No One Fits All: From Fixed Prompting to Learned Routing in Multilingual LLMs

2026-04-18

AI Summary

This work addresses the inconsistent performance of fixed prompting strategies—such as always using native or translated prompts—across languages and tasks in multilingual large language models, highlighting their lack of universality. The authors propose an instance-level prompt strategy routing mechanism that employs a lightweight classifier to dynamically select, for each input, whether to use a native or a translated prompt. This approach overcomes the limitations of static prompting strategies and reveals that the level of language resource availability is a key factor influencing the effectiveness of translated prompts. Experimental results demonstrate that the proposed method significantly outperforms fixed strategies across four multilingual benchmarks and effectively generalizes to task formats unseen during training.

Technology Category

• Natural Language Processing: Prompt Engineering / Prompting
• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

Translation-based prompting is widely used in multilingual LLMs, yet its effectiveness varies across languages and tasks. We evaluate prompting strategies across ten languages of different resource levels and four benchmarks. Our analysis shows that no single strategy is universally optimal. Translation strongly benefits low-resource languages even when translation quality is imperfect, high-resource languages gain little, and prompt-based self-routing underperforms explicit translation. Motivated by these findings, we formulate prompting strategy selection as a learned decision problem and introduce lightweight classifiers that predict whether native or translation-based prompting is optimal for each instance. The classifiers achieve statistically significant improvements over fixed strategies across four benchmarks and generalize to unseen task formats not observed during training. Further analysis reveals that language resource level, rather than translation quality alone, determines when translation is beneficial.

Problem

• multilingual LLMs
• prompting strategy
• translation-based prompting
• language resource level
• strategy selection

Innovation

• learned routing
• multilingual LLMs
• prompting strategy
• translation-based prompting
• language resource level