Machine translation (MT) is a hot topic within the localization world. The improvements in MT have been impressive and it has become a more common method of translating content. But is machine translation always appropriate and should it be used as frequently as it is? Is machine translation the way forward or a black-hole?

One of the great things about working for a marketing localization agency is that you gain real insight into all aspects of localization. Be it global PPC campaigns or translation strategy for multilingual websites. Part of our localization/translation strategy at Wordbank is the use of a multi-purpose translation process which includes machine translation and I am actually a fan of machine translation – it’s fast, cost-effective and relatively simple.

However, there are definitely do’s and don’ts to machine translation. Below are 3 reasons for and against using machine translation instead of human translation. I’ve tried to give a more marketing focused perspective than the normal cost vs. quality discussion and focus on the integration of what machine translation means for your company and your customer on the web.

For Machine Translation

1. It’s VERY Fast

Cost-effectiveness aside, one of the strongest elements to machine translation is speed. Machine translation allows you to upload multilingual content almost instantly. Daily posts, blog articles, news articles etc are not a problem because the content is fairly instantly available. (I should add that you should never use MT without a post editing stage but this is still a quicker process than 4-stage translation.)

But what does this mean for your brand and customers? Well, firstly you are up-to-date. That the message you are displaying is accurate for that moment in time and users know they are getting the latest information from your site – whatever their language. Secondly, we know that from a search perspective, Google’s ‘freshness’ update does mean that updated content is pushed higher in search engines. Not only does this give you an increase in your rankings but it also allows you to be part of an emerging search trend (e.g. reacting to news stories) and then puts your brand front of mind for that individual product or service.

2. It’s Evolved

Remember that for businesses and brands looking to utilise machine translation, this isn’t the same machine which is creating the hilarious Google translate mistakes we see dotted around the world. The evolution of machine translation has been one of the most impressive aspects of this localization form.

You can add translation memories and glossaries into a machine translation package. The premise is still a translation engine like Google but it is now layered with bespoke translations which are specific to holding on to brand identity or create part of your description for a product/service. These systems can be built into pre-existing translation tools like Studio which can then merge with a post-editing system. Essentially, it’s making the hilarious miss-translations a thing of the past. There is much more impressive tech stuff on MT but I’m not a tech person so let’s assume I’ve listed them here.

 3. It Runs off a Majority Wins Situation

Often perceived as one of the negatives of machine translation, the algorithms set up by MT technology like Google utilises a majority wins situation where the most common translation across the web (as crawled by Google) is used as translation. Why is this a good thing? Well if you are targeting a market, speaking in the tone which they most often use could create a positive response when users engage with your site.

Equally, it’s almost tried and tested content with the ‘one-person opinion’ translation over-ruled by the masses which will ultimately be your customer base. I would never say that this means machine translation should always win nor is it an accurate reflection of how your target audience engage and speak, but it is a good stepping stone for getting your English content there.

Against Machine Translation

1. Is it a good representation of your company?

Machine translation never flows as well as human translation. Despite its evolution, it will still translate the words rather than the concept you are trying to put across. This is an important element to think about when visitors first enter your site. If a customer searches on Google and finds your company, the first engagement with your company would be tainted by off-message copy and literal translation rather than brand development. And it’s also important to remember that customers don’t just enter via your homepage so this is something you should be thinking about at any point of your site.

I’m not saying that machine translation can never be used on any page on your website just in-case someone enters the site via this page but it’s worth looking at analytics and seeing what traffic your current pages are getting. The higher the relative volumes, the less you should consider machine translation.

 2. It doesn’t work in marketing

That’s right! Machine translation does not work on marketing collateral and nor should it. Your marketing collateral is creative, dynamic and forward thinking. Companies pay millions to develop a brand identity and to deliver a product/service in a compelling and idealistic fashion. Machine translation uses logic not creative thinking and this will significantly dampen any creative flair added to your marketing material. What a waste of money if you have spent time and effort to develop clever and decisive marketing content for it to be flattened by a machine.

In localization, we use brand guidelines, concept testing and transcreation to develop marketing collateral which continues to be creative and locally engaging but to hold on to a central brand identity. These are the type of steps taken to ensure that the time taken in your English marketing efforts is rewarded internationally.

 3. Doesn’t help in Search

Where are your visitors coming from? Are they mainly coming through search portals? If this is the case then machine translation is not the answer for you and very rarely can a brand ignore search as a visitor generation method.

Firstly there is the issue of keywords. Keywords should always be researched rather than translated to find what local customers are actually searching for. Don’ t make the mistake of thinking that the only difference between markets is language! These keywords need to be woven into your content both in meta-data and in-content itself.

Making your page targeted to the terms you are looking to appear for is just the start to SEO, but it is a very important step. Machine translation will straight translate your English optimized content, but the keywords will be translated, not researched so your content will most likely be irrelevant to the key search terms you are actually aiming for.

Secondly let’s also remember the Google/Yandex assessors. When you are targeting key search terms, these users will assess the quality of the websites which appear within that SERP and that quality is also reflected in the use of content. We know that Yandex takes a particularly dim-view of machine translated content in this assessment and Google is quickly taking the same stance. Poor assessment means a quick drop in rankings – which is extremely difficult to counter once this has happened.

Conclusion

There are hundreds of pros and cons for both machine translation and human translation approach. Ultimately it comes down to defining factors such as content use, target audience, sector etc. A good mixture of both generally creates an effective multilingual website which doesn’t cost the earth. But content is not one of the areas you always want to scrimp on. Remember 3 important rules to machine translation:

  • Do not utilise without a post-editing stage
  • Always remember the customer not the budget
  • Never use machine translation on a page targeting search

Do you agree? Do you think machine translation has a place as part of website localization?