For a long time now, there have been rumors that machine translation will replace human translators, and sometimes we hear statements like, "Translations done by humans and Google Neural Machine are nearly indistinguishable," when Google announced the launch of its neural machine translation system (GNMT). Of course, neural networks have made huge strides in their development and are becoming increasingly integrated into everyday life, but has artificial intelligence really solidified its place in the translation arena to the extent that it can replace humans?
Yes, time does not stand still. Globalization processes connect people, regions, cities, and countries into a single network where anyone can access information located at another point on the globe (of course, if they have paid for the internet). People are increasingly captivated by foreign cultures, traditions, and particularly literature, often in the original language; typically, they receive such information processed and translated into a comprehensible language by relevant communities or groups, public pages, or news sites. However, there are times when information arrives in its original form, like a tome in the original language, but unfortunately, a person does not always have a translation of that tome (new literature is emerging so rapidly that it can be challenging to translate everything, with popular works prioritized), and they may lack the skills to read and understand what is written in the book. In this case, they have a few options: wait for an official translation (and if the work is not popular, that wait could be long), wait for an amateur translation (yes, there are indeed brave souls who take on such work), or use makeshift tools, like Google Translate.
The first two paths are similar, as you rely on human labor; the second one is a bit more questionable, but not every official translator is good, so let's conditionally combine them into one. The second path is far less viable, although some people are already prepared to perceive it as a finished and final product, and this poses a greater threat than the quality of the machine translator itself, which is convenient as a tool intended to ease the routine work of a translator, but nothing more. To not yield to this 'opponent,' which is supported primarily by those lenient toward the quality of translation, you must adhere to the following principles, which will be outlined below.
1. You translate the meaning of the text, not the words. If I don't understand it — I don’t translate.
The machine operates according to algorithms. And these are indeed complex interlingual algorithms utilizing dictionaries and grammatical rules, it must be acknowledged. But! Translating text is not merely translating words from one language to another; it is a much more complicated process. A significant drawback of machine translation is that it cannot grasp the meaning of the text.
Therefore, Translator-Human, develop your knowledge of the language being translated, up to the level of idiomatic phrases, proverbs, and sayings, and phraseological units. Meaning is the main and foremost thing you must grasp from the text!
2. Learn your native, dear to your heart, great and mighty Russian language. The translation must fully adhere to the norms of the language into which it is being translated, in our case, Russian.
Yes, I consider this point just as important as knowing the foreign language from which the translation is done. It is often the case that those who take up the craft of translation make mistakes in their own... When chaos and disorder reign in your own home, how can you go to someone else's house and teach its owners how to maintain order? Correct answer: you can't.
I generally advocate for domestication in translation strategy, and therefore I believe that any attempts to present cultural differences in the text through means not characteristic of the Russian language are local forms of *-mania, where instead of the asterisk, you could substitute, for example, gallo- or anglo-, and so on. Of course, a certain range of words, like titles specific to the country (wali, shah, king, etc.), forms of address (mister, sir, gentleman), can be transformed, but this would be unreasonable.
Love your language. Cherish it.
And no matter what the experts say about preserving the cultural nuances of the text, the main thing in the text is its plot, characters, emotions, and meanings, and you can understand the cultural environment through other means, for example, by learning the original language. The translator is needed to convey the text in a format accessible to the reader, namely, in their native language.
3. Don't be afraid to transform foreign text.
I won't delve into translation theory, but there are several specific transformation techniques in translation. Additional elements can be added, omitted, or rearranged in the translated text—all determined based on an analysis of the text being translated, which also implies a good native foundation. By the way, in this regard, machine translation falls far behind human translators. The machine translates 'as is', while a person can decide 'what's best' and act accordingly.
4. And lastly, be patient and diligent.
For translating text is very hard work, requiring a lot of energy and time, as well as knowledge, broad horizons, and the ability to adapt.
As for me, I translate from Japanese, which presents me with a number of additional barriers, and it certainly doesn't make life easier for the machine translator, as pattern recognition for Eastern languages plays a very important role. But during the time I've spent translating foreign texts, I have developed the aforementioned four principles that distinguish translation as Translation, rather than a simple copy of the foreign text, and which I believe are unique in any case, whether it's Japanese or English, for example.
And, to sum up, in what exactly does a translator not yield to a machine?
A person will not concede to a machine translator in the ability to understand what is not explicit — meaning. The machine understands words, combinations of words, grammar, and vocabulary; at times, it distinguishes homonyms, but it will not grasp meaning as a composite element of the text anytime soon. However, for a person to understand the meaning of the text, they must master their native language, while the reader must keep in mind that the result of machine translation can be quite far from the true meaning of the text.
Everything else, I believe, falls within the realm of common knowledge.
Source: habr.com
