Opinion A translation revolution for an inclusive, prosperous India
Manish Sabharwal and Arunava Sinha write: A translation revolution is blurring aesthetic, linguistic and political borders. This will raise inclusiveness, prosperity and soft power
It’s unfair to talk about translation without dealing with the elephant in the room — English as a link language, scale tool, and software vehicle. (Illustration by C R Sasikumar) 
-
Share -
-
-
In 1835, Thomas Babington Macaulay wrote, “A single shelf of a good European library is worth the whole of native literature in India”. This view reflected a colonialist’s need to create legitimacy for invasion but also captured his inability to understand a country that, even 187 years after his racist quip, has 22 official languages, newspapers in 35 languages, and speaks 1,200 languages. Our linguistic diversity is a gift but keeps a treasure chest of Indian knowledge locked. We believe the translation revolution underway will generate more output by India@100 than in the 75 years since Independence. This revolution — driven by policy, technology, philanthropy, and universities — will expand the global knowledge base, unlock Indian treasures for every Indian, and raise the share of the internet in Indian languages.
India’s linguistic diversity matters. Gandhiji massified our fight for freedom by organising the movement’s operating units around language rather than British administration units like Madras Presidency, United Provinces, Bombay Presidency, etc. The contentious language debates between the 299 remarkable people who wrote our Constitution captured the imagination of India as a nation, articulated regional identities, and created unity of purpose in founding modern India. The world’s largest democracy has been created on the infertile soil of the world’s most diverse society. But translations provide more glue; imagine if a webinar in Hindi could be heard live by a participant in Tamil. Imagine a food wholesaler receiving a WhatsApp message in Kannada written by a farmer in Bengali. And imagine if a book published in English could simultaneously be available in 22 Indian languages.
The first phase of Digital India’s unique open architecture ecosystem is now shifting from digital transactions — identification (Aadhaar), payments (UPI), vaccination certification (CoWIN), documents (Digilocker), toll collection (Fastag), education (NDEAR), health (NDHM), and taxation (GSTN) — to the second phase of using data to build artificial intelligence. A pioneer is Bhashini — under the National Language Translation Mission of the Ministry of Electronics and Information Technology — a translation ecosystem that aligns central ministries, state governments, big tech companies, start-ups, publishers, universities, NGOs, and citizens. Early results are exciting: 100+ models have been uploaded into Bhashini’s ULCA (Universal Language Contribution API), Bhasha Daan (creating datasets by crowdsourcing) has begun, and this model of collaborative AI is a global first.
Bhashini is catalysing responses. The AI for Bharat (AI4Bharat) Centre at IIT Madras, launched last week with support from Rohini and Nandan Nilekani and Microsoft, aims to “bring parity with respect to English in AI technologies for Indian languages with open-source contributions in datasets, models, and apps”. Contributions include IndicBERT (language model in 12 languages), IndicTrans (translation model used by India’s Supreme Court), IndicXlit (transliteration model in 20 languages), IndicWav2Vec (speech-recognition model) and IndicBART (language generation model). The Centre for Translations at Ashoka University brings together student translators and experts from 28 states and 300 cities in Indian languages to build a national library of literature in translation and aims to “replace global prejudices, stereotypes and monolithic views of Indian knowledge with plurality”. Additionally, non-profit foundations are testing translation fellowships. A guesstimate of “knowledge” non-fiction books suggests a stock of 5,000+ books from an Indian language into English, a stock of 750+ books from one Indian language to another Indian language, and a flow of 200 books a year. Innovation and resources are needed to tackle the pending 25,000+ books.
It’s unfair to talk about translation without dealing with the elephant in the room — English as a link language, scale tool, and software vehicle. In 1919, Gandhiji wrote an article in Young India suggesting real education was impossible through a foreign medium. B R Ambedkar supported English adoption in the Constituent Assembly debates because it was equidistant from all communities and would blunt traditional advantages. One of us who works in jobs believes English is a vocational skill that creates labour mobility, wage premiums, and resume signalling. But any conversation about English in India must embrace multilingualism. The most interesting poets of Hindi and Urdu at Allahabad University — Harivansh Rai Bachchan and Firaq Gorakhpuri — were both professors of English Literature. It is also churlish not to acknowledge the impact of the recent Booker Prize (the English translation of Hindi novel, Ret Samadhi, by Geetanjali Shree) and commercial success (the English translation of Kannada novel, Gachar Gochar, by Vivek Shanbag) as useful in raising the viability of regional publishers (they had become printers), regional language writers (they have bigger playgrounds), and regional language translators (they can do it full-time).
There is a long tradition of writers thinking about translations’ art, craft, and ambiguities. A 1941 essay by Vladimir Nabokov identifies three sources of evil in translation — ignorance, laziness and prejudice. A 1997 essay by Steven Rendall thinks a work’s translatability depends on an adequate translator and its essence allowing translation. An excellent new book, Translating Myself and Others, by Jhumpa Lahiri believes that many Indians are exposed to translation as children (in her case, using mom or ma for a card), draws upon Ovid’s Greek myth of Echo and Narcissus to explore the distinction between translating and writing, and has a beautiful reconsideration of the words original, authentic, and authorship. Pushback on the limitations of translation by software based on uniquely human skills and emotions must be acknowledged. We agree that for books, this technology will assist translators, not replace them.
Linguist Claude Hagege suggests languages are not a collection of words but living, breathing organisms holding the connections of a culture. Himachal Pradesh’s 16 languages have 200 words for snow, including one that means ”falling when the moon is up”. Increasing the quantity, quality, and speed of technologies and people that translate our languages are India’s infrastructure for inclusiveness, prosperity, and soft power. Thankfully, policy is catalysing the partnerships between technology, philanthropy, and academia that are fuelling an overdue translation revolution.
The writers are with Teamlease Services and the Centre for Translation, Ashoka University respectively
-
Share
-
-
-
Share
