Beyond English: The Impact of Prompt Translation Strategies across Languages and Tasks in Multilingual LLMs

Despite advances in the multilingual capabilities of Large Language Models (LLMs) across diverse tasks, English remains the dominant language for LLM research and development. This has led to the widespread practice of pre-translation, i.e., translating non-English task prompts into English before inference. Selective pre-translation, a more surgical approach, focuses on translating specific prompt components. However, its current use is sporadic and lacks a systematic research foundation. Consequently, the optimal selective pre-translation strategy for various multilingual settings and tasks remains unclear. In this work, we aim to uncover the optimal setup for selective pre-translation by systematically assessing its use. Specifically, we view the prompt as a modular entity, composed of four functional parts: instruction, context, examples, and output, either of which could be translated or not. We evaluate pre-translation strategies across 35 languages covering both low and high-resource languages, on various tasks including Question Answering (QA), Natural Language Inference (NLI), Named Entity Recognition (NER), and Abstractive Summarization. Our experiments show the impact of factors as similarity to English, translation quality, and the size of pretrained data, on the model performance. We suggest practical guidelines for choosing optimal strategies in various multilingual settings.