The 50-ton elephant in any translator’s room these days is machine translation, or MT for short. Not that all translators are necessarily worried about it, but it’s a big beast that refuses to be ignored. And more importantly, it’s being embraced by the non-translation community, lovingly embraced in fact, which means linguists everywhere should get used to its presence and maybe even hop on to the castle on top and guide it.
First, let’s look at what MT is. Essentially, machine translation is the use of computer software to automatically translate from one language to another. Humans are not involved other than feeding in the text and walking away with the results. It is mostly commonly used on the Internet where users can translate web pages at the click of a button, or enter text into a box to get the results. It can also be integrated into other software or made into an app for mobile devices. The latest breakthrough is translation of the human voice, a form of simultaneous interpretation. MT promises speed, cost savings, ease-of-use and an exciting futuristic world in which we can all communicate freely in whatever language we want and still understand each other.
So far, so great. And I want to add that I am a huge fan of technological advances. I embrace almost all innovations and automation if it makes our lives easier, healthier or simply better in some way. It truly is amazing what technology can do these days and MT is another example of some remarkably advanced technology. Computer translation has evolved from simple word-for-word lookups, to analysis of complex phrases against previously translated content to come up with appropriate phrasing and even metaphors.
It is obvious to see why some translators would be worried. Robots are coming to take over our jobs! The same thing happened in the industrial revolution and the same thing is happening today: technology does, to some extent, replace humans. The fear is that MT will get so good that human translators will no longer be needed and the profession will actually die out. Certainly, there are those who are already turning to MT instead of paying a professional. Numerous websites are popping up that have been translated by a machine. Emails are often fed into Google Translate or Bing Translator to get a quick translation.
However, how many of you would actually say that these translations are any good? Certainly, in my experience, the results from MT range from mediocre to terrible. I’m not talking about simple phrases here, but a lengthy piece of content. It is still remarkable what MT can do, but it is still a long way off what a good professional translator can do. And the simple reason is that computers are still not smart or creative enough. They are « smart » in a different way. They are very smart at processing data and using various heuristic or algorithmic functions. They can draw on huge databases of text and work at lightning speeds. Huge strides have been made in artificial intelligence, but few would argue they are anywhere near humans in this regard. They are still a long, long way off the same level of critical thinking.
Nothing demonstrates this gulf better than a creative translation. Building layers of imagery, alliteration, connotation and humour into a marketing slogan requires a real wordsmith. How do you think MT would cope with this slogan for a fictional sandwich spread: « Such lickety-sticky, golden-gooey goodness that even the bread will propose a toast. » Okay, so maybe it won’t sell, but I have a deadline to meet and this is the best I can come up with on the spot. Let’s try Google, what do we get: « Une telle lickety collante, la bonté d’or gluant que même le pain va proposer un toast » in French. « Tal santiamén pegajosa bondad, de oro-pegajosa que incluso el pan va a proponer un brindis » in Spanish. Hmm, not sure if it’s quite so appetizing, but maybe the pun still works in French (assuming the Anglicism of « pain grillé » is acceptable)?
So, if a good professional translator can do a better job of translation, why is MT such a booming phenomenon? Well, of course the big draw is that it’s cheap, or free. Google and Bing’s online services don’t cost a cent if used on their webpages. But money alone is not the only factor, they have to offer acceptable results. And depending on what it is used for, there is a case to say that MT offers acceptable results. If you want the gist of a web page or an email, MT can do a good job. There will be many inaccuracies and often complete gibberish contained within the translation, but it should be enough for you to understand the essence of the content. MT is also useful for basic word-for-word translation and some common phrases. It’s on-demand nature should not be overlooked either. To be able to punch in a menu description on your smartphone when out at a restaurant and see a translation almost immediately is much more practical than sending out a request to a translation agency.
In our next blog, we will look at MT more in depth and see ways in which it can be useful to translators, what its limitations are and what the future may hold. It cannot be ignored that MT is improving all the time and is here to stay.