---
title: "Shughni Machine Translation Enhanced by Donor Languages"
description: "Shughni is an Eastern Iranian language of Tajikistan and Afghanistan with about 90,000 speakers, no standard… 14.6 BLEU. 23 references."
authors: ["Dmitry Novokshanov", "Innokentiy S. Humonen", "Ilya Makarov"]
year: 2026
venue: "The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family"
source: "https://aclanthology.org/2026.silkroadnlp-1.12/"
doi: "10.18653/v1/2026.silkroadnlp-1.12"
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/novokshanov-2026-shughni-machine-translation-enhanced-donor/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/novokshanov-2026-shughni-machine-translation-enhanced-donor/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/43/) / Shughni Machine Translation Enhanced by Donor Languages

# Shughni Machine Translation Enhanced by Donor Languages

**Authors:** [Dmitry Novokshanov](https://md.research.elamur.ai/free/en/authors/dmitry-novokshanov/), Innokentiy S. Humonen and [Ilya Makarov](https://md.research.elamur.ai/free/en/authors/ilya-makarov/) · **Year:** 2026 · **Venue:** [The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family](https://md.research.elamur.ai/free/en/venues/silkroadnlp/) ([Proceedings of the Workshop on NLP and LLMs for the Iranian Language Family 2026](https://md.research.elamur.ai/free/en/venues/silkroadnlp/2026/)) · **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/2026.silkroadnlp-1.12/) · [PDF](https://aclanthology.org/2026.silkroadnlp-1.12.pdf) · [DOI 10.18653/v1/2026.silkroadnlp-1.12](https://doi.org/10.18653/v1/2026.silkroadnlp-1.12) · [OpenAlex](https://openalex.org/W7140125943)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Shughni is an Eastern Iranian language of Tajikistan and Afghanistan with about 90,000 speakers, no standard orthography, and almost no digital text. Before this work it had no machine translation, and current commercial LLMs handle it badly. The authors built Shughni–Russian training data from a Shughni corpus and a digitized Shughni–Russian dictionary, converting everything into a single Cyrillic script. They then fine-tuned NLLB-200 in both directions. They also tested whether adding an equal-sized parallel corpus in a second "donor" language helps. Donors were chosen either for typological similarity to Shughni or because earlier work found them broadly helpful as "super donors". Pashto, a close relative, gave the best average gains. The final system combines Pashto data with back-translated Shughni text. It outperforms the plain NLLB baseline and several large open-source LLMs, and native-speaker evaluation confirms the improvement. The result matters because it gives Shughni its first usable MT resource, small enough to run on local devices. It also offers a practical recipe for languages that start with almost no MT data.

## Key points
- **Data:** 69,227 training pairs, drawn from 4,256 corpus sentences (folk tales, oral texts, Gospel passages) and the dictionary (21,871 entries plus 43,300 example sentences). There are two test sets: 123 BivalTyp sentences and 110 NLLB-MD sentences translated by speakers from Khorog.
- **Donor languages tested:** Pashto and Somali (typologically close), Vietnamese and Zulu (distant), and the "super donors" Afrikaans and Slovenian. Pashto was best overall, while Slovenian helped only for translation into Russian.
- **Final model:** NLLB-200 1.3B with Pashto, plus back-translation of a Shughni novel (8,138 sentences) and 195 Shughni–Tajik Gospel sentences.
- **Scores (Shughni→Russian, combined test set):** 36.3 chrF++, 39.3 METEOR and 14.6 BLEU, with 45.7 chrF++ on BivalTyp alone. The 600M variant is better for translation into Shughni (26.2 chrF++).
- **Comparison and demo:** Large open LLMs score far lower; for example, Qwen 3 (235B) reaches 17.5 chrF++. Human XSTS ratings favour the final system in both directions, with κ = 0.65. A Hugging Face demo also translates into English using Russian as a pivot (42.2 chrF++ on BivalTyp).

*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

- [National Research University Higher School of Economics](https://md.research.elamur.ai/free/en/universities/national-research-university-higher-school-of-economics/) — University, Moscow, Russia · 23 works · 39,585 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 (1): [🔁 Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/) · 1,087 works

## Cite

Cite as: Dmitry Novokshanov, Innokentiy S. Humonen and Ilya Makarov. 2026. Shughni Machine Translation Enhanced by Donor Languages. The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family.

