A Computational Approach for Halegannada to Hosakannada Translation Using Dictionary Mapping and API Integration

Halegannada, an archaic form of the Kannada language used in historical literature from the 9th to 14th centuries, poses significant challenges for translation into modern Hosakannada and English due to its linguistic divergence and lack of direct support in conventional translation tools. This paper proposes a novel multi-stage pipeline to address this gap, enabling the translation of Halegannada texts for contemporary audiences. The approach begins with digitizing a traditional Halegannada-to-Kannada dictionary into a JSON format, creating a structured dataset for word-level mappings. A Python-based system then processes Halegannada input by mapping words to their Hosakannada equivalents and performing an initial English translation using a Python translator library. To overcome the incoherence of word-by-word translations, an API-based online translator refines the raw Hosakannada text into grammatically coherent English sentences, which are subsequently re-translated into Hosakannada. Additionally, the system generates multiple translation variations to enhance comprehension. Experimental results demonstrate the pipeline’s effectiveness, producing meaningful translations for sample Halegannada sentences, such as converting "ನಾನು ಗ್ರಾಮದಲ್ಲಿ ವಾಸಿಸುತ್ತ ೇನೆ" into "I live in the village" (English) and its refined Hosakannada equivalent. This work contributes to the digital preservation of Halegannada literature, making it accessible to modern Kannada speakers and English readers, and lays the foundation for future advancements in low-resource language translation.