KozKreolMRU WMT 2025 CreoleMT System Description: Koz Kreol: Multi-Stage Training for English–Mauritian Creole MT

Mauritian Creole (Kreol Morisyen), spoken by approximately 1.5 million people world-wide, faces significant challenges in digital language technology due to limited computational resources. This paper presents "Koz Kreol," a comprehensive approach to English-Mauritian Creole machine translation using a three-stage training methodology: monolingual pretraining, parallel data training, and LoRA fine-tuning. We achieve state-of-the-art results with 28.82 BLEU score for EN → MFE translation, representing a 74% improvement over ChatGPT-4o. Our work addresses critical data scarcity through use of existing datasets, synthetic data generation, and community-sourced translations. The methodology provides a replicable framework for other low-resource Creole languages while supporting digital inclusion and cultural preservation for the Mauritian community. This paper consists of both a systems and data subtask submission as part of a Creole MT Shared Task.

• Publication year

  2025

• Venue

  Proceedings of the Tenth Conference on Machine Translation

• Publication date

  Unknown publication date

• Fields of study

  Not labeled

• Identifiers
  DOI 10.18653/v1/2025.wmt-1.92
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Findings of the First Shared Task for Creole Language Machine Translation at WMT25

  2025cited by this paper
• Limited-Resource Adapters Are Regularizers, Not Linguists

  2025cited by this paper
• Kreyòl-MT: Building MT for Latin American, Caribbean and Colonial African Creole Languages

  2024influential reference
• African Substrates Rather Than European Lexifiers to Augment African-diaspora Creole Translation

  2023cited by this paper
• CreoleVal: Multilingual Multitask Benchmarks for Creoles

  2023cited by this paper
• A Benchmark for Learning to Translate a New Language from One Grammar Book

  2023cited by this paper
• What a Creole Wants, What a Creole Needs

  2022cited by this paper
• KreolMorisienMT: A Dataset for Mauritian Creole Machine Translation

  2022influential reference
• LoRA: Low-Rank Adaptation of Large Language Models

  2021cited by this paper
• Multilingual Translation with Extensible Multilingual Pretraining and Finetuning

  2020cited by this paper
• Multilingual Denoising Pre-training for Neural Machine Translation

  2020cited by this paper
• Plurality and definiteness in Mauritian and Haitian creoles

  2019cited by this paper
• The Curious Case of Neural Text Degeneration

  2019cited by this paper
• (Preprint)

  2018cited by this paper
• Rapid Adaptation of Neural Machine Translation to New Languages

  2018cited by this paper
• Decoupled Weight Decay Regularization

  2017cited by this paper
• Transfer Learning for Low-Resource Neural Machine Translation

  2016cited by this paper
• Creolization in Anthropological Theory and in Mauritius

  2016cited by this paper
• chrF: character n-gram F-score for automatic MT evaluation

  2015cited by this paper
• The handbook of Pidgin and Creole studies

  2008cited by this paper
• Linguists' most dangerous myth: The fallacy of Creole Exceptionalism

  2005cited by this paper
• Bleu: a Method for Automatic Evaluation of Machine Translation

  2002influential reference
• Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem

  1989cited by this paper

• No citing papers are available for this paper.