Author(s)

ZiQing Zhong, XiaoMing Lu^*, TianTian Gong

Affiliation(s)

School of Interpreting and Translation Studies, Guangdong University of Foreign Studies, Guangzhou 510420, Guangdong, China.

Corresponding Author

XiaoMing Lu

ABSTRACT

Translating euphemisms and politeness expressions between languages with distinct cultural norms presents significant challenges for automated translation systems, particularly for low-resource language (LRL) pairs. This study examines the effectiveness of integrating a custom-built knowledge base (KB) into large language models (LLMs) to enhance the translation of euphemistic expressions and politeness markers between Persian, Chinese, and English, with a focus on the Persian cultural schema of taarof. Two qualitative translation assessments were conducted: a Persian ceremonial speech translated into Chinese, and English audiovisual subtitles translated into Chinese and subsequently into Persian. Five translation versions were produced using LLMs with and without a domain-specific knowledge base, and evaluated by one Chinese lecturer and two native Persian speakers. Findings reveal that LLMs integrated with the knowledge base consistently outperform standalone models in terms of semantic accuracy and politeness transfer. However, divergent evaluations between evaluators highlight that politeness perception remains culturally contingent. This study demonstrates the value of knowledge base integration for LRL translation while underscoring the continued necessity of human’s machine translation post editing efforts.

KEYWORDS

Taarof; Low-resource languages; Knowledge base; Large language models

CITE THIS PAPER

ZiQing Zhong, XiaoMing Lu, TianTian Gong. Improving real-time low-resource language euphemism translation with agent knowledge base. AI and Data Science Journal. 2025, 2(3): 19-25. DOI: https://doi.org/10.61784/adsj3029.

REFERENCES

[1] Dong Xiaojun. A Discussion on the Communicative Functions of Euphemisms in English Subtitle Translation: A Case Study of the British Drama Downton Abbey. English Square, 2016(10): 5-7. DOI:10.16723/j.cnki.yygc.2016.10.002.

[2] Naumenko S, Kryvonis M. Linguistic means of forming euphemisms in the modern Chinese language. Cognition, Communication, Discourse, 2023(27): 44-55.

[3] Kamehkhosh N, Larina T V. Cultural values and politeness strategies in British and Persian family discourse. Proceedings of INTCESS, 2020: 7th.

[4] Tayebi T. Heterogeneous distribution of cultural conceptualizations and (im) politeness evaluations. International Journal of Language and Culture, 2020, 7(1): 84-103.

[5] Sharifian F, Tayebi T. Perception of (im) politeness and the underlying cultural conceptualisations: A study of Persian. Pragmatics and Society, 2017, 8(2): 231-253.

[6] Kádár D Z, House J, Liu H. Beyond the deferential view of the Chinese V pronoun nin. Pragmatics, 2025, 35(2): 155-184.

[7] Yaghoubzadeh S. A study of requests made by negative politeness in English movies and their Persian subtitles. Theory and Practice in Language Studies, 2021, 11(2): 216-223.

[8] Zohreh Ameri Safiabadi. A Comparative Study of Politeness Formulae in Chinese and Persian. Zhejiang University, 2024. DOI:10.27461/d.cnki.gzjdx.2024.001957.

[9] Ji Kaiyun, Yang Chen Ying. A Cultural Interpretation of Iranian Taarof from the Perspective of Cultural Anthropology. Middle East Studies, 2021(01): 147-164+265-266.

[10] Izadi A. Over-politeness in Persian professional interactions. Journal of Pragmatics, 2016, 102: 13-23.

[11] Wang X, Yang Q, Qiu Y, et al. Knowledgpt: Enhancing large language models with retrieval and storage access on knowledge bases. arXiv preprint arXiv:2308.11761, 2023.

[12] Yaqubi M. On subtitling of ta’arof apologies. Journal of Language and Translation, 2018, 8(1): 31-42.