---
title: "Steering with Multilingual Lexicons for Sentiment Control in Low-Resource Languages"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.03317v1"
permalink: "https://feed.astralor.com/items/itm_4d3360ad71cec4e6a8e31866"
published: "2026-11-02T15:44:02.000Z"
score: 57
channel: "ai"
tags: ["RAG","Steering","Multilingual Lexicons","Sentiment Control","Low-Resource Languages Activation","Activation","English-derived","Large-magnitude English"]
---

# Steering with Multilingual Lexicons for Sentiment Control in Low-Resource Languages

## At a glance

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 multi…

## Key points

- 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 m…
- The steering efficacy is mechanistically dictated by both representational density and pre-training priors.

## Full text

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.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_4d3360ad71cec4e6a8e31866