocessing%2DUsin%2DPython%2E+Original%2Ddate%3A+2019%2D08%2D05T09%3A30%3A04Z%2E)
16. 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. 2021. Beyond english-centric multilingual machine translation. J. Mach. Learn. Res., 22(1):107:4839–107:4886. [find](https://scholar.google.com/scholar?q=Angela+Fan%2C+Shruti+Bhosale%2C+Holger+Schwenk%2C+Zhiyi+Ma%2C+Ahmed+El%2DKishky%2C+Siddharth+Goyal%2C+Mandeep+Baines%2C+Onur+Celebi%2C+Guillaume+Wenzek%2C+Vishrav+Chaudhary%2C+Naman+Goyal%2C+Tom+Birch%2C+Vitaliy+Liptchinsky%2C+S)
17. Negasi Haile, Nuredin Ali, and Asmelash Teka Hadgu. 2023. ERROR ANALYSIS OF TIGRINYA EN-GLISH MACHINE TRANSLATION SYSTEMS. [find](https://scholar.google.com/scholar?q=Negasi+Haile%2C+Nuredin+Ali%2C+and+Asmelash+Teka+Hadgu%2E+2023%2E+ERROR+ANALYSIS+OF+TIGRINYA+EN%2DGLISH+MACHINE+TRANSLATION+SYSTEMS%2E)
18. Pratik Joshi, Sebastin Santy, Amar Budhiraja, Kalika Bali, and Monojit Choudhury. 2020. [The State and Fate of Linguistic Diversity and Inclusion in the NLP World](https://md.research.elamur.ai/free/en/papers/joshi-2020-state-fate-linguistic-diversity-inclusion/). In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6282–6293, Online. Association for Computational Linguistics.
19. Raviraj Joshi, Kanishk Singla, Anusha Kamath, Raunak Kalani, Rakesh Paul, Utkarsh Vaidya, Sanjay Singh Chauhan, Niranjan Wartikar, and Eileen Long. 2025. Adapting Multilingual LLMs to Low-Resource Languages using Continued Pre-training and Synthetic Corpus. arXiv preprint. ArXiv:2410.14815 [cs]. [↗](https://arxiv.org/abs/2410.14815)
20. Shreya Khare, Ashish Mittal, Anuj Diwan, Sunita Sarawagi, Preethi Jyothi, and Samarth Bharadwaj. 2021. Low Resource ASR: The Surprising Effectiveness of High Resource Transliteration. In Interspeech 2021, pages 1529–1533. ISCA. [find](https://scholar.google.com/scholar?q=Shreya+Khare%2C+Ashish+Mittal%2C+Anuj+Diwan%2C+Sunita+Sarawagi%2C+Preethi+Jyothi%2C+and+Samarth+Bharadwaj%2E+2021%2E+Low+Resource+ASR%3A+The+Surprising+Effectiveness+of+High+Resource+Transliteration%2E+In+Interspeech+2)
21. Bushra Kidanemariam. 2019. Amharic-NLP-Tools-in-JAVA. [find](https://scholar.google.com/scholar?q=Bushra+Kidanemariam%2E+2019%2E+Amharic%2DNLP%2DTools%2Din%2DJAVA%2E)
22. Diederik P. Kingma and Jimmy Ba. 2017. [Adam: A Method for Stochastic Optimization](https://md.research.elamur.ai/free/en/papers/kingma-2014-adam-method-stochastic-optimization/). arXiv preprint. ArXiv:1412.6980 [cs]. [↗](https://arxiv.org/abs/1412.6980)
23. Surafel M. Lakew, Matteo Negri, and Marco Turchi. 2020. Low Resource Neural Machine Translation: A Benchmark for Five African Languages. arXiv preprint. ArXiv:2003.14402 [cs]. [↗](https://arxiv.org/abs/2003.14402)
24. Daniel Mekuriaw and Arman Cohan. 2024. BENCHMARK DATASET AND PARAMETER-EFFICIENT CROSS-LINGUAL TRANSFER LEARNING FOR AMHARIC TEXT SUMMA-RIZATION. Technical report. [find](https://scholar.google.com/scholar?q=Daniel+Mekuriaw+and+Arman+Cohan%2E+2024%2E+BENCHMARK+DATASET+AND+PARAMETER%2DEFFICIENT+CROSS%2DLINGUAL+TRANSFER+LEARNING+FOR+AMHARIC+TEXT+SUMMA%2DRIZATION%2E+Technical+report%2E)
25. Matt Post. 2018. [A call for clarity in reporting BLEU scores](https://md.research.elamur.ai/free/en/papers/post-2018-call-clarity-reporting-bleu-scores/). In Proceedings of the Third Conference on Machine Translation: Research Papers, pages 186– 191, Brussels, Belgium. Association for Computational Linguistics. Merriam-Webster. Definition of HOMOPHONE.
