Analysis of the Effectiveness of Subword Tokenizers in a Low-Resource Linguistic Environment: Implementation Experience for the Tajik Language
Main Article Content
Abstract
This paper examines modern approaches to subword tokenization of texts as applied to the low-resource Tajik language, which is characterized by a complex morphological structure and a high degree of word-form variability. In the course of the study, a large-scale heterogeneous corpus was compiled and preprocessed, comprising 99 books and 134,497 textual articles of various genres and topics, with a total volume exceeding 33 million tokens. The corpus was cleaned of noise, normalized, and used as a basis for training and subsequent testing of subword models.
Based on this corpus, five tokenization models implementing the BPE, WordPiece, and Unigram algorithms were trained and analyzed using the Hugging Face Tokenizers and SentencePiece libraries. Comparative evaluation was conducted using a set of key metrics, including the proportion of out-of-vocabulary (OOV) words, the degree of text representation compression, tokenization speed, as well as characteristics of n-gram distribution, which make it possible to assess the ability of the models to capture the morphological and structural organization of the language. The experimental results made it possible to identify the strengths and weaknesses of different approaches to subword segmentation and to determine the most effective tokenization strategies under conditions of the morphological complexity of the Tajik language. The findings obtained can be used in the development of language models and applied NLP tools for Tajik and other low-resource languages, contributing to the expansion of their presence in the digital environment.
Article Details
References
2. Vasiliev V.O., Petrov A.A. Problemy obrabotki maloizvestnykh yazykov v sovremennykh NLP sistemakh [Problems of Processing Low-Resource Languages in Modern NLP Systems] // Zhurnal vychislitel'noi lingvistiki i intellektual'nykh tekhnologii. 2021. No. 2(25). P. 45–58.
3. Sennrich R., Haddow B., Birch A. Neural Machine Translation of Rare Words with Subword Units // Proc. of the 54th Annual Meeting of the Association for Computational Linguistics (ACL). Berlin, Germany, 2016. Vol. 1. P. 1715–1725.
4. Kudo T. Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates: preprint arXiv:1804.10959 [cs.CL]. 2018. 10 p. URL: https://arxiv.org/abs/1804.10959 (accessed: 06.04.2025).
5. Kudo T., Richardson J. SentencePiece: A Simple and Language Independent Subword Tokenizer and Detokenizer for Neural Text Processing // Proc. of the 2018 Conference on Empirical Methods in Natural Language Processing: System Demonstrations (EMNLP). Brussels, Belgium, 2018. P. 66–71.
6. Arabov M.K., Sedykh V.V. Sravnitel'nyi analiz metodov modelirovaniya semanticheskikh predstavlenii slov v usloviyakh ogranichennykh yazykovykh resursov: sluchai tadzhikskogo yazyka [Comparative Analysis of Methods for Modeling Semantic Representations of Words in Low-Resource Settings: The Case of Tajik Language] // Nauchno-tekhnicheskii vestnik Povolzh'ya. 2025. No. 6. P. 196–198.
7. Arabov M.K., Makhmadaliev Kh.S., Khabibullozoda K.Kh. Creating a multiformat text corpus for the Tajik language to train modern language models // Science and Innovation. Series of Geological and Technical Sciences. 2025. No. 2. P. 131–136.
8. Devlin J., Chang M. W., Lee K., Toutanova K. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding // Proc. of NAACL-HLT. Minneapolis, USA, 2019. P. 4171–4186.
9. Gage P. A New Algorithm for Data Compression // C Users Journal. 1994. Vol. 12, No. 2. P. 23–38.
10. khovar.tj – News Portal of the Republic of Tajikistan. URL: https://www.khovar.tj (accessed: 06.04.2025).
11. Asia Plus – Tajik Information Service. URL: https://asiaplustj.info (accessed: 06.04.2025).
12. Ovoz i Tojik – Independent Online Media. URL: https://ovozitojik.tj (accessed: 06.04.2025).
13. Farazh – Online Newspaper of Dushanbe. URL: https://farazh.tj (accessed: 06.04.2025).
14. Bartenev O.O. Otsenka effektivnosti metodov tokenizatsii teksta [Evaluating the Efficiency of Text Tokenization Methods] // Vestnik MEI. 2023. No. 6. P. 15–28.
15. Bostrom A., Durrett G. Comparative Analysis of BPE and Unigram Tokenization in RoBERTa Models. Research Report. 2024. URL: https://iris.ru.is/ws/files/240198035/Language_Representation_Models_for_Low_and_Medium_Resource_Languages.pdf (accessed: 10.12.2025).
16. Comparative Analysis of Subword Tokenization Approaches for Indian Languages // Emergent Mind. 2025. URL: https://www.emergentmind.com/articles/2505.16868 (accessed: 10.12.2025).
17. Mikaberidze B., Nadareishvili T., Abashidze M. A Comparison of Different Tokenization Methods for the Georgian Language // Proc. of ICNLSP 2024. 2024. P. 199–208.
18. Park D., Mehta S., Kudo T. Effects of Subword Segmentation on Multilingual Language Models // Proc. of EMNLP. 2023. P. 3504–3518.
19. Arabov M.K. Tajik Language Tokenizers (v1.1). URL: https://huggingface.co/ArabovMK/tajik-tokenizers-v1 (accessed: 06.04.2025).
This work is licensed under a Creative Commons Attribution 4.0 International License.
Presenting an article for publication in the Russian Digital Libraries Journal (RDLJ), the authors automatically give consent to grant a limited license to use the materials of the Kazan (Volga) Federal University (KFU) (of course, only if the article is accepted for publication). This means that KFU has the right to publish an article in the next issue of the journal (on the website or in printed form), as well as to reprint this article in the archives of RDLJ CDs or to include in a particular information system or database, produced by KFU.
All copyrighted materials are placed in RDLJ with the consent of the authors. In the event that any of the authors have objected to its publication of materials on this site, the material can be removed, subject to notification to the Editor in writing.
Documents published in RDLJ are protected by copyright and all rights are reserved by the authors. Authors independently monitor compliance with their rights to reproduce or translate their papers published in the journal. If the material is published in RDLJ, reprinted with permission by another publisher or translated into another language, a reference to the original publication.
By submitting an article for publication in RDLJ, authors should take into account that the publication on the Internet, on the one hand, provide unique opportunities for access to their content, but on the other hand, are a new form of information exchange in the global information society where authors and publishers is not always provided with protection against unauthorized copying or other use of materials protected by copyright.
RDLJ is copyrighted. When using materials from the log must indicate the URL: index.phtml page = elbib / rus / journal?. Any change, addition or editing of the author's text are not allowed. Copying individual fragments of articles from the journal is allowed for distribute, remix, adapt, and build upon article, even commercially, as long as they credit that article for the original creation.
Request for the right to reproduce or use any of the materials published in RDLJ should be addressed to the Editor-in-Chief A.M. Elizarov at the following address: amelizarov@gmail.com.
The publishers of RDLJ is not responsible for the view, set out in the published opinion articles.
We suggest the authors of articles downloaded from this page, sign it and send it to the journal publisher's address by e-mail scan copyright agreements on the transfer of non-exclusive rights to use the work.