Everyone is familiar with automatic translation through services such as Interglot, translate.nu and Google Translate. Here you enter words or short sentences to get a correct translation. These translations work fine for a word or short sentence, but it should not be really complex. Because with long sentences, it quickly leads to comical situations as soon as an automatic translator goes over it. The sentences then run smoothly, to say the least, not really well…
But.. this is going to change! Automatic translation is going to be the thing. In recent months, a major technological breakthrough has occurred in the world of computer translation. For example, it is already possible to correctly translate complex texts in more than ten language combinations. The driver of this innovation? google! The translation world is shaking. Should translators fear for their jobs? And do translation agencies still get assignments?
Translations by self-learning neutral networks
Until recently, Google worked with the so-called phrase-based method to compile the correct translations. Separate phrases were translated and pasted together on the basis of probability. This method is fine for translating simple texts, but less so for complex sentences. Earlier this year, Google therefore launched Neural Machine Translation: machine translations based on a self-learning neural network that view and translate sentences in their entirety. Our brains work in a similar way, and it’s the technology that’s being used for self-driving cars as well. Neural networks can provide particularly good translations. Errors in long, complex sentences are reduced by 50 to 85 percent!
Ideal for webshops
This new way of translating seems to be an absolute must for web shops, for example. Static or descriptive content can be offered quickly in multiple languages in this way. Ideal for players such as Amazon, who can immediately enter the Dutch market for a while. International e-commerce players can expand very quickly and enter new markets. But by no means all content can be translated automatically.
Translation agencies are not yet concerned and this seems justified. Because as soon as texts become creative, a computer cannot handle this. A computer simply cannot gauge sentiment very well. For example, when a press release is written in a promotional or humorous way, it still requires the human eye to translate it correctly. And this actually applies to all forms of creative writing. A also agrees translation agency from Amstelveen: “Translation is about understanding the purpose of the text. Does something need to be described, is it about persuasion or does a product need to be sold? Then this requires different ways of writing and thus translating. A computer simply cannot do this. ”
What is clear is that the balance between man and machine is constantly changing. Machines can already be used more than a year ago and so automatic translation is definitely here to stay.
(Visited 424 times, 1 visits today)