Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multilingual Prompt Engineering in Large Language Models: A Survey Across NLP Tasks

Shubham Vatsal, Harsh Dubey, Aditi Singh

TL;DR

This survey provides a structured overview of multilingual prompt engineering for large language models across 30 NLP tasks and 250 languages. It introduces a taxonomy of 39 prompting techniques, emphasizing zero-shot and few-shot settings, and maps them to SoTA results across datasets such as MGSM, XCOPA, MLQA, XL-Sum, and WinoMT. The analysis highlights cross-lingual strategies (e.g., Translate-En-CoT, XLT, CLP, CLSP), dictionary-augmented prompts (CoD, DIPMT), and multi-step methods (InterCPT, DecoMT, TALENT) that improve cross-language reasoning and translation in low-resource contexts. It also investigates language-family distribution and resource-level coverage, identifying underrepresented languages and suggesting directions for more inclusive, globally representative NLP research. Overall, the paper provides a comprehensive framework and actionable insights for researchers and practitioners aiming to enhance multilingual LLM performance through prompting alone, without extensive fine-tuning.

Abstract

Large language models (LLMs) have demonstrated impressive performance across a wide range of Natural Language Processing (NLP) tasks. However, ensuring their effectiveness across multiple languages presents unique challenges. Multilingual prompt engineering has emerged as a key approach to enhance LLMs' capabilities in diverse linguistic settings without requiring extensive parameter re-training or fine-tuning. With growing interest in multilingual prompt engineering over the past two to three years, researchers have explored various strategies to improve LLMs' performance across languages and NLP tasks. By crafting structured natural language prompts, researchers have successfully extracted knowledge from LLMs across different languages, making these techniques an accessible pathway for a broader audience, including those without deep expertise in machine learning, to harness the capabilities of LLMs. In this paper, we survey and categorize different multilingual prompting techniques based on the NLP tasks they address across a diverse set of datasets that collectively span around 250 languages. We further highlight the LLMs employed, present a taxonomy of approaches and discuss potential state-of-the-art (SoTA) methods for specific multilingual datasets. Additionally, we derive a range of insights across language families and resource levels (high-resource vs. low-resource), including analyses such as the distribution of NLP tasks by language resource type and the frequency of prompting methods across different language families. Our survey reviews 36 research papers covering 39 prompting techniques applied to 30 multilingual NLP tasks, with the majority of these studies published in the last two years.

Multilingual Prompt Engineering in Large Language Models: A Survey Across NLP Tasks

TL;DR

This survey provides a structured overview of multilingual prompt engineering for large language models across 30 NLP tasks and 250 languages. It introduces a taxonomy of 39 prompting techniques, emphasizing zero-shot and few-shot settings, and maps them to SoTA results across datasets such as MGSM, XCOPA, MLQA, XL-Sum, and WinoMT. The analysis highlights cross-lingual strategies (e.g., Translate-En-CoT, XLT, CLP, CLSP), dictionary-augmented prompts (CoD, DIPMT), and multi-step methods (InterCPT, DecoMT, TALENT) that improve cross-language reasoning and translation in low-resource contexts. It also investigates language-family distribution and resource-level coverage, identifying underrepresented languages and suggesting directions for more inclusive, globally representative NLP research. Overall, the paper provides a comprehensive framework and actionable insights for researchers and practitioners aiming to enhance multilingual LLM performance through prompting alone, without extensive fine-tuning.

Abstract

Large language models (LLMs) have demonstrated impressive performance across a wide range of Natural Language Processing (NLP) tasks. However, ensuring their effectiveness across multiple languages presents unique challenges. Multilingual prompt engineering has emerged as a key approach to enhance LLMs' capabilities in diverse linguistic settings without requiring extensive parameter re-training or fine-tuning. With growing interest in multilingual prompt engineering over the past two to three years, researchers have explored various strategies to improve LLMs' performance across languages and NLP tasks. By crafting structured natural language prompts, researchers have successfully extracted knowledge from LLMs across different languages, making these techniques an accessible pathway for a broader audience, including those without deep expertise in machine learning, to harness the capabilities of LLMs. In this paper, we survey and categorize different multilingual prompting techniques based on the NLP tasks they address across a diverse set of datasets that collectively span around 250 languages. We further highlight the LLMs employed, present a taxonomy of approaches and discuss potential state-of-the-art (SoTA) methods for specific multilingual datasets. Additionally, we derive a range of insights across language families and resource levels (high-resource vs. low-resource), including analyses such as the distribution of NLP tasks by language resource type and the frequency of prompting methods across different language families. Our survey reviews 36 research papers covering 39 prompting techniques applied to 30 multilingual NLP tasks, with the majority of these studies published in the last two years.
Paper Structure (79 sections, 9 figures)

This paper contains 79 sections, 9 figures.

Table of Contents

  1. Introduction
  2. Focus of Our Survey
  3. Literature Search Process
  4. Outline
  5. Prompt Engineering Techniques
  6. En-Basic/Standard/Vanilla/Direct Prompting
  7. Native-Basic/Standard/Vanilla/Direct Prompting
  8. X-Basic/Standard/Vanilla/Direct Prompting
  9. Native-CoT
  10. En-CoT
  11. Translate-En-CoT
  12. Translate-En
  13. Cross-Lingual-Thought Prompting (XLT)
  14. Cross-Lingual-Prompting (CLP)
  15. Cross-Lingual Self-Consistent Prompting (CLSP)
  16. ...and 64 more sections

Figures (9)

  • Figure 1: Taxonomy Diagram of Prompt Engineering Methods and Languages Applied Across Different Multilingual NLP Tasks
  • Figure 2: NLP Task Distribution by Language Family
  • Figure 3: Number of Prompting Techniques by Language Family
  • Figure 4: Number of Papers by Language Family
  • Figure 5: Language Coverage by Language Family
  • ...and 4 more figures