Our latest news & views
Using automatic machine translation (MT) tools to translate your website’s content will reduce your website’s visibility, and damage your brand’s reputation. It’s quite a statement, but the reasons are clear.
Recently published statistics state that the e-tourism sector in France – worth an estimated €18.5 billion – is losing more than €120 million annually due to the incorrect use of automatic translation tools.
When multilingualism is fundamental to sales success and budgets are tight, some companies seek out automatic machine translation tools, such as Google Translate or Bing, to translate their web content for free. Understandable perhaps, but reaching your international customers this way will not only damage your reputation, it will also reduce your website’s visibility. Here’s why:
The use of automatically translated web content is against Google’s Webmaster Guidelines, pure and simple.
The search engine does not deem raw translated texts from Google Translate or any other automatic machine translation tool as high quality content, and it will penalise your site for using them. When this happens, your website could be unranked for months, and your customers will have difficulty finding you.
When a website is translated accurately by a human translator, and exists in two different language versions, Google makes comparisons between the two and stores the correlation to improve the accuracy of Google Translate, little by little. Google senses auto-generated content because it creates a dangerous feedback loop, endlessly recycling its own translations, making the service less and less accurate over time. That’s exactly why it doesn’t even give you a chance – as soon as you publish automatic machine translation content as your own content, Google will stop indexing the site.
In case you didn’t already know, Google is the world’s most powerful search engine with more than one billion monthly visitors. Bing ranks second (350 million), and its guidelines require ‘rich and engaging’ content, which rules out anything automatically generated. Yahoo!, which ranks third (300 million users), has very specific guidelines about not using automatically generated content. Businesses ignore these search engine rules at their peril.
Accessible MT tools, such as Google Translate, Bing and Reverso, appeal to many as fast and cheap alternatives to hiring translators or using a translation agency. But they are openly and honestly designed for shorter texts. The translations they provide are intended for informational purposes only, not for publication, and the tools’ developers make no pretence about this.
Many of these basic automatic translation systems do not apply traditional grammatical rules, but rather statistical analysis. This form of automatic MT is based on information theory (a branch of applied mathematics, electrical engineering and computer science) and is well known for having major shortcomings and limitations when it comes to translation quality.
Word-by-word and sentence-by-sentence translation models struggle with idioms and nuances of speech, resulting in unnatural translations with a purely superficial fluency. These systems are simply not designed for translating longer texts.
In addition, statistical automatic MT does not work well between languages that have significantly different word orders, which complicates translation at the best of times. Theoretically, there are six possible language word orders, and most of the world’s languages use one of three:
Automatic machine translation tools simply cannot overcome these word order complications, and this explains why better performance is achieved when the source language is English and it’s being translated into other European languages.
In addition, when considering keywords for international SEO, there is often very little correlation between key search terms in English and those in foreign languages, so by directly translating what you assume to be a common and effective keyword, you could be including something which is rarely searched for.
Example: If a Chinese website wanted to sell an umbrella to an English-speaking market, it could end up marketing a ‘shelterwood’ or a ‘bumbershoot’ instead (try it on Google Translate). It doesn’t need to be said how infrequently those terms would be used when looking for an umbrella to buy.
Having a human translator is so valuable when dealing with any serious translation project – no machine will ever be able to match the idiomatic complexity and cultural knowledge of a native language learned slowly over a lifetime. That said, there are alternatives to consider if you’re on a particularly tight budget.
If cost is forcing you into using automatic MT, it is advisable to hire a proof-readers to check your translated copy after it has been through the machine translation system. This popular and effective route is called “machine translation with human post-editing” and is an excellent option.
There’s no doubt about the power of machine translation when used correctly. Here at Brightlines we know that when it counts, there’s no substitute for a human translator – for you, your reputation, and your business.
Editor’s note. This article was first published in 2014. In November 2016 the article was completely revamped and updated for accuracy and comprehensiveness.