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

```bibtex
@inproceedings{novokshanov2026shughni,
  title = {{Shughni Machine Translation Enhanced by Donor Languages}},
  author = {Dmitry Novokshanov, Innokentiy S. Humonen and Ilya Makarov},
  booktitle = {The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family},
  year = {2026},
  doi = {10.18653/v1/2026.silkroadnlp-1.12},
  url = {https://doi.org/10.18653/v1/2026.silkroadnlp-1.12}
}
```

- [← previous Short Story-based Arabic Syntax Learning Media: Development Research at Universitas Negeri Jakarta](https://md.research.elamur.ai/free/en/papers/bahtiar-2026-short-story-based-arabic-syntax/)
- [next → Shuttle Between Symbolic Instructions and Neural Parameters of Large Language Models](https://md.research.elamur.ai/free/en/papers/sun-2026-shuttle-between-symbolic-instructions-neural/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/43/)

## References

23 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Tokhir Kalandarov. 2018. Pamirskie narody, ih jazyki i perepis’: etnicheskij diskurs [The Peoples of Pamir, Their Languages and the Census: The Ethnic Discourse]. Etnograficheskoe obozrenie, (5):162–178. [find](https://scholar.google.com/scholar?q=Tokhir+Kalandarov%2E+2018%2E+Pamirskie+narody%2C+ih+jazyki+i+perepis%E2%80%99%3A+etnicheskij+diskurs+%5BThe+Peoples+of+Pamir%2C+Their+Languages+and+the+Census%3A+The+Ethnic+Discourse%5D%2E+Etnograficheskoe+obozrenie%2C+%285%29%3A162%E2%80%931)
2. Yunsu Kim, Petre Petrov, Pavel Petrushkov, Shahram Khadivi, and Hermann Ney. 2019. Pivot-based transfer learning for neural machine translation between non-english languages. arXiv preprint arXiv:1909.09524. [find](https://scholar.google.com/scholar?q=Yunsu+Kim%2C+Petre+Petrov%2C+Pavel+Petrushkov%2C+Shahram+Khadivi%2C+and+Hermann+Ney%2E+2019%2E+Pivot%2Dbased+transfer+learning+for+neural+machine+translation+between+non%2Denglish+languages%2E+arXiv+preprint+arXiv%3A1909)
3. Philipp Koehn and Rebecca Knowles. 2017. Six challenges for neural machine translation. arXiv preprint arXiv:1706.03872. [find](https://scholar.google.com/scholar?q=Philipp+Koehn+and+Rebecca+Knowles%2E+2017%2E+Six+challenges+for+neural+machine+translation%2E+arXiv+preprint+arXiv%3A1706%2E03872%2E)
4. Daniel Licht META AI, Cynthia Gao META AI, Janice Lam META AI, Francisco Guzmán META AI, Mona Diab, and Philipp Koehn. 2022. Consistent human evaluation of machine translation across language pairs. Volume 1: MT Research Track. [find](https://scholar.google.com/scholar?q=Daniel+Licht+META+AI%2C+Cynthia+Gao+META+AI%2C+Janice+Lam+META+AI%2C+Francisco+Guzm%C3%A1n+META+AI%2C+Mona+Diab%2C+and+Philipp+Koehn%2E+2022%2E+Consistent+human+evaluation+of+machine+translation+across+language+pairs%2E+V)
5. Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2014. [Neural machine translation by jointly learning to align and translate](https://md.research.elamur.ai/free/en/papers/bahdanau-2014-neural-machine-translation-jointly-learning/). arXiv preprint arXiv:1409.0473.
6. Gaëtan Caillaut, Raheel Qader, Mariam Nakhlé, Jingshu Liu, and Jean-Gabriel Barthélemy. 2024. Scaling laws of decoder-only models on the multilingual machine translation task. Preprint, arXiv:2409.15051. [find](https://scholar.google.com/scholar?q=Ga%C3%ABtan+Caillaut%2C+Raheel+Qader%2C+Mariam+Nakhl%C3%A9%2C+Jingshu+Liu%2C+and+Jean%2DGabriel+Barth%C3%A9lemy%2E+2024%2E+Scaling+laws+of+decoder%2Donly+models+on+the+multilingual+machine+translation+task%2E+Preprint%2C+arXiv%3A2409%2E150)
