So, if your goal is to truly understand local lingo and become a native speaker, you are better off taking time to learn the language rather than relying on Google Translate. On average, there are 2.77 mistakes for every 10 words analysed when translating everyday phrases native to each country, such as “Break a leg” and “Echar una mano”.Īs an Artificial Intelligence (AI) tool, it is difficult for Google Translate to understand the uniquely human and emotional nuances needed to convey the meaning of a region-specific phrase into the target language. Our experiment shows that Google Translate struggles to translate colloquial phrases, with only a 72% accuracy rate on average across all languages analysed, compared to 94% for business websites’ content and 96% for literary text. The average mistakes and readability score of Google translate across different types of texts: Type of text The Google translate experiment Lack of human understanding leads Google Translate to fail at translating colloquial phrases

To determine how accurate Google Translate is, we tested its ability to translate business sites, literary texts and everyday phrases – from English to Italian, German, and Spanish and vice versa – comparing Google’s translations with the know-how of true bilingual language experts. Yet, despite the improvements Google has tried to implement, can Google Translate really match the skill of bilingual speakers? We can’t be the only ones who’ve noticed questionable phrasing when trying to translate a website or even simple expressions… Since its launch in April 2006, Google Translate’s quality of translation has improved a lot thanks to advances in machine learning, such as the GNMT (Google Neural Machine Translation) model. Google Translate is the world’s number one translation software.

English to japanese translator google