“Machine translation”. Translators shudder to hear those words! It is partly in disgust, due to a firmly-held belief that a computer will never replace a superior human translator (like us!), partly because we are scared stiff that it will! So we either vehemently deprecate machine translation, or we carefully skirt around the subject and hope, for example, that our customers won’t find out about the Serbian-English-Serbian translation tool, recently made available for free by the almighty Google (link below)!
Because the fact is that Google’s translation tool, which now provides automatic translation into English of Serbian websites and of copy-pasted blocks of Serbian text, is really surprisingly good (we will not discuss Google’s English-Serbian translation tool in this article, i.e. the reverse Translate to English direction, as it is pretty awful right now)!
Rather than acting as if it didn’t exist, we think it is better to get this subject out in the open and examine its implications for the clients of translation companies and for the translation industry in general. So this will be the first in what is planned to be a series of articles looking at automatic and machine translation, both in the context of Serbian-English translation and of translation in general. In this article we will look briefly at the quality of Google’s automatic Serbian-English translation and explain why we do not think translators and translation companies working in the Serbian-English pair should be too concerned for their livelihoods right now.
An example of Google’s Serbian English translation
Let’s carry out a little experiment first. We took a paragraph of Serbian text (taken from a Serbian Wikipedia article) and pasted it into the Google Serbian-English translation tool.
A human translation from the Serbian to English would read something like this:
A translation memory is comprised of segments of text in the source language and of their translation into one or more target languages. These segments can be passages, paragraphs, sentences or phrases. Individual words are not handled by translation memories, these are dealt with by terminology bases. Research has shown that many companies using multilingual documents use translation memory-based systems.
Within a few seconds, Google Translate outputs the following translation into English:
Translation memory consists of segments of the text in the original language and their translation into one or more target languages. These segments can be passages, paragraphs, sentences or phrases. Individual words are not in the field of translation memory, but they deal with terminoloske database. Research shows that many companies have multilingual documentation systems used to translating memory.
Can you understand it? Apart from a few problems the translator had in identifying passive/active constructions and an unknown word, of course you can! It’s certainly a lot better than any Serbian-English machine translation tool we’ve tried before. If you look at what an old-style machine translation (which shall remain nameless) did to this paragraph, maybe you can begin to appreciate how good Google Translate is:
Prevodilacka store sastoji oneself off segmenata textual on izvornom jeziku too njihovog prevoda on unity whether over ciljanih jezika. Those segmenti might lie flinders,pasusi,recenice whether fraze. Pojedinacne reci did not of domenu prevodilacke memorije,vec oneself to them bave terminoloske baze. Istra%u017Eivanja pokazuju ought mnoge kompanije wo there are visejezicku dokumentaciju koriste sisteme with prevodilackom memorijom.
I beg your pardon? That was supposed to be English, in case you were wondering! And NO, we did not doctor this in any way! Also, if anyone can tell us what “flinders” are, then they know more Middle English than we do!
Google Translate is perhaps not as successful with all texts as it was with this one, but it is certainly a major improvement over the above example in practically all cases! So perhaps translators should think twice before discounting machine translation from Serbian to English (and other languages, if this is anything to go by).
What makes Google Translate different?
Google’s system is a little different to previous machine translations in that it uses a statistical method to analyse existing translations from Serbian to English and applies what it has learned to the new text. Old-style systems merely use a dictionary to translate texts word-for-word by “brute force” and tend not to be very successful. However, it should be noted that Google themselves have recognised that their statistical method has now hit a wall of diminishing returns and it is unlikely that, as the technology currently stands, the standard of translation will be able to improve appreciably, and that goes not just for Serbian and English, but for all language combinations.
Death-knell for human translators?
So are we crazy to tell you all this? After all, translation companies rely on the (paid) work of human translators! What happens if all your clients go off and begin using Google Translate free of charge? Indeed, we have already seen examples of amateur translators supplying “translations from Serbian into English” that have clearly been carried out using this tool! It is only a matter of time before translation companies begin receiving “previously-translated” texts (texts that suspiciously resemble Google translations!) from clients and being asked to “just proof-read this” for a rate considerably lower than a translation from scratch would cost.
Well, we would like to talk about a few reasons why you and your clients should know about Google Translate for Serbian and English and why we think translation companies need not fear for their business:
A translation business should value transparency and seek to work within the realities of the market – it does not make long-term business sense to “hide” valuable resources like this from our clients! Besides, they will find out about it sooner or later! Rather, we should accept the reality that tools such as this bring to the translation industry – the market will always be changing and we need to be prepared to adapt, not cling to an outdated reality.
We should want our customers to use Google Translate for Serbian-English translation! After all, the vision of a translation company should be to enable their customers to communicate with other markets and cultures. So if this tool helps a client who only understands English to understand a text in Serbian, then you have surely gone some way to achieving this vision!
But the core of the issue and the reason translation companies have nothing to “fear” from Google Translate is what you have been suspecting all along: computerized, automatic translation is not going to replace professional human translation from Serbian to English (or any other language) any time soon. Or let’s phrase it as a question: would you, as the marketing manager of, say, a Serbian company wanting to do business in the West, entrust the translation of your website or of your corporate magazine into English to a machine translation tool? The simple reality is that, no, you would not.
This is not necessarily to knock automatic translation tools – they are after all a soft target for us superior human translators! They may well have their applications, and we may discuss this in another article. This is merely to say that any business that is serious about a given market, given the current quality of machine translation, will settle only for a professional, human translation of their promotional materials. After all, we said Google’s Serbian English translation was good, but it’s not THAT good! In fact it’s not nearly good enough.
Perhaps in a future article we will also take a look at some of the differences between machine translation and human translation and investigate some of the reasons why, despite the remarkable advances, and the positive things we have said about Google Translate, automatic translation software is not currently a serious choice for professional translation – from Serbian to English or in any other language combination – and why it may never be. Indeed, we have some deep concerns about possible misuses of a tool like this, in an environment where even now translation is often not taken seriously enough.