7. Yong Cheng. 2019. Joint training for pivot-based neural machine translation. In Joint training for neural machine translation, pages 41–54. Springer. [find](https://scholar.google.com/scholar?q=Yong+Cheng%2E+2019%2E+Joint+training+for+pivot%2Dbased+neural+machine+translation%2E+In+Joint+training+for+neural+machine+translation%2C+pages+41%E2%80%9354%2E+Springer%2E)
8. Yong Cheng, Yang Liu, Qian Yang, Maosong Sun, and Wei Xu. 2016. Neural machine translation with pivot languages. arXiv preprint arXiv:1611.04928. — [arXiv:1611.04928](https://arxiv.org/abs/1611.04928) [find](https://scholar.google.com/scholar?q=Yong+Cheng%2C+Yang+Liu%2C+Qian+Yang%2C+Maosong+Sun%2C+and+Wei+Xu%2E+2016%2E+Neural+machine+translation+with+pivot+languages%2E+arXiv+preprint+arXiv%3A1611%2E04928%2E+%E2%80%94+%5BarXiv%3A1611%2E04928%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F1611%2E04928%29)
9. Daria Chistiakova and Daria Ryzhova. 2023. Bivalent patterns in Shughni. BivalTyp: Typological database of bivalent verbs and their encoding frames, (5):(Available online at [https://bivaltyp.info](https://bivaltyp.info), Accessed on 26 May 2025.). [↗](https://bivaltyp.info)
10. Jacob Cohen. 1968. Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. Psychological Bulletin, 70(4):213–220. [find](https://scholar.google.com/scholar?q=Jacob+Cohen%2E+1968%2E+Weighted+kappa%3A+Nominal+scale+agreement+provision+for+scaled+disagreement+or+partial+credit%2E+Psychological+Bulletin%2C+70%284%29%3A213%E2%80%93220%2E)
11. Matthew S. Dryer and Martin Haspelmath, editors. 2013. WALS Online (v2020.4). Zenodo. [find](https://scholar.google.com/scholar?q=Matthew+S%2E+Dryer+and+Martin+Haspelmath%2C+editors%2E+2013%2E+WALS+Online+%28v2020%2E4%29%2E+Zenodo%2E)
12. Edel’man and Jusufbekov. 2000. Shugnanskij jazyk [Shughni Language]. In Jazyki mira: Iranskie jazyki. III. Vostochnoiranskie jazyki, page 225. [find](https://scholar.google.com/scholar?q=Edel%E2%80%99man+and+Jusufbekov%2E+2000%2E+Shugnanskij+jazyk+%5BShughni+Language%5D%2E+In+Jazyki+mira%3A+Iranskie+jazyki%2E+III%2E+Vostochnoiranskie+jazyki%2C+page+225%2E)
13. Behrooz Ghorbani, Orhan Firat, Markus Freitag, Ankur Bapna, Maxim Krikun, Xavier Garcia, Ciprian Chelba, and Colin Cherry. 2021. Scaling laws for neural machine translation. Preprint, arXiv:2109.07740. — [arXiv:2109.07740](https://arxiv.org/abs/2109.07740) [find](https://scholar.google.com/scholar?q=Behrooz+Ghorbani%2C+Orhan+Firat%2C+Markus+Freitag%2C+Ankur+Bapna%2C+Maxim+Krikun%2C+Xavier+Garcia%2C+Ciprian+Chelba%2C+and+Colin+Cherry%2E+2021%2E+Scaling+laws+for+neural+machine+translation%2E+Preprint%2C+arXiv%3A2109%2E07740)
14. J. Richard Landis and Gary G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics, 33(1):159–174. [find](https://scholar.google.com/scholar?q=J%2E+Richard+Landis+and+Gary+G%2E+Koch%2E+1977%2E+The+measurement+of+observer+agreement+for+categorical+data%2E+Biometrics%2C+33%281%29%3A159%E2%80%93174%2E)
