Cross-Lingual Prompt Steerability: Towards Accurate and Robust LLM Behavior across Languages

TL;DR

This paper investigates how system prompts influence large language models' performance across multiple languages, proposing a framework to optimize prompts for enhanced accuracy and robustness in multilingual settings.

Contribution

It introduces a unified evaluation framework and an optimization method for system prompts, improving multilingual LLM behavior and reasoning consistency.

Findings

Certain prompt components like CoT, emotion, and scenario enhance multilingual robustness.

Prompt optimization improves metrics by 5-10%.

More effective prompts lead to structured reasoning and less language-switching.

Abstract

System prompts provide a lightweight yet powerful mechanism for conditioning large language models (LLMs) at inference time. While prior work has focused on English-only settings, real-world deployments benefit from having a single prompt to operate reliably across languages. This paper presents a comprehensive study of how different system prompts steer models toward accurate and robust cross-lingual behavior. We propose a unified four-dimensional evaluation framework to assess system prompts in multilingual environments. Through large-scale experiments on five languages, three LLMs, and three benchmarks, we uncover that certain prompt components, such as CoT, emotion, and scenario, correlate with robust multilingual behavior. We develop a prompt optimization framework for multilingual settings and show it can automatically discover prompts that improve all metrics by 5-10%. Finally, we analyze over 10 million reasoning units and find that more performant system prompts induce more structured and consistent reasoning patterns, while reducing unnecessary language-switching. Together, we highlight system prompt optimization as a scalable path to accurate and robust multilingual LLM behavior.