Google Translate has announced the addition of 110 new languages to its platform, including South Africa’s TshiVenda and siSwati. This expansion, part of Google’s 1,000 Languages Initiative, aims to support the 1,000 most spoken languages globally through the use of AI models. The latest update utilizes the PaLM 2 large language model and follows a previous expansion in 2022 that introduced 24 languages using Zero-Shot Machine Translation.
Focus on African languages
The expansion prominently features African languages, with about a quarter of the newly supported languages coming from the continent. TshiVenda and siSwati join Afrikaans, English, Sesotho, Sepedi, Xhosa, XiTsonga, and Zulu as the South African languages now available on Google Translate.
Read: Google’s electricity bill pushes greenhouse gas emissions up by 48%
Global reach
The newly added languages, spoken by an estimated 614 million people, represent around 8% of the global population. Among these are Fon, Kikongo, Luo, Ga, and Wolof, languages primarily spoken in various African countries.
The PaLM 2 language model has been instrumental in enabling this extensive addition of languages. Google’s approach prioritizes the most commonly used varieties of each language, considering factors such as dialects and specific spellings to provide relevant translations.
The update also includes languages with unique cultural significance, such as Afar, Cantonese, Manx, NKo, Punjabi (Shahmukhi), Tamazight (Amazigh), and Tok Pisin. Each language presents distinct challenges and contributions, reflecting the diverse linguistic landscape.
Future developments
Google plans to continue collaborating with linguists and native speakers to further enhance its translation capabilities. The goal is to support more language varieties and spelling conventions as technology advances, improving the accuracy and usefulness of Google Translate for a broader audience.