magazinelogo

Translation and Foreign Language Learning

ISSN Online: 3069-0315 ISSN Print: 3070-3077 CODEN:
Frequency: monthly Email: tfll@hillpublish.com
Total View: 429776 Downloads: 110776 Citations: 8 (From Dimensions)
ArticleTranslation Theories and Skills http://dx.doi.org/10.26855/tfll.2026.02.008

A Comparative Study on Prompt Strategies for Culture-loaded Word Translation Based on DeepSeek

Zhiyuan Feng, Rong Chen*

School of Foreign Studies, Xi'an University of Posts and Telecommunications, Xi'an 710121, Shaanxi, China.

*Corresponding author: Rong Chen

Postgraduate Innovation Fund of Xian University of Posts and Telecommunications, 2025 (Project No.: CXJJYW2025018) The Educational Reform of Xi’an University of Posts and Telecommunications for Postgraduate Programs “Mecha-nisms and Effectiveness Evaluation of PBL-Based Intercultural Competence Development in MTI Educa-tion” (YJGJ2025040)
Published: February 28, 2026

Abstract

The accurate translation of culture-loaded words remains a major challenge. This study selects 30 material, social, and linguistic culture-loaded words from A Dream of Red Mansions and uses four prompt types with DeepSeek to translate them. The translations are evaluated using BLEU, METEOR, ChrF++, COMET, BERTScore, and blind human assessment. The results reveal criterion-dependent differences: the chain-of-thought prompt performs most strongly in the human assessment and in selected overlap-based metrics, whereas the example-based prompt performs well on several reference-oriented lexical and semantic similarity metrics. The zero-shot baseline and background-instruction prompts show comparatively uneven results across measures. These findings indicate that prompt effectiveness depends on the evaluation criteria and task goals: chain-of-thought prompt is preferable when cultural appropriateness, fluency, and pragmatic effectiveness are prioritized, while example-based prompt may be useful when alignment with a particular reference translation is the primary objective. Overall, the study provides empirical evidence for applying prompt strategies to DeepSeek-based translation of culture-loaded words.

Keyword

Large Language Model; Prompting Strategies; Culture-Loaded Word; DeepSeek

References

Blatz, J., Fitzgerald, E., Foster, G., Gandrabur, S., Goutte, C., Kulesza, A., Sanchis, A., & Ueffing, N. (2004). Confidence estimation for machine translation. Proceedings of the 20th International Conference on Computational Linguistics, 315-321.

Chen, W. (2020). The deconstruction of translator subjectivity by machine translation: Also on the future orientation of human translation. Foreign Language Research, 37(2), 76-83.

Chen, Y. (2025). How prompts as communicative intermediaries shape human-machine intimacy: A case study of generative AI. New Media Research, (14), 12-19, 40.

Deng, J., & Wei, S. (2025). Effectiveness of AI applications in the translation of Chinese medicine terminology. Journal of Hunan University of Chinese Medicine, 45(11), 2196-2201.

Dong, X., Qin, K., & Chen, H. (2025). Quantifying the impact of emotion-driven prompts on translation accuracy of GLLMs: A comprehensive evaluation based on chrF++ and BERTScore. Translation Research and Teaching, (2), 122-131.

Feng, Q. (2025). A model of translation revision competence: Construction, interpretation, and value. Foreign Language World, 86-93.

Feng, Z., & Zhang, D. (2022). Machine translation and human translation boost each other. Journal of Foreign Languages, 45(6), 77-87.

Gao, Y., Wang, R., & Hou, F. (2024). How to design translation prompts for ChatGPT: An empirical study. Proceedings of the 6th ACM International Conference on Multimedia in Asia Workshops, 1-7.

Hakami, A. H., & Abomoati, G. S. (2024). Exploring the impact of prompt formulation in AI chatbots on the translation of English-to-Arabic and Arabic-to-English idioms: A case study. Pakistan Journal of Life and Social Sciences, 22(2), 21371-21381.

Hou, Q., & Hou, R. (2021). A study of neural machine translation: Insights and prospects. Foreign Language Research, (5), 54-59.

Lavie, A., & Denkowski, M. J. (2009). The METEOR metric for automatic evaluation of machine translation. Machine Translation, 23(2), 105-115. https://doi.org/10.1007/s10590-009-9059-4

Li, Y. (2025). TQFLL: An analytic framework for translation quality assessment for large language models of allusions in multilingual corpora (Master’s thesis). Shanghai International Studies University, Shanghai.

