Machine Translation: Overview

There have been great improvements in Machine Translation. Many people have used tools such as Google’s translate when attempting to read text in a foreign language. The mobile version of this has also come in handy when one visits a foreign country. This is made possible by using advanced algorithms. The two best-known algorithms are the rule-based methodology and the probabilistic approach.

The rule-based method uses a list of quite comprehensive linguistic rules for the sentence structure of each language and another set of rules which combine the two sentences together. Probabilistic Machine Translation uses statistical theory to link source and target text. For each sentence in the source language, there are any number of sentences in the target language with some being more likely than the others. Those with the most probability of being correct are then chosen.

Despite the advances that have been made in MT, the technology still falls behind the accuracy of human translation. This is because Machine Translation places too much emphasis on the words in a sentence rather than on the meaning behind the text. The statistical algorithms behind machine translation don’t apply language specific grammatical rules. These machine translations can’t understand particular neologisms and nuances of speech. There is any number of examples where MT has led to humorous results. For companies and websites that have a reputation to uphold in a global market, these mistakes can be costly.

Hence, when it comes to translation of languages, context is king. Positioning a product to a consumer in another language requires an understanding of the target audience and the meaning of the original message. Up until now, there are still no machines which can accomplish this task effectively.

Machine Translation and SEO

There is, however, an even more important consideration when dealing with online content. This comes down to the impact of Machine Translation on your search engine rankings. When it comes to optimizing your content for a foreign market, one has to focus not just on the correct translation but also on appropriately localizing it. Foreign markets have unique keywords that differ from a literal translation of the word. For example, ‘coche’ is the Spanish term word for «car». Yet, in some Latin American countries, it refers to a baby stroller. Hence, your “car” keyword strategy should be adjusted for your target country.

It is clear, therefore, that Machine Translation is unlikely to help your SEO strategy in other languages and markets. An even more important consideration is whether using Machine Translated content in your website will actually harm your rankings. For example, in the video below by Google’s Matt Cuts, he explains that search algorithms may interpret MT content as auto-generated content.

In fact, according to Google’s definition of «Automatically Generated Content», “Text translated by an automated tool without human review or curation before publishing” is listed as an example. This implies that your content may be considered as spam by the search engines.

One can understand Google’s aversion to presenting Machine Translated content. Their search algorithms rely on high quality, well worded content. Similarly, Google translate itself relies on quality translations in order to improve it’s translation ability. It degrades the effectiveness of the service if they are constantly recycling their own translations. Hence, Googlebot will try to avoid your content (or site even) if it suspects any Machine Translation.

Moreover, Google emphasizes the importance of content that adds value to website users. Poorly translated text can negatively affect your visitor’s experience. Apart from harming your rankings, negative user experiences harms your brand and may drive potential users, customers and business partners away.