26. Shivalika Singh, Angelika Romanou, Clémentine Fourrier, David I. Adelani, Jian Gang Ngui, Daniel Vila-Suero, Peerat Limkonchotiwat, Kelly Marchisio, Wei Qi Leong, Yosephine Susanto, Raymond Ng, Shayne Longpre, Wei-Yin Ko, Sebastian Ruder, Madeline Smith, Antoine Bosselut, Alice Oh, Andre F. T. Martins, Leshem Choshen, Daphne Ippolito, Enzo Ferrante, Marzieh Fadaee, Beyza Ermis, and Sara Hooker. 2025. [Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation](https://md.research.elamur.ai/free/en/papers/singh-2025-global-mmlu-understanding-addressing-cultural/). arXiv preprint. ArXiv:2412.03304 [cs]. [↗](https://arxiv.org/abs/2412.03304)
27. Abraham Negash. 2017. The Origin and Development of Tigrinya Language Publications (1886 ... [find](https://scholar.google.com/scholar?q=Abraham+Negash%2E+2017%2E+The+Origin+and+Development+of+Tigrinya+Language+Publications+%281886+%2E%2E%2E)
28. Hellina Hailu Nigatu, Atnafu Lambebo Tonja, Benjamin Rosman, Thamar Solorio, and Monojit Choudhury. 2024. The Zenos Paradox of Low-Resource Languages. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 17753–17774, Miami, Florida, USA. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Hellina+Hailu+Nigatu%2C+Atnafu+Lambebo+Tonja%2C+Benjamin+Rosman%2C+Thamar+Solorio%2C+and+Monojit+Choudhury%2E+2024%2E+The+Zenos+Paradox+of+Low%2DResource+Languages%2E+In+Proceedings+of+the+2024+Conference+on+Empirica)
29. Rubungo Andre Niyongabo, Qu Hong, Julia Kreutzer, and Li Huang. 2020. KINNEWS and KIRNEWS: Benchmarking cross-lingual text classification for In Proceedings of Kinyarwanda and Kirundi. the 28th International Conference on Computational Linguistics, pages 5507–5521, Barcelona, Spain (Online). International Committee on Computational Linguistics. [find](https://scholar.google.com/scholar?q=Rubungo+Andre+Niyongabo%2C+Qu+Hong%2C+Julia+Kreutzer%2C+and+Li+Huang%2E+2020%2E+KINNEWS+and+KIRNEWS%3A+Benchmarking+cross%2Dlingual+text+classification+for+In+Proceedings+of+Kinyarwanda+and+Kirundi%2E+the+28th+Intern)
30. Team NLLB, 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. arXiv preprint. ArXiv:2207.04672 [cs]. [↗](https://arxiv.org/abs/2207.04672)
31. Jane Chinelo Obasi. 2018. Structural Irregularities within the English Language: Implications for Teaching and Learning in Second Language Situations. [find](https://scholar.google.com/scholar?q=Jane+Chinelo+Obasi%2E+2018%2E+Structural+Irregularities+within+the+English+Language%3A+Implications+for+Teaching+and+Learning+in+Second+Language+Situations%2E)
32. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. [Bleu: a method for automatic evaluation of machine translation](https://md.research.elamur.ai/free/en/papers/papineni-2002-bleu-method-automatic-evaluation-machine/). In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational Linguistics. [↗](https://doi.org/10.3115/1073083.1073135)
33. Maja Popović. 2015. [chrF: character n-gram F-score for automatic MT evaluation](https://md.research.elamur.ai/free/en/papers/popovic-2015-chrf-character-n-gram-f/). In Proceedings of the Tenth Workshop on Statistical Machine Translation, pages 392–395, Lisbon, Portugal. Association for Computational Linguistics. [↗](https://doi.org/10.18653/v1/w15-3049)
34. Martha Yifiru Tachbelie, Solomon Teferra Abate, and Laurent Besacier. 2014. Using different acoustic, lexical and language modeling units for ASR of an under-resourced language Amharic. Speech Communication, 56:181–194. [↗](https://doi.org/10.1016/j.specom.2013.01.008)
35. Chantal N. van Dijk, Merel van Witteloostuijn, Nada Vasi, Sergey Avrutin, and Elma Blom. 2016. The Influence of Texting Language on Grammar and Executive Functions in Primary School Children. PLoS ONE, 11(3):e0152409. [↗](https://doi.org/10.1371/journal.pone.0152409)
36. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. 2017. [Attention is All you Need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/).
37. Kimi Wenzel and Geoff Kaufman. 2024. Designing for Harm Reduction: Communication Repair for Multicultural Users’ Voice Interactions. In Proceedings of the CHI Conference on Human Factors in Computing Systems, pages 1–17, Honolulu HI USA. ACM. [find](https://scholar.google.com/scholar?q=Kimi+Wenzel+and+Geoff+Kaufman%2E+2024%2E+Designing+for+Harm+Reduction%3A+Communication+Repair+for+Multicultural+Users%E2%80%99+Voice+Interactions%2E+In+Proceedings+of+the+CHI+Conference+on+Human+Factors+in+Computing+)
38. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi 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 Rush. 2020. [Transformers: State-of-the-art natural language processing](https://md.research.elamur.ai/free/en/papers/wolf-2020-transformers-state-art-natural-language/). In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 38–45, Online. Association for Computational Linguistics. [↗](https://doi.org/10.18653/v1/2020.emnlp-demos.6)
39. Seid Muhie Yimam, Abinew Ali Ayele, Gopalakrishnan Venkatesh, Ibrahim Gashaw, and Chris Biemann. 2021. [Introducing Various Semantic Models for Amharic: Experimentation and Evaluation with Multiple Tasks and Datasets](https://md.research.elamur.ai/free/en/papers/yimam-2020-introducing-various-semantic-models-amharic/). Future Internet, 13(11):275. Number: 11 Publisher: Multidisciplinary Digital Publishing Institute. Target Lang. Source Reference No-Norm H-Only Tir The discourse will also answer such questions as these : How often should Christians commemorate this event ? ብተወሳኺ እቲ ዝርርብ ንኸምዚ ዚስዕብ ዝኣመሰለ ሕቶታት መልሲ ኺህብ እዩ � -ክርስትያናት ነዛ ዕለት እዚኣ ኼብ-ዕልዋ ዘለዎም ክንደይ ሳዕ እዩ ፧ [↗](https://doi.org/10.3390/fi13110275)

*Updated 1 November 2026*