arXiv cs.CL· Lucinda Laurent, Muhammad Dehan Al Kautsar, Alfan Farizki Wicaksono, Fajri Koto·· 12 小时前AI 评分26
用多语言情感词典引导低资源语言的情感控制
Steering with Multilingual Lexicons for Sentiment Control in Low-Resource Languages
AI 导读
多语言情感词典可作为激活引导的替代信号，在三个模型家族中，聚合多语言词典对中低资源语言的效果优于英语衍生引导向量，而单语词典在模型预训练专精的语言上仍更强。引导效果由表征密度和预训练先验共同决定：大幅度的英语向量会压垮低资源语言的紧凑表征，造成"语言崩塌"，多语言聚合的范数抑制可缓解这一问题。双语专精惩罚和潜在数值刚性等架构偏差也制约跨语言引导响应。
正文 · 原文
Abstract:Activation steering offers a lightweight mechanism for controlling model behavior, but existing methods depend on sentence-level contrastive data that is unavailable for most of the world's languages. We show that multilingual sentiment lexicons are a viable substitute across three model families: aggregated multilingual lexicons outperform English-derived steering vectors for mid- and low-resource languages, while monolingual lexicons remain stronger for languages a model has specialized in during pre-training. The steering efficacy is mechanistically dictated by both representational density and pre-training priors. Large-magnitude English vectors overwhelm the compact representations of low-resource languages, causing "linguistic collapse." This fragility is mitigated by the norm-dampening of multilingual aggregation. Architectural biases, including bilingual specialization penalty and latent numerical rigidity, also heavily constrain cross-lingual steering responsiveness. Ultimately, our work establishes lexicons as a practical steering signal and offers a comprehensive mechanistic account of cross-lingual steering dynamics.
| Comments: | 30 pages, 12 figures, 11 tables |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.03317 [cs.CL] |
| (or arXiv:2611.03317v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.03317 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Lucinda Laurent [view email]
[v1]
Mon, 2 Nov 2026 15:44:02 UTC (451 KB)
来源：arXiv cs.CL · arxiv.org