In a major breakthrough for language preservation and digital inclusion in Nepal, Kathmandu University’s Information and Language Processing Research Laboratory (ILPRL) has officially launched the beta version of a Google-funded trilingual machine translation system. The new AI-powered tool enables translation between Tamang, Nepali, and English, marking a historic moment for one of Nepal’s most widely spoken indigenous languages.
What Is the Google Tamang–Nepali–English Translation System?
The Tamang–Nepali–English Trilingual Machine Translation (TMT) System is now publicly available in beta at
https://tmt.ilprl.ku.edu.np/.
This AI-based platform uses advanced neural machine translation models to support seamless multilingual communication.
The system currently supports the following translation pairs:
- Tamang ↔ Nepali
- Nepali ↔ English
- Tamang ↔ English
The models have been trained on more than 23,000 parallel sentences covering critical sectors such as agriculture, healthcare, education, culture, and everyday communication. The project is open for public testing, allowing users to submit feedback to improve accuracy before the final release.
Google Research Award and Kathmandu University Collaboration
This initiative is funded under the Google Academic Research Award 2025 (Society-Centric AI category), where only 30 projects were selected globally. Officially titled
“Empowering Information Access Rights: Developing a Trilingual Machine Translation System for English, Nepali, and Tamang”,
the project is led by expert researchers at Kathmandu University.
Why Tamang Language Translation Matters in 2025
The Tamang language is spoken by approximately 1.5 million people, representing about 5.6% of Nepal’s population according to the 2021 census. Despite its cultural significance, Tamang faces increasing challenges due to the dominance of Nepali and English in education, governance, and digital platforms.
This new Tamang AI translation system helps address these challenges by:
- Improving access to health, education, and agricultural information
- Reducing language barriers in public services
- Supporting e-governance and digital inclusion in Nepal
- Preserving indigenous languages in the digital era
Community Involvement and Future Expansion
According to Principal Investigator Prof. Balkrishna Bal, the project is now entering a community-driven phase. Planned initiatives include workshops, training programs, and the launch of a public Corpus Annotation Tool to crowdsource linguistic data from Tamang speakers worldwide.
The research team aims to expand the dataset from 23,000 to 100,000 parallel sentences, significantly improving translation accuracy.
Will Tamang Be Added to Google Translate?
One of the most anticipated outcomes of this project is the potential integration of Tamang language support into Google Translate. Researchers are actively working with Google toward this goal, which would bring global visibility and long-term sustainability to the language.
As Prof. Bal notes, “This will be a major step toward internationalizing the Tamang language.”
A Model for Indigenous Language AI Worldwide
With thousands of indigenous languages at risk of extinction, the Google Tamang machine translation system serves as a powerful example of how artificial intelligence can promote linguistic diversity and social good.
If you are a Tamang speaker, linguist, or technology enthusiast, visit the beta platform, try the Tamang to English translator, and contribute feedback to help shape the future of inclusive AI.
Frequently Asked Questions (FAQs)
Is Tamang available on Google Translate?
No, Tamang is not yet available on Google Translate. However, researchers are working with Google to enable future integration.
Who developed the Tamang translation system?
The system was developed by Kathmandu University’s Information and Language Processing Research Laboratory (ILPRL) with funding from Google.
Where can I test the Tamang–Nepali–English translator?
You can test the beta version at https://tmt.ilprl.ku.edu.np/.
How accurate is the Tamang AI translation?
The system is currently in beta and trained on 23,000+ sentences. Accuracy will continue to improve through user feedback and expanded datasets.
Sources: Techpana, Kathmandu University ILPRL announcements, and the official TMT project website.
