arXiv cs.CL· Dongxu Wu, Sara Court, Micha Elsner·· 11 小时前AI 评分29
研究：LLM 翻译中如何处理词汇知识冲突？九款模型评测
Are These Programmers Ripe Enough to Eat? Evaluating How LLMs Handle Lexical Knowledge Conflicts in Translation
AI 导读
研究者构建受控的普通话到英语翻译任务，将熟悉源词赋予新含义，系统评估词频、句子上下文和提示词框架对九款 LLM 输出的影响。低频源词及支持所给含义的句子上下文会提高模型对上下文知识（CK）的依赖，九款模型中有七款即使被明确要求忽略所给含义仍偏向 CK。模型家族内较大模型更倾向使用 CK，后续实验显示情态弱化可降低 CK 依赖，重复效应取决于提示词布局，推理的影响因模型而异。
正文 · 原文
Abstract:External lexical resources can help large language models (LLMs) translate texts in low-resource languages and specialized domains, but these resources may conflict with models' prior lexical knowledge. To isolate this conflict, we construct a controlled Mandarin-to-English translation task in which familiar source words are assigned new meanings within and across semantic categories. We treat these prompt-supplied meanings as contextual knowledge (CK) and systematically evaluate the effects of word frequency, sentence context, and prompt framing on the translated outputs. Across nine models, CK reliance is greater for less frequent source words and in sentence contexts that support the supplied meanings. Prompt framing can reduce CK reliance, but seven of nine models favor CK even when explicitly instructed to ignore the supplied meanings. Within model families, larger models are more likely than smaller models to utilize CK in their outputs. Three follow-up experiments further show that modal weakening lowers CK reliance, the effect of repetition depends on prompt layout, and the effect of reasoning varies across models. Overall, our controlled study reveals how CK reliance varies across conditions and models, motivating new methods to detect and evaluate lexical conflicts and guide models' lexical choices in LLM-based translation.
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2611.09785 [cs.CL] |
| (or arXiv:2611.09785v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.09785 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Sara Court [view email]
[v1]
Fri, 6 Nov 2026 17:44:14 UTC (1,017 KB)
来源：arXiv cs.CL · arxiv.org