How accurate is Google Translate? (2023)

Is Google Translate accurate enough to trust your company?

Simply. Immediate. Free. These are some of the reasons why Google Translate is one of themost popular translation tools. It definitely comes to the rescue when trying to navigate a foreign land.

However, it can be unreliable for translating complex sentences for one main reason: it is not mature enough to take into account all possible nuances and contexts. Still, the question arises: how accurate is Google Translate? And can you rely on that when translating your website?

To use:Are you ready to translate your website? Skip the article andstart your free trial.

How does Google Translate work?

Ever wonder how Google Translate built its translation database? ANDstrongly based on Europarl corpus, a human-translated collection of documents from sessions of the European Parliament. In addition, it is based on several digital resources and common language translations.

When it was first released in 2006, it usedautomatic statistical translationprovide the translated text immediately. But it quickly became clear that this would not work in the long run. Since it translated single words, it worked more efficiently with short sentences. But it started offering weird translations for longer, more complex sentences.

Google knew it needed to switch to machine translation (MT) technologies to improve its accuracy. 10 years later, in 2016, the company developed its own framework, Google Neural Machine Translation Technology (GNMT). The move made leaps and bounds in its algorithm and changed its approach to translation. Instead of translating each word, he lookedthe meaning of the whole sentence.

The result? Much more faithful translations, even considering the jargon and slang. The change was huge: it reduced translation errors inmore than 55%-85%in many major language pairs.

With this new learning system, Google Translate no longer uses English as an intermediary to translate any language. insteadtranslated directly between two languages. That is, it was translated from French into Japanese, not from French into English and then into Japanese. By eliminating the middleman, it worked faster, more efficiently and, above all, more accurately.

Is Google Translate correct?

As long as Google Translate is availablemore than 130 languages– making it a translation tool with the widest range of support – also varies in terms of accuracy rate. For example, as Spanish is one of the most popular languages, the translation accuracy is usually over 90%.

In fact, a 2014 study found that only Google Translate has this ability.57.7% accuracywhen used to translate complex medical terms. A 2021 study published byUCLA Medical CenterI found that google translatorkept the general meaning for 82.5% of the translations. But cross-language accuracy ranged from 55% to 94%.

Sometimes Google Translate's accuracy is surprisingly good. results ofour own study on the state of machine translationfor website translation showed that 10 out of 14 translation editors were positively surprised by the quality of the translations presented to them. That meansautomatic translation worked betterthan they expected.

One thing to note about the accuracy of Google Translate is that it works best when translating literary texts into English. The situation is different with colloquial expressions: he showed a 72% accuracywhen converting casual texts from English to other languages. Because of this, you can sometimes get weird translations that don't make sense when trying to translate everyday phrases into other languages.

Is Google Translate reliable for translating websites?

Of course, efficiency is the name of the game here, and not everyone has the resources or time to leave thousands of words in the hands of a human translator.This is where machine translation comes into play.

Machine translation has come a long way since its inception in the 1950sneural machine translation(NMT), has become more reliable than ever. However, being as fluid as speech, it is not always perfect.

English translation remains Google Translate's greatest strength. after aStallion 2013jwho evaluated Google Translate's accuracy in extracting data from languages ​​other than English, translated articles generally took longer to extract than articles in English.

In our study, one of the translation editors observed that when MT failed to recognize the context of a given term, it provided a generic translation. Incorrect translations produced due to lack of context. However, when enough context was provided, the translation turned out to be correct. But can't the same be said about human translation and the treatment of texts out of context?

The reason for all this is quite simple: the accuracy of Google Translate depends on it.how much data is available for the target language. Outside ofAlmost 60% of websites are in English, Google Translate has a lot of data to work with. Therefore, it has the highest accuracy for English language pairs. On the other hand, as only 2% of web pages are in Portuguese, Google Translate will struggle to provide a very accurate translation into Portuguese.

While a small percentage of websites may be available in a given language, that doesn't mean there isn't demand for it.73% of consumersThey prefer product reviews in their native language when browsing online. If you want to expand your business across borders, it's important to target native speakers in your target area.site locationThis will help you define your message and ensure nothing is lost in translation.