Lin, Y., & Wei, M. (2024). AI prompts as logistical media: How large language models shape human linguistic practices. Journalism and Writing, (12), 77-88.

Liu, X. (2026). Generative AI-enabled literary translation quality assessment and strategy optimization: An eco-translatological analysis of DeepSeek-assisted translations of Faulkner’s The Town. Ancient and Modern Cul-tural Creativity, (6), 108-110.

Mu, Y., & Zheng, S. (2025). AI-empowered translation knowledge network construction and a new paradigm of human-machine collaborative teaching. Computer-Assisted Foreign Language Education, 52-61, 112.

Papineni, K., Roukos, S., Ward, T., & Zhu, W. (2002). BLEU: A method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics (pp. 311-318). Philadelphia, PA: Association for Computational Linguistics.

Popović, M. (2017). chrF++: Words helping character n-grams. Proceedings of the Second Conference on Machine Translation, 612-618.

Rei, R., Stewart, C. A., Farinha, A. C., & Lavie, A. (2020). COMET: A neural framework for MT evaluation. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, 2685-2702.

Ren, S. (2012). Liao Qiyi: Exploring contemporary western translation theories. Journal of West China Philology, (2), 212-215, 250.

Song, B., & Liu, L. (2025). Research on the application of large language models in text analysis: A case study of transitivity system annotation in political speech texts. Foreign Language Research, 42(6), 18-25.

Song, X., Xu, D., Zhang, Z., & Zhang, L. (2025). Influence of prompt reformulation types on performance of AI generated content. Library Tribune, 45(4), 129-138.

Sun, L., & Han, C. (2021). English translation of culture-loaded words in Folding Beijing from the perspective of eco-translatology. Shanghai Journal of Translators, (4), 90-94.

Translators Association of China. (2024). China translation industry development report. Beijing: Translators Associa-tion of China.

Wang, Q., Zhao, D., Wang, Y., et al. (2025). Practical pathways for AI-driven transformation and upgrading of the pub-lishing industry. China Publishing Journal, (4), 33-40.

Wang, D., Zhao, Y., Jing, Y., Huang, Q., & Liu, W. (2025). An investigation of AI prompt guides: Feature analysis and a framework for cultivating prompt literacy. Information Studies: Theory & Application, 48(12), 94-105.

Wen, X., & Tian, Y. (2024). The effectiveness of ChatGPT in translating China-specific discourse text. Shanghai Journal of Translators, (2), 27-34, 94-95.

Yang, Y. (2025). Research on application strategies of prompts in design-oriented courses. Art Education, (12), 179–182.

Yu, L. (2024). Lexical diversity and syntactic complexity in ChatGPT translation. Foreign Language Teaching and Re-search, 56(2), 297-307, 321.

Zhai, Q. (2025). LLM-driven written translation teaching: A study on the design of culturally informed prompts based on cognitive load theory. Asia-Pacific Journal of Humanities and Social Sciences, 5(1), 133-142.

Zhang, T., Kishore, V., Wu, F., Weinberger, K. Q., & Artzi, Y. (2019). BERTScore: Evaluating text generation with BERT. arXiv preprint arXiv:1904.09675. https://doi.org/10.48550/arXiv.1904.09675

Zhang, S., & Zhao, C. (2025). Examining the suitability of large language models for literary translation: A multi-indicator assessment of the English translations of Biancheng. Foreign Languages in China, 22(4), 85-95.

Zhou, J., Wang, D., Dai, Q., & Xia, S. (2025). Exploration of the effectiveness of prompt strategies in generative AI conversations. Data Analysis and Knowledge Discovery, 9(9), 49-59.

Copyright

© 2026 by the author(s).
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial-NoDerivatives (CC BY-NC-ND) license, which permits non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited and is not modified or adapted.
https://creativecommons.org/licenses/by-nc-nd/4.0/

How to cite this paper

A Comparative Study on Prompt Strategies for Culture-loaded Word Translation Based on DeepSeek

How to cite this paper: Zhiyuan Feng, Rong Chen. (2026). A Comparative Study on Prompt Strategies for Culture-loaded Word Translation Based on DeepSeek. Translation and Foreign Language Learning2(2), 219-229.

DOI: http://dx.doi.org/10.26855/tfll.2026.02.008