Google Translate Expands with 110 New Languages Including Punjabi, Cantonese, and More

Google Translate just got a whole lot more multilingual! In its largest expansion ever, Google has added support for 110 new languages, bringing the total number of supported languages to over 220. This update furthers Google's commitment to breaking down language barriers and making communication more accessible for everyone.

Harnessing the Power of AI

At the core of this expansion is Google's powerful PaLM 2 large language model. This AI model helps translate learn new languages and improve translation accuracy, particularly for languages with fewer speakers or limited training data. This is a significant step forward in Google's 1,000 Languages Initiative, their ambitious project to build AI models that can translate the world's 1,000 most spoken languages.

Reaching New Communities

The new languages encompass a diverse range of communities, from major world languages like Cantonese to those spoken by small Indigenous groups. This update represents a significant expansion of African languages supported by Google Translate, including Fon, Kikongo, Luo, and Wolof. Overall, these new languages represent over 614 million speakers, making communication and information access easier for a significant portion of the global population.

Highlights of the New Languages

• Afar: This tonal language spoken in Djibouti, Eritrea, and Ethiopia saw a high volume of contributions from volunteers during the development process.
• Cantonese: A highly requested language, Cantonese presents a unique challenge due to its overlap with Mandarin in written form.
• Manx: A Celtic language of the Isle of Man, Manx saw a near extinction but has been undergoing a revitalization effort in recent decades.
• NKo: This standardized form of West African Manding languages uses a unique alphabet and is actively supported by researchers and language communities.
• Punjabi (Shahmukhi): Written in the Perso-Arabic script, this variety of Punjabi is the most spoken language in Pakistan.
• Tamazight (Amazigh): This Berber language of North Africa has multiple dialects but a standardized written form supported by Google Translate in both Latin and Tifinagh scripts.
• Tok Pisin: An English-based creole language and the lingua franca of Papua New Guinea, Tok Pisin offers a glimpse into a unique linguistic blend.

Addressing Language Variation

When adding new languages, Google Translate considers regional variations, dialects, and different spelling standards. For instance, the Romani language translations blend Southern Vlax Romani with elements from other dialects to offer the most commonly used forms.

Source

Image Credits - Google