Essentially,Google Translate is only as good as its users, or at least those who improve the quality of their translation.. The more input, the better the output, so it's important to continue to feed tool articles from publications in languages ​​other than English. While it's a convenient base for translating your site, it works best when used with other translation tools and, of course, a human eye to check things out.

Are there more accurate translation tools than Google Translate?

Other popular translation tools are, for example,DeepL, Amazon Translator and Microsoft Translatorthat use machine learning to inform and improve their services. DeepL leverages Linguee's extensive database of hand-translated phrases, sentences, expressions and snippets developed by the companies themselves. Amazon's service is based on neural networks designed for translation between a source language and a target language. Like Amazon, Microsoft Translator relies on NMT for its translations.

Our study found that DeepL, which uses the same machine translation approach as Google Translate, had the fewest unacceptable translations into Italian (it-IT). But it also had the fewest contactless translations, meaning it didn't need a human to edit it for the same language. DeepL also has the fewest supported languages ​​at 28, but performed best in translations into Spanish (es-ES).

Meanwhile, Amazon Translate, ranked as the worst performing machine translation engine, outperformed its peers by delivering the most contactless translations from French (fr-FR). It also had the fewest unacceptable Simplified Chinese (zh-CN) translations. Their support is right in the middle in 75 languages.

Microsoft Translator, which supports the second most languages ​​at 111, showed very consistent performance. It performed well with German contactless translations, but also had the lowest number of Portuguese contactless segments.

All tools worked well with European languages ​​and provided high quality Arabic translations from a human editing point of view. brief,no translation software beats the other– work best when used together.

Should You Use Google Translate?

However, Google Translate is still a great tool that can save you a lot of time and money in translating your website. Accuracy largely depends on your intent, style, and the person receiving your message. So, if you just need to translate a few short and simple sentences, this will suit you perfectly.

Google Translate can be useful in these cases:

• Edit Low Visibility Content such as B. Site Footers
• Instructional and standard content such as B. Guide end users to important information to solve a simple problem
• Product reviews, as little high-quality, comprehensive feedback is expected.

On the other hand, Google Translate may not be suitable in these cases:

• Translate everyday phrases that require more context or that don't have an exact equivalent in your target language
• Use it like a dictionary to translate individual words as it may offer a different meaning than the author intended.
• Translate source text that may not have followed conventional grammar rules into your own language or may have a different syntax than the target language, for example B. Spanish to Korean.
• Translate sarcastic or ironic expressions

As in these cases the quality of the translation can make all the difference in the tone and delivery of the message, it would certainly be beneficial to incorporate professional translation into your website translation workflow.

The best of both worlds: machine translation and human editing

Google Translate is capable of many wonderful features, howevershines best when working with other translation tools and human editors.

According to our study, 99% of the translation work produced worldwide is not done by professional human translators. And, on average, only 30% of machine-translated content is edited. It's actually a huge asset for anyone looking to use machine translation. While a human editor is still needed to detect inaccuracies (artificial intelligence is not perfect), the use of translation software is very valuable. This means that the website translations done by MT are definitely usable and only require a few tweaks here and there.

Sounds like a lot of work, especially for small businesses. The good news? A website translation solution like Weglot makes this easy.

More than60,000 global brandsUse Weglot to translate your websites with the bestautomatic translation and machine translation. While the two seem similar, machine translation involves the entire workflow of proofreading, editing, optimizing the content for SEO, and then uploading the content back to the website. In this way, you are guaranteed to get accurate translations.

Weglot works by selectionthe most suitable MT enginefor a specific language pair to provide the most accurate result. Thanks toneural machine translation, Weglot instantly recognizes and translates any content on your website. You don't have to tediously go through each page and translate them all yourself. Best of all, it also shows these translations in different language versions of your website.

It's intuitive to use and lets you translate and publish your content right away. This makes it even easier to localize and optimize your content in over 100 different languages, even RTL like Hebrew and Arabic.

Curious to know how it works? Experience Weglots10 days free trialto see how easy it is to set up a multilingual website.