It has been 60 years since IBM sang the praises of the first machine capable of performing translations. It was called the701 computer. Its first major translation achievement was translating Russian sentences into English. This success led researchers to believe that soon machines would be doing all the German English translations and every other conceivable language too.
However, IBM was first to admit that a sentence or two of translation needed 2 ½ times the instructions inputted into a computer than those required to guide a missile. Google uses vector space maths, which is anything but simple, but Google Translate seems so straightforward and fast when applied. However, this type of machine translation is not perfect enough to replace the face of German English translation companies.
For most professional English German translation jobs, there is more involved than knowledge about the 2 languages. Translators often have to create an appropriate language in order to get out the message required. They try to craft precise sentences so that the text has the same impact in the translated language as the source language. Some of the best translators create the best translation because they have been studying the two languages side by side for decades. A machine translator does not have this wealth of experience to draw upon..
The diversity of a language shows its colours when a text is translated by a hundred human translators as there is little doubt that none of these translations will be identical. Machine German translation tools have up to now indicated that their output is far more limiting for the requirements of most translation tasks.
In any language one word may have a multitude of meanings and each word’s context draws out the meaning. According to Peter Gilliver, an Oxford English Dictionary lexicographer, the word “run” has up to 645 quite distinct meanings. For a machine German English translator to be as reliable as a human translator it will have to be able to reproduce all the different meanings as required, otherwise it’s not paying for itself. Word-for-word translation is not possible as it doesn’t show the relationship between the words. Even humans create special word relationships that machine translators could never keep up with.
A business knows that words are important when marketing a product. The voice of a brand is dependent on the words that are chosen to represent it. How would even the best professional translation company that uses machine translation know which words or phrases on a possible list would best fit a business’s customised language?