EthioMT: A New Parallel Corpus for Low-Resource Ethiopian Languages to Advance Machine Translation

Core Concepts

This paper introduces EthioMT, a new parallel corpus for 15 low-resource Ethiopian languages paired with English, and presents benchmark results for machine translation performance on this corpus using transformer and fine-tuning approaches.

Abstract

The paper discusses the creation of the EthioMT parallel corpus, which covers 15 Ethiopian languages from the Afro-Asiatic and Nilo-Saharan language families. It provides details on the languages included, their language families, number of speakers, and dataset sizes.
The authors collected datasets for the languages, primarily from religious domains, and aligned the sentences with their English translations. They then preprocessed the data and split it into training, development, and test sets.
To evaluate the usefulness of the new corpus, the authors conducted baseline experiments using two approaches: a transformer model and fine-tuning a multilingual M2M100-48 model. The results show that the fine-tuning approach outperformed the transformer model in both translation directions (from English to Ethiopian languages and vice versa). The performance was better for languages with larger dataset sizes, such as Amharic, Afaan Oromo, Somali, and Tigrinya, compared to languages with smaller datasets.
The authors conclude that the EthioMT corpus can foster collaboration and facilitate research and development in low-resource Ethiopian languages. They plan to expand the corpus size and explore additional machine translation approaches in the future.

Stats

Amharic has a dataset size of 1.5M.
Afaan Oromo has a dataset size of 2.9M.
Somali has a dataset size of 1.2M.
Tigrinya has a dataset size of 140K.

Quotes

"Recent research in natural language processing (NLP) has achieved impressive performance in tasks such as machine translation (MT), news classification, and question-answering in high-resource languages. However, the performance of MT leaves much to be desired for low-resource languages."
"Ethiopia is a country that stands out for its remarkable cultural and linguistic diversity, with over 85 spoken languages. Only a few languages of Ethiopia have received attention in the area of NLP research and application development. Most languages have been left behind due to resource limitation."

Key Insights Distilled From

EthioMT

by Atnafu Lambe... at arxiv.org 03-29-2024

https://arxiv.org/pdf/2403.19365.pdf

Deeper Inquiries

What other techniques or approaches could be explored to further improve machine translation performance for the low-resource Ethiopian languages in the EthioMT corpus

To further enhance machine translation performance for low-resource Ethiopian languages in the EthioMT corpus, several techniques and approaches can be explored:

Data Augmentation: Implementing data augmentation techniques such as back-translation, where the parallel corpus is translated back and forth between languages to generate additional training data, can help improve translation quality.

Transfer Learning: Leveraging pre-trained language models like BERT or GPT to initialize the translation model can aid in capturing language nuances and improving translation accuracy.

Domain Adaptation: Fine-tuning the translation model on specific domains beyond religious texts, such as legal, medical, or educational content, can enhance the model's ability to translate accurately in diverse contexts.

Hybrid Models: Combining statistical machine translation with neural machine translation techniques can potentially improve translation quality by leveraging the strengths of both approaches.

Active Learning: Implementing active learning strategies to select the most informative data points for manual annotation can help in expanding the corpus with high-quality translations.

Multimodal Translation: Integrating visual or audio inputs with text data for translation tasks can lead to more accurate and contextually relevant translations.

How can the EthioMT corpus be expanded to include more diverse domains beyond the religious texts currently covered

Expanding the EthioMT corpus to include more diverse domains beyond religious texts can be achieved through the following strategies:

Crowdsourcing: Engaging the local community to contribute translations from various domains such as healthcare, agriculture, technology, and culture can help diversify the corpus.

Collaboration with Institutions: Partnering with educational institutions, research organizations, and governmental bodies to access and collect texts from different domains can enrich the corpus.

Text Mining: Utilizing text mining techniques to extract and translate publicly available data from websites, social media, and other online sources can help incorporate diverse content into the corpus.

Professional Translation Services: Engaging professional translators to translate documents from different domains can ensure high-quality translations for inclusion in the corpus.

Government Initiatives: Collaborating with government agencies to access official documents, policies, and reports in various sectors can provide valuable content for the corpus.

What potential applications and use cases could the improved machine translation capabilities for Ethiopian languages enable, and how could they benefit the local communities

Improved machine translation capabilities for Ethiopian languages can enable various applications and benefit local communities in the following ways:

Enhanced Communication: Facilitating seamless communication between speakers of Ethiopian languages and other languages can promote inclusivity and understanding in diverse settings.

Educational Support: Providing accurate translations for educational materials can enhance learning opportunities for students who speak Ethiopian languages, thereby improving literacy and educational outcomes.

Cultural Preservation: Enabling the translation of cultural texts, folklore, and historical documents can contribute to the preservation and promotion of Ethiopia's rich cultural heritage.

Healthcare Access: Translating medical information and resources can improve healthcare access for speakers of Ethiopian languages, ensuring better understanding of health-related instructions and services.

Economic Growth: Supporting translation in business and trade contexts can facilitate international partnerships, market expansion, and economic growth for local businesses and entrepreneurs.

Legal Assistance: Offering legal translation services can aid in ensuring access to justice and legal rights for individuals who speak Ethiopian languages, promoting fairness and equity in legal proceedings.