Integrating artificial intelligence with a morphology-based digital corpus framework for Toba Batak language preservation

This study proposes a conceptual and computational framework, supported by a pilot implementation, for integrating artificial intelligence (AI) technologies with a morphology-based digital corpus to support Toba Batak language preservation. As a low-resource agglutinative language with complex morphological structures, Toba Batak faces significant challenges in digital processing due to its productive affixation system and morphophonemic variations. This study adopts a qualitative conceptual approach based on linguistic analysis and computational modeling, complemented by a small-scale pilot experiment using a rule-based morphological analyzer. The findings indicate that the complex affixation patterns and morphophonemic characteristics of Toba Batak create significant computational challenges, highlighting the need for structured morphological representation within AI-based language systems. The proposed framework applies linguistic knowledge through a hybrid approach that combines finite-state transducers with neural architectures for applications such as automatic speech recognition (ASR), machine translation (MT), text-to-speech (TTS), and conversational agents. This study demonstrates the potential of morphology-based digital corpora to support the development of more linguistically informed AI technologies for low-resource languages. The proposed approach provides a scalable and adaptable foundation for future AI-assisted preservation and revitalization efforts for Toba Batak and other morphologically complex endangered languages.