Tools of Translation: Machine Translation

 

In the past, numerous jokes have been made about the awkward nature of machines, especially when they were used to simulate typical human actions. Robots were portrayed with clumsy movement and stilted speech; texts produced entirely by a computer were either nonsensical or unfocused. Even today, one can personally witness the difficulty machines have with human words and speech by selecting YouTube’s options for automatic transcription or translation. However, this clumsiness is becoming a thing of the past, as technological advances and increasingly refined data morph the capabilities of machines from science-fiction jokes to frighteningly useful tools. The translation industry is no stranger to this, where machine translation can potentially achieve results on par with that of a certified translation company in Singapore.

Though clients seeking guarantees of quality may still turn to humans, particularly for crucial work such as a notarized translation, one key reason for opting for machine translation would simply be volume, or rather the capability of converting a high volume of pages from one language to another very quickly. After all, it may be simple enough to provide a translation service for a dozen or so pages, but if the client requires the translation of hundreds of pages in a limited window of time, then it may be too unmanageable for only a few translators to do on their own, or it may be that the certified document translation services in Singapore necessary to finish that work before the specified deadline would be too costly for the client. Of course, clients aiming to seek out cheaper translation services in general is a key reason of its own.

On the other hand, the adage of “you get what you pay for” applies here as well. Machine translation programs may have advanced in quality as tools since their creation, but on its own, machine translation still has a reputation of producing noticeable errors as well. A computer may have whole encyclopedias of various languages in its database, but it may end up making obvious mistakes if it is unable to understand context like a human. For example, if a legal translation service is being provided, then it requires more formal language. Some machine translation programs use a brute-force method of running words through countless dictionaries for their equivalent in another language, but if they run into the names of peoples or places (such as in a PR application translation) and are not instructed to leave them alone, the program can make the grievous error of translating a name into something unrecognizable.

Mind you, this is not making an argument against machine translation, as it is a tool just like any other. Rather, it is an argument against blindly using machine translation. A dictionary may know each word’s meaning, but it cannot put together a letter. It is still up to thinking humans to direct the meaning of the text, just as it is still up to a certified translation professional in Singapore to compensate for what a machine translation program lacks. This process of machine-translated results being later reviewed and rectified by a human translator has a specific name: Machine Translation Post-Editing.