15. Ilya Loshchilov and Frank Hutter. 2019. Decoupled weight decay regularization. Preprint, arXiv:1711.05101. [find](https://scholar.google.com/scholar?q=Ilya+Loshchilov+and+Frank+Hutter%2E+2019%2E+Decoupled+weight+decay+regularization%2E+Preprint%2C+arXiv%3A1711%2E05101%2E)
16. Yury Makarov, Maksim Melenchenko, and Dmitry Novokshanov. 2022. Digital resources for the Shughni language. In Proceedings of the workshop on resources and technologies for Indigenous, endangered and lesser-resourced languages in Eurasia within the 13th language resources and evaluation conference,… [find](https://scholar.google.com/scholar?q=Yury+Makarov%2C+Maksim+Melenchenko%2C+and+Dmitry+Novokshanov%2E+2022%2E+Digital+resources+for+the+Shughni+language%2E+In+Proceedings+of+the+workshop+on+resources+and+technologies+for+Indigenous%2C+endangered+and+)
17. Vitaly Protasov, Elisei Stakovskii, Ekaterina Voloshina, Tatiana Shavrina, and Alexander Panchenko. 2024. Super donors and super recipients: Studying cross-lingual transfer between high-resource and low-resource languages. In Proceedings of the Seventh Workshop on Technologies for Machine Translatio… [find](https://scholar.google.com/scholar?q=Vitaly+Protasov%2C+Elisei+Stakovskii%2C+Ekaterina+Voloshina%2C+Tatiana+Shavrina%2C+and+Alexander+Panchenko%2E+2024%2E+Super+donors+and+super+recipients%3A+Studying+cross%2Dlingual+transfer+between+high%2Dresource+and+l)
18. Sergey Sergeevič Say. 2020. BivalTyp: Typological database of bivalent verbs and their encoding frames. (Available online at [https://bivaltyp.info](https://bivaltyp.info), Accessed on 1 January 2026.). Institute for Linguistic Studies Russian Academy of Sciences. [↗](https://bivaltyp.info)
19. Rico Sennrich, Barry Haddow, and Alexandra Birch. 2015. Improving neural machine translation models with monolingual data. arXiv preprint arXiv:1511.06709. [find](https://scholar.google.com/scholar?q=Rico+Sennrich%2C+Barry+Haddow%2C+and+Alexandra+Birch%2E+2015%2E+Improving+neural+machine+translation+models+with+monolingual+data%2E+arXiv+preprint+arXiv%3A1511%2E06709%2E)
20. 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, and 20 o… [find](https://scholar.google.com/scholar?q=NLLB+Team%2C+Marta+R%2E+Costa%2Djuss%C3%A0%2C+James+Cross%2C+Onur+%C3%87elebi%2C+Maha+Elbayad%2C+Kenneth+Heafield%2C+Kevin+Heffernan%2C+Elahe+Kalbassi%2C+Janice+Lam%2C+Daniel+Licht%2C+Jean+Maillard%2C+Anna+Sun%2C+Skyler+Wang%2C+Guillaume+We)
21. Jörg Tiedemann. 2012. Parallel data, tools and interfaces in OPUS. In Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’12), Istanbul, Turkey. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=J%C3%B6rg+Tiedemann%2E+2012%2E+Parallel+data%2C+tools+and+interfaces+in+OPUS%2E+In+Proceedings+of+the+Eight+International+Conference+on+Language+Resources+and+Evaluation+%28LREC%E2%80%9912%29%2C+Istanbul%2C+Turkey%2E+European+Langu)
22. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. [Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/). Advances in neural information processing systems, 30.
23. Antje Wendtland. 2009. The position of the pamir languages within east iranian. Orientalia Suecana, 58:172–188. [find](https://scholar.google.com/scholar?q=Antje+Wendtland%2E+2009%2E+The+position+of+the+pamir+languages+within+east+iranian%2E+Orientalia+Suecana%2C+58%3A172%E2%80%93188%2E)

*Updated 1 November 2026*