When Bikram Biruly returned home after completing his PhD last month, residents of Matkam Sahi, a Ho Adivasi village in Odisha’s Mayurbhanj district welcomed with traditional dama, dumaṅ and mandar drums, songs and dance. It was a felicitous reception to a 29-year-old working to preserve the Ho language using AI, significant for a tribe long battling threats to its cultural identity.
Bikram, who traces his roots to Jharkhand’s Kolhan region, earned his PhD at Odisha’s Kalinga Institute of Social Sciences (KiSS) in September, the first Ho researcher to use AI to preserve the language, which, according to the 2011 Census, is spoken by approximately 1.4 million people, most of them in Jharkhand.
The Ho community has for long been demanding inclusion of the Ho language in the Eighth Schedule of the Constitution, which would grant it official protection. The Eighth Schedule currently includes 22 languages recognised by the government.
The Ho Adivasis settled in Matkam Sahi, Bikram’s village, in the 1970s when the Icha-Kharkai Dam under the Subernarekha Multipurpose Project displaced thousands in Jharkhand’s Kolhan region.
Bikram completed his MSc and PhD at KiSS, specialising in Indigenous Knowledge and Science and Technology — a programme that blends traditional wisdom with modern science. He says his interest in linguistics grew “naturally” as part of his indigenous upbringing. “When I took my cows grazing in the jungle, I would take my chalk and slate along to study my mother tongue,” he said.
But things weren’t easy. He says members of the Ho community were often bullied for speaking the language: “A popular saying in many parts of Odisha is, ‘Ulloh ku jete sijheile sijhibo nahi, Kulho ku jete bujheile bujhibo nahi’ — meaning, just as uloh, a hard-to-boil vegetable, never softens, no matter how long you cook it, the Ho will never understand what you say no matter how much you try.”
Ho is part of the Austroasiatic language family, which includes Vietnamese, Khmer and Santhali. Closely related to Mundari, it is also spoken in parts of West Bengal, apart from Odisha and Jharkhand. In 2011, it received the ‘second official language’ status in Jharkhand.
Story continues below this ad
Bikram worked to preserve the language even before he got his PhD. He first taught the Ho Warang Kshiti script locally, then launched a web portal, Ho Samaj Live. In 2019, Tata Steel Foundation honoured him with the ‘Top Tribal Youth Change-Maker’ award at the Samvaad Tribal Conclave.
For his PhD research, Bikram trained AI using datasets in three categories: Automatic Speech Recognition (ASR) to transcribe spoken Ho into text; Entity Recognition to classify words like names, places and dates; and Part of Speech (POS) tagging to understand grammar.
“I sourced a large dataset — 100,000 sentences — from open repositories like IIT Madras, primarily in Odia, Hindi and English, and converted around 20,000 sentences into Ho using the Warang Kshiti script,” he says.
With his PhD now complete, Bikram plans to focus on implementing his model and hopes to approach the Ministry of Tribal Affairs to include Ho on digital platforms like Adi Vaani and Google Translate.
Story continues below this ad
“Adi Vaani is an app made by the government for tribal languages where individuals can write, translate and listen to programmes in Gondi, Bhili, Mundari and Santhali. I want to include Ho in such apps to digitally store our language,” the young scholar says.
