CMEDIA: Tech giant, Google has reportedly for the first time added Canada’s most widely spoken Indigenous language to its translation service.
Google Translate, which translates text, documents and websites from one language into another has added Inuktut, a broad term encompassing different dialects spoken by Inuit in Canada, Greenland and Alaska.
As part of Google initiative to develop a single artificial intelligence language model, the latest addition is to support 1,000 of the most spoken languages in the world.
With roughly 40,000 Inuktut speakers in Canada as suggested by Statistics Canada’s data, the criteria of the number of speakers alone is not enough to determine whether a language can be included in Google Translate, said Isaac Caswell, a senior software engineer with the platform.
With sufficient online text data being a prerequisite to pull from to create a language model, other Indigenous languages in Canada had “simply too little data to have any usable machine translation model,” said Caswell.
Spoken by more than 86,000 people in Canada, Cree was also considered by the engineers for addition to Google Translate Service, but found fewer websites in the language to pull from.
“We don’t want to put anything on the product which just produces broken text or nonsense,,,Inuktut really stands out…lot of clean and a lot of well written data, because, I think, the community is increasingly online,” said Caswell.
When adding a language to Google Translate, the tech company looks at two main things: a desire or need from the community and its technical feasibility.
After determining if Google’s model could recognize Inuktut, it began to consult with language speakers and organizations.
Reaching out to Inuit Tapiriit Kanatami, the national organization that represents about 70,000 Inuit in Canada, Google wanted to ensure the validity of the Inuktut language model for development, including the ability to translate both of the language’s writing systems.
Inuktut uses qaniujaaqpait, or syllabics, and qaliujaaqpait, based on the Roman alphabet.
Having developed its own data set of common characters that can be used to write in any dialect of Inuktut, Inuit Tapiriit Kanatami wanted to ensure ease of written communication among the different Inuit regions.
“If we hadn’t had their help, we would have just been able to launch in syllabics, which undermines some of their current work,” said Caswell.
Welcoming Google’s work to include Inuktut, and citing the need to revitalize, protect and promote Inuit languages, the organizations said,
“This is another way in which to make our language relevant, easily accessible…to be able to interact with it…is reconciliation in action…to keep our language strong and to celebrate our language.,” Natan Obed, president of Inuit Tapiriit Kanatami, said in an interview.
Aiming to represent a group often overlooked by the tech sector. the introduction of Inuktut, Google says.
“I hope, maybe if anything…a little bit more seen by a big tech (company)…Indigenous communities have had a lot of experiences being overlooked by technology,” said Caswell.
Having the ability to translate written Inuktut to English and vice versa. Google Translate’s users would also be provided with other options, including the verbal translation tool, which may come at a later time, said Caswell.
Caswell added that although the promotion of Indigenous languages with the use of AI is not without its limitations, this will change with improved technology as more and more languages are unlocked.
