Blessing of Multilinguality: A Systematic Analysis of Multilingual In-Context Learning

TL;DR

This paper systematically analyzes multilingual in-context learning, demonstrating that using mixed high-resource language demonstrations improves performance on low-resource languages and highlighting the benefits of multilingual exposure.

Contribution

It provides a comprehensive understanding of when and why multilingual ICL works well, emphasizing the effectiveness of mixed-language prompts and multilingual exposure.

Findings

Mixed HRL demonstrations outperform English-only prompts.

Including irrelevant non-English sentences in prompts improves performance.

Multilingual exposure helps bridge the gap for underrepresented languages.

Abstract

While multilingual large language models generally perform adequately, and sometimes even rival English performance on high-resource languages (HRLs), they often significantly underperform on low-resource languages (LRLs). Among several prompting strategies aiming at bridging the gap, multilingual in-context learning (ICL) has been particularly effective when demonstration in target languages is unavailable. However, there lacks a systematic understanding of when and why it works well. In this work, we systematically analyze multilingual ICL, using demonstrations in HRLs to enhance cross-lingual transfer. We show that demonstrations in mixed HRLs consistently outperform English-only ones across the board, particularly for tasks written in LRLs. Surprisingly, our ablation study shows that the presence of irrelevant non-English sentences in the prompt yields measurable gains, suggesting the effectiveness of multilingual exposure itself. Our results highlight the potential of strategically leveraging multilingual resources to bridge the performance gap for underrepresented languages.