Google Translate Adding 100+ New Languages Thanks to AI

Google today announced that it is adding support for a whopping 110 new languages to Google Translate, marking the service’s largest expansion to date.

The move was made possible by Google’s PaLM 2 AI model, which helped Translate learn these new languages. PaLM 2 was especially adept at teaching Translate languages closely related to each other — Awadhi and Marwadi, for example, which are close to Hindi.

Back in 2022, Google added 24 new languages to Translate thanks to Zero-Shot Machine Translation, a method where a translation model learns a new language without being trained on any examples. However, the latest expansion far dwarfs that with its 110 new languages.

The new languages available in Translate include Cantonese, which Isaac Caswell, a senior software engineer working on Google Translate, noted “has long been one of the most requested languages” for the machine translation service.

“From Cantonese to Qʼeqchiʼ, these new languages represent more than 614 million speakers, opening up translations for around 8% of the world’s population,” explained Caswell.

“Some are major world languages with over 100 million speakers. 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.”

Other languages now supported by Google Translate include Afar, Manx, NKo, Punjabi (Shahmukhi), Tamazight (Amazigh), and Tok Pisin, among many others.

According to Caswell, Google’s strategy in choosing between the varieties that many languages tend to have thanks to regional differences, dialects, and different spelling standards has been to prioritize the most commonly used varieties of each language.

Caswell went on to add that Google Translate is committed to supporting even more language varieties and spelling conventions in the future. Google previously announced its 1,000 Languages Initiative, which seeks to train AI models that will help support the 1,000 most commonly spoken languages around the world.