Google revealed on Thursday that powerful artificial intelligence (AI) algorithms were used to improve its Translate platform, adding 110 new languages for users.
The IT giant’s deployment of the PaLM 2 large language model (LLM) has enabled the largest expansion that Google Translate has ever experienced.
According to a press release, Isaac Caswell, senior software engineer at Google Translate, “from Cantonese to Q’eqchi’, these new languages represent more than 614 million speakers, opening up translations for around 8% of the world’s population.”
“Some are major world languages with over 100 million speakers,” Caswell said. “Others are spoken by small communities of Indigenous people, and a few have almost no native speakers but active revitalization efforts. About a quarter of the new languages come from Africa, representing our largest expansion of African languages to date, including Fon, Kikongo, Luo, Ga, Swati, Venda and Wolof.”
Google Translate’s announcement highlighted several noteworthy additions, one of which being Cantonese. Caswell stated that Cantonese has “long been one of the most requested languages” for the tool, but adding it was difficult because it frequently overlaps with Mandarin in writing, making it “tricky to find data and train models.”
Afar, a tonal language spoken in Djibouti, Eritrea, and Ethiopia, and Shahmukhi, a variant of Punjabi that is the most spoken language in Pakistan, were also added. The release mentioned that Afar had the greatest volunteer community contributions.
The Isle of Man’s Celtic language, Manx, was also included when its final native speaker passed away in 1974, almost bringing the language to extinction. Thousands of speakers now exist on the island as a consequence of a revival campaign.
Caswell pointed out that since Tok Pisin is an English-based creole, app users who speak English should try translating into it since they “might be able to make out the meaning!” Tok Pisin is the primary language of Papua New Guinea. It was added to Google Translate.
In an effort to fulfill its previously stated 1,000 Languages Initiative, which commits the business to developing AI models supporting the 1,000 most widely spoken languages in the world, Google said it intends to gradually add more languages to Translate. According to the business, that process will accelerate even further when AI technologies like PaLM 2 develop.
“PaLM 2 was a key piece to the puzzle, helping Translate more efficiently learn languages that are closely related to each other, including languages close to Hindi, like Awadhi and Mardwadi, and French creoles like Seychellois Creole and Mauritian Creole,” Caswell stated.
“As technology advances, and as we continue to partner with expert linguists and native speakers, we’ll support even more language varieties and spelling conventions over time,” he said.