Comparative Analysis of Large Language Models in Moroccan Darija Translation

The advent of large language models (LLMs) has significantly advanced the field of natural language processing, particularly in machine translation. However, translating dialects and lesser-studied languages like Moroccan Darija presents unique challenges due to their linguistic complexity and the scarcity of high-quality parallel corpora. This chapter explores the translation performance of four cutting-edge LLMs—GPT-4, Gemini 1.5 Flash, Claude 3.5 Sonnet, and LLama3-70B—in converting Moroccan Darija to English. By employing the Levenshtein Distance to measure the similarity between original and translated texts, the study calculates the accuracy of each model, providing a quantitative basis for assessing their reliability and consistency. The chapter delves into the mean accuracy and standard deviation of each model’s translations, offering a comprehensive evaluation of their performance. Through detailed analysis and visualization of the distribution of translation accuracy, the study highlights the strengths and weaknesses of each model, offering valuable insights for researchers and practitioners. The findings underscore the importance of selecting the right model for effective translation, contributing to the ongoing development of more accurate and reliable machine translation technologies.