Share on linkedin
Share on twitter
Share on facebook

Machine translation: Does it really have a place in website localization?

Share on linkedin
Share on twitter
Share on facebook

Literal word-for-word translations gone wrong or just plain gibberish – machine translation gets its fair amount of negative press time and time again. 

Like you, I’ve certainly had my fair share of terribly returned translations when using machine translation providers and it can quickly become embarrassing if you’re using said translations to navigate your way around a new country… 

So, why then do we at Weglot believe there’s a place for machine translation when it comes to even more complex translation projects such as website translation. Why do we welcome the use of machines in the localization process and ultimately how does Weglot solve some of the drawbacks machine translation can have? 

To understand that we need to take a look at some of the common misconceptions related to machine translation:

  • Accuracy
  • Machine translation is Google Translate  
  • Machines can’t think 
  • Time 
  • Expertise 
  • Context  

Then, see what role machine translation can actually play in making your website multilingual. 

Accuracy

Undoubtedly, accuracy is likely the first doubt people have when you mention using machine translation in website localization. 

The thing is, its accuracy really depends on the text and languages you’re translating into. It can produce good results for simple text in common languages, but do badly for complicated text in less common languages. 

You also need to consider what the text is being used for. If it’s a simple product description, machine translation can work pretty well. If it’s a more integral piece of content on your website, such as your homepage, you’ll likely want to send the machine translation for a final review, whether that be by you, your team or a professional. 

But, overall, accuracy isn’t a problem when you couple it with the post editing features a translation solution like Weglot can provide. Using machine translations as a way to give you a first layer of translation makes it an ideal way to kickstart a website localization project.

Machine translation is Google Translate

This is a common one and one we hear often at Weglot. Machine translation has wrongly been associated with Google Translate – it’s probably the first machine provider you think about yourself and rightly so, it’s the best-know tool. 

A few people even go as far as to think Weglot is just like Google Translate. Spoiler alert, it’s not. Yes we use machine translation providers for a first layer of website translation – but we don’t just use Google Translate. 

At Weglot, we constantly carry out performance tests on the leading machine providers: DeepL, Google Translate, Microsoft/Bing Translate and Yandex (alongside other platforms), in all language pairs, to make sure that we’re delivering the most up-to-date, natural translations to our users.

Secondly, and importantly, translation is only one part of the website localization process. Weglot also handles how you actually display those translations. Plus, those translations can also be manually edited so you never have to rely fully on machine translation providers. Google Translate is a machine translation provider and that’s about it. 

Machines can’t think 

Whilst it’s true that machines can’t think (not yet anyway 👀) what they can do is learn. Machine translation providers, such as Google, DeepL and Microsoft are powered by large amounts of data

What that means they can leverage the millions of interactions and conversations happening on their platforms in hundreds of languages, daily. 

They provide translation based on real conversations instead of pre-input dictionaries. Of course the dictionaries still serve their purpose, but it’s the learning from bilingual texts that’s helping them get the nuances of human language.

This then brings us back to the first point about accuracy. The better machine translation providers learn, the more accurate they are. We only need to look back a couple of years to see how far machine translation has come.

Machines are learning fast and whilst they continue to improve, you can leverage those learning capabilities within website localization. 

For example, translation memory. Weglot saves small sentences of similar content across your website so you don’t need to make the same manual edits time and time again. 

Lose time 

Losing time is most certainly a machine translation myth. The whole purpose of machine translation within website localization is to use it to speed up the manual process of translating thousands of words – not to mention it’s more cost effective! In fact, it’s generally how pro translators actually work

It’s estimated that a pro translator can translate 2000 words a day, whereas machine translation can generate thousands of words a minute. When you look at it like that you can see how manual translation simply cannot compete. 

It doesn’t mean we don’t see the enormous benefits of post editing, whether by a professional or not. Machine translation for speed simply means you can put professional translators time to better use. They can concentrate on adding more value to your machine translated text. It means a website translation project can go from months of work, to just days depending on how many edits you want to make. 

It’s not an expert 

Ah, well yes, but was it ever pretending to be? Whilst it’s true machine translation will give you a pretty like-for-like return, it’s the job of you, your team or a professional translator in that specific field to bring the expertise. 

Not everything needs to be machine translated if you have a broad range of content that needs a different kind of expertise. For example, your terms and conditions – it probably makes more sense to send to translators that deal specifically with legal documents. 

But, remember, If you’re handling a website localization project with Weglot, you don’t actually need to use the first layer of machine translation we give you. You’re free to import your own professionally translated text or add a pro translator from the Weglot Dashboard. 

Expertise can always be enhanced by humans. 

It doesn’t understand context

Context and emotions are generally harder for machine translation to understand. It means that they won’t pick up if something is a joke or whether the direct translation will remain culturally appropriate. 

It’s not really a question of machine vs human translation. It’s more a case of understanding where the strengths and weaknesses lie on this particular point.

For example, machine translation is probably going to a good job of translating text without emotion, such as a manual, set up guides or product descriptions. Whereas a marketing text where the original content was specific to your target market, will be enhanced by the input of a human. 

It also can’t understand things like brand names, for instance if your company name was Pear, it’s going to give a direct translation in your target language. 

To echo previous points, post editing is always going to be beneficial here – especially if you’re translating very creative website content, or as mentioned above, something that might be culturally sensitive. 

When it comes to issues such as brand names – using a translation solution that utilizes machine translation and post editing features – gives you the capability to amend this without having to manually adjust every single place your brand name is mentioned. Benefit from Weglot’s glossary feature so you can make ‘translation rules’ such as ‘never translate’. 

Website localization and machine translation: can it work? 

The long and short answer will always be a firm yes here at Weglot. And that’s mainly because we’ve seen thousands of our users quickly launch into new markets because of it. 

It’s probably time we took a step back and cleared our skepticism towards machine translation. 

Yes, if you run your whole website through machine translation and expect it to be perfect, you’re more than likely to be a little disappointed because it’s still working on humor, slang, idioms, but that’s ok. 

Is it a powerful tool that will save you time, money and resources? Yes. Our stance here at Weglot is to use its many benefits to your advantage and then work on perfecting it afterwards with pro translators or the help of your team (if that’s what your brand needs). 

For those needing to launch a website localization project on a tight deadline, using Weglot as your translation solution means you can leverage machine translation from the best providers on the market: DeepL, Microsoft, Google and Yandex. 

There’s absolutely nothing stopping you launching with machine translation, keeping an eye on your sales in your new markets and making translation adjustments where needed later down the line. 

Conclusion. Machine translation is the perfect way to kickstart your website localization project. 

About the author
READY TO GET STARTED?
The best way to understand the power of Weglot is to see it for yourself. Get your website multilingual live in minutes.
Subscribe to our newsletter

Stay up to date with the latest Weglot news and tips!