Google Translate received a big upgrade on Thursday and will now support an additional 110 languages. The new languages were added with the assistance of artificial intelligence (AI). The tech giant used its in-house PaLM 2 large language model (LLM), and it was particularly effective in learning those languages that were related to each other or were a regional dialect of a language. Among the new additions, Indian regional languages Awadhi and Marwadi have also been added to Google Translate.
Google Translate gets 110 new languages
In a blog post, the company said it is the largest expansion Google Translate has ever received. Earlier, the translation tool supported a total of 133 languages. After the expansion, the number goes up to 243 languages. In November 2022, Google announced its 1,000-language initiative. Under the initiative, the company committed to building AI models to support the 1,000 most spoken languages around the world. The 110 new languages are part of that initiative.
To choose the new languages, Google prioritised the most commonly used varieties of each language. Citing an example, it said when considering the language Romani which has many dialects across Europe, the AI model picked the Southern Vlax Romani language, which is a commonly used variety online. The choice was also dictated by the availability of text on which the AI model could be trained.
Some of the notable new additions are as follows. Cantonese has finally been added. Google says it was one of the most requested languages for Google Translate. Afar, a tonal language spoken in Djibouti, Eritrea, and Ethiopia was also added after the company found volunteer community contributions. Notably, one-fourth of the total new languages are from Africa.
