Ask any foreign language student about using online tools such as Google Translate, Bing Translator, Babblefish or any other translator and you’ll hear stories about how they got burned using these services. In other words, the translation was at best, incorrect, and at worst, offensive. Although these tools are becoming more intelligent leading to increased usage, they’re only good for casual translations and not suitable for full-fledged documents.

The basic reason for this problem is that machine translation uses data gathered over time to produce a statistical equivalent of what the best translation may be, thus leaving lots of room for error. What happens is that smart machines use the best possible match, which may or may not be right. When it comes to translations for business purposes, you cannot afford to have major mistakes to happen. Translation by humans trained in your target language, is a better choice because people can ascertain certain nuances that machines cannot. Following are areas where machine translations fall short.

Word Choice

The world’s major languages have different active word counts. For example, English has more than 171,000 active words, while Spanish and French have only 100,000. Conversely, Chinese has 340,000 while Japanese has 600,000. Words in some languages can have multiple meanings, resulting in confusion. For example, the Spanish word “intoxicado” is often used to describe someone who is drunk or intoxicated. It also has several other meanings, including one that indicates “poisoning.” Human translators can understand context, something that machines cannot, thus avoiding poor word choices.

Medical and Legal Writing

Both have highly specific vocabulary that requires higher comprehension. In addition, legal requirements differ from country to country, even in places that have the same language such as Germany and Austria. Live translators trained in such languages are aware of the differences and will work accordingly.

Culture Differences

Machines are unable to ascertain cultural requirements. Latin American countries value charisma, while Germans place an emphasis on efficiency that is not seen anywhere else in the world. People in South Korea want research and negotiation. If your translations don’t reflect these requirements, your business can ultimately fail under these circumstances.

Clearly, in more ways than one, machines fall short when it comes to translation. A machine’s “best guess” at what a phrase means will almost never come close to a human translation, because each act of communication is as unique as the person that spoke or wrote it in the first place, and the cultural context it came out of. Machines treat language as a science based on data, while a highly trained human translator has the ability to look at the true meaning behind the words. We may get to a point where machine translation can come close to human translation, but for now, there’s simply no substitute.

This post is written by Arlene Miles, a content writer with Ulatus.