Parallel Corpus for Indigenous Language Translation: Spanish-Mazatec and Spanish-Mixtec

本研究介绍了首个针对墨西哥两种土著语言Mazatec和Mixtec与西班牙语的平行语料库，旨在支持机器翻译任务。该数据集包含9799句Mazatec语和13235句Mixtec语，主要来源于宗教和宪法领域。数据集的创建过程涉及文本对齐和预处理，以确保数据质量。此数据集的应用领域主要集中在提升低资源语言的机器翻译性能，特别是在从西班牙语翻译到土著语言的方向上显示出较好的效果。

This study introduces the first parallel corpus between two indigenous Mexican languages, Mazatec and Mixtec, and Spanish, aiming to support machine translation tasks. The corpus consists of 9,799 sentences in Mazatec and 13,235 sentences in Mixtec, mainly sourced from religious and constitutional domains. The development of this corpus includes text alignment and preprocessing steps to guarantee data quality. The primary application of this corpus is to enhance the machine translation performance of low-resource languages, and it has demonstrated promising results especially in the translation direction from Spanish to these indigenous languages.