In its recent tweet, Google Webmasters stated that multilingual pages shouldn’t be word for word translation. Some website owners believe that Google requires word for word translation when you use hreflang and make a multilingual website. That is absolutely not the case. In fact, it is best to translate the page naturally and not do word for word translation. Appoint a professional translator fluent in both languages to translate it in a natural way, easy for visitors to understand and engage with. That means some words might not be on the page at all but bad translation can have a catastrophic impact on your business and its sales.

What is Google Translate and how it works?

Today, businesses need digital content to be translated quickly, almost on the fly, and a single or even team of translators may not be able to provide acceptable turnaround times.

In an effort to accelerate the process, many businesses are turning to Google Translate for website translation. Google Translate is Google’s free tool that instantly translates words, phrases, and web pages between English and over 100 other languages. Google Translate uses a process called “statistical machine translation”. This means that Google computers generate translations based on patterns found in large amount of text. Instead of teaching their computers all the rules of a language, they let their computers discover the rules themselves. They do this by analysing documents that have already been translated by human translators and scanning these texts, looking for statistically significant patterns.

Because the Google Translate widget uses an algorithm that is based on probability, the system is most effective when used with smaller chunks of language to convey simple messages. This makes is fine for casual conversation, internal communications but not for technical documents or important emails due to of inaccuracies around grammar and sometimes meaning. Error in medical translation can cost lives.

Can businesses rely on Google Translate for accurate translation?

Using a translation method that relies on probability increases the chances for errors. The translated text will only be as good as the machine allows it to be, but never as good as human translation will make it.

An article for The Telegraph says that “humans have shorter attention span than goldfish, thanks to smartphones” – if your website content is hard to understand and not localised, your visitors will bounce off and go somewhere else. Can you risk that? Or is the initial investment worth the outcome?