"Machine translation". Translators shudder to hear these words! This is partly disgusting because it is firmly convinced that the computer will never replace an excellent human translator (like us!), Partly because we are afraid that he will threaten us! So we strongly reject the translation of the machine or carefully cover the subject and hope that our customers will not be aware of the Serbian-English-Serbian translation tool recently owned by the Almighty Google (free below)!
Because the fact that Google's translation tool, which now automatically obtains English-language texts on Serbian web pages and blocking Serbian texts, is really surprisingly good (we will not discuss the Google English-Serbian translation tool in this article, that is, in the reverse direction, as it is pretty awful now)!
Instead of acting as if it did not exist, we think it is better to keep this subject open and examine its implications for the computing companies and the translation industry. So this is the first to be considered as a series of articles that look at automated and machine translation in both the Serbian-English translation and the overall translation. In this article we will briefly review the quality of Google's automatic English translation and explain why we do not think translators and translation companies in the Serbian-English pair need to worry too much about their livelihoods.
An example of Google's Serbian translation
First, do a little experiment. We made texts in Serbian (from a Serbian Wikipedia article) and inserted into Google's Serbian translation tools. The translation from Serbian to English reads like this:
Translation memory consists of the source language text segments and translation in one or more target languages. These segments can be sections, paragraphs, sentences, or phrases. Some words are not translated by the translation memories, they are handled by terminological basics. Research has shown that many companies using multilingual documents use translation memory-based systems.
Within seconds, Google Translator translates the following translation into English:
consists of parts of the original language and translation into one or more target languages. These segments can be sections, paragraphs, sentences, or phrases. The words are not in translation memory, but they deal with terminology databases. Research has shown that many companies have a multilingual documentation system for rewriting memory.
Can you understand? In addition to some problems, the translator had to identify passive / active constructs and an unknown word, of course! This is certainly much better than any Serbian-English translation tool we've tried before. If we look at this old type of machine translation (without a name) this paragraph, you may begin to estimate how good Google Translate is:
Prevodilacka stores itself from the segment's text source beyond these translations its unity, even more than the target language. Those segments can be like pastes. Individuals say they do not use interpreting memories, but they deal with terminology bases themselves. Istra% u017Eivanja show that many companies have visejezicku documentation that use the translation memory.
Do you ask for mercy? You should have been English if you were curious. And no, we did not handle it in any way. Furthermore, if anyone can tell us what the "flinders" are, they know more about Middle English than we do!
Google Translator may not be as successful with all texts as this, but there is certainly a significant improvement in the above example in virtually every case! Thus, translators may have to think twice before abandoning machine translation from English to English (and other languages if it is anything). it differs little from previous machine translations by using a statistical method to analyze existing translations from English to English and apply what you learned to the new text. Old-style schemes use only one dictionary to translate the words "raw power" literally and are generally not very successful. It should be noted, however, that Google themselves have recognized that their statistical methods are now facing a wall of declining yields and it is unlikely that, according to the current state of the technology, the standard of translation may be noticeably improved and this will not only go to Serbian and English for all language combinations.
Is death for human translators?
So are we crazy to say all this? After all, translation companies rely on the paid job of human translators! What happens if all of your customers start out and use Google Translator for free? Indeed, we have already seen examples for amateur translators who "translate English into English" and have clearly done this with the tool! It is only a matter of time for translation companies to start "previously translated" texts (texts that are suspect to Google translations) from customers and are asked to "read this"
Well, for some reasons we want to talk that you and your clients should know about Google Translator in Serbian and English and why we think that translation companies do not have to be afraid of their business:
Maybe a future article will look at the differences between machine translation and human translation and examine the reasons why the progress and the positive things we've said about Google Translator are remarkable, translation software is currently not a serious choice for professional translation – from Serbian to English or other language combinations – and why it can not. Indeed, we are deeply concerned about the potential abuses of such devices in an environment where the translation is still not taken seriously.
In the meantime, however, look at the device and maybe open a whole new world of Serbian Internet content that you could not reach before. Try Google's Serbian-English-English-Serbian translation here .
Supported by sbobet