Download Brochure

The Scientific and Technological Research Council of Turkey (TÜBİTAK) announced on Wednesday that it has taken considerable steps toward the Turkish...


When you spend money on a localization project, you expect a level of translation quality that matches your investment. Clients have every right to be upset...

The Yiddish language is a wonderful source of rich expressions, especially terms of endearment (and of course, complaints and insults). This article is a...

Other blog entries...

Machine Translation Is Just Another Tool In A Translator’s Arsenal Of Modern Tools

05 October 2012 Friday

Many people reading these words will not remember it, but a few decades ago, the only tools that a translator was able to use while being engaged in a peculiar activity involving for example translation of a very complicated patent were: 1 – a typewriter, and 2 – a few specialized dictionaries, which by the time they made it from the printer to the translator’s desk were inevitably already obsolete.

Oh, I almost forgot – there was also the precious tool No. 3 in the arsenal of tools available to translators – the life-saving “whiteout”, the greatest invention since sliced bread as far as translators pounding the typewriter keys to produce their masterpieces were concerned.

Things started looking up for our profession around the year 1980 when personal computer with a printer replaced the traditional typewriter-whiteout combo. The cheapest PC model used to cost about two thousand dollars, but it came with a word processor. Back then when WordPerfect was the king of word processors, there were many word processors, before the wise heads at a cartel called Microsoft decided to get rid of them so that a couple of decades later, they could start selling Microsoft Word as a yearly subscription through the Internet tied to a single PC, which means that over a period of 10 years, they could easilly wring out more than a thousand dollars from a single user of a single piece of software that used to cost about one to three hundred dollars.

Will they get away with it? I would hope not, but the chances are that they probably will. In the meantime, I started learning OpenOffice in case I need to switch to it as I am looking at it now in a completely different light.

If you buy something digital in the modern world, something digital that can be delivered through Internet, something like software, or music, or a book, you don’t really own it anymore as used to be the case in the predigital world. In the new, digital  world, you don’t own anything. Instead, the cartels own you because now they can sell you a temporary license to use whatever it is that they are selling so that you will have to buy the same thing over and over again every year.

Things have certainly changed, haven’t they? Mostly for the worse, but in some respect probably also for the better. Translators now have many more tools at their disposal.

In addition to a powerful computer and word processor, we have the Internet where most of the time, we can find life-saving context and databases including bilingual lists of terms that we need for our work.

We now also have machine translation which we can use to get a basic idea about the text that we are translating, or a rough simile of what our translation might eventually look like.

Or not.

The problem is, many people who know nothing about translation, (and some who are selling translation for a living), are often unable (or unwilling) to make a distinction between machine translation and real translation.

Machine translation is not a real translation. It is a tool that can suggest a possible translation, which in some cases may be right, and often is completely wrong. A tool does not replace a human translator, whether the tool is a dictionary on paper, on a disk, or in a database, or a fairly sophisticated software package called a machine translation program.

It is understandable why people who don’t know much about the translating process would make this mistake. In the nineties, companies were training people to become operators of complicated machines manufacturing various products such as car parts, only to eventually replace these operators by robots capable of performing all of the manufacturing stages that used to be performed by humans.

Should it then not be possible to apply the same approach to translation?

A similar robotization of translation would be possible if the translation process could be divided into a great number of small but precisely measurable segments, which could then be reproduced with a software package. It would then be possible to replace a human operator called translator by software, while the product of this robotized translation would perhaps still be examined by a human operator who would be basically performing the same task as a file input clerk.

The problem is, words are not car parts. When you put car parts together the right way, you have a car. When you put words together the right way …. you have an expression of human mind and soul.

We use words as substitutes for mysterious processes occurring in a largely unexplored universe called the human brain. At this point, we really have no idea  how the brain works. We know a little bit about the trillions of processes taking place in the brain probably every second. But there is much more that we don’t know about the brain than what we do know about it – otherwise we would be able to cure for instance Alzheimer’s disease.

Translation tools and aids that translators are fortunate to be able to use now, as opposed to the situation 10, 20 or 30 years ago, including machine translation, should not be mistaken for what they are not, namely human brain that can react to and recreate the world around us in more languages than just the one that was originally programmed into our brain in our infancy.

To be sure, some translations are so simple and repetitive that robotization of the task with machine translation which can then be quickly edited by a somewhat bilingual human operator does make  sense and it is probably something that is already being used in many cases.

But for the most part, the results of such a process will be quite unpredictable.

If a creative bilingual or multilingual human brain is not a part of the equation controlling the entire process from the very beginning, the resulting product will be at best only a blueprint for translation that still needs to be put to test and approved by a real translator.

And if human creativity was not included at the initial stage when software and hardware took control over the decision making process instead of letting a human brain work its magic, the results may be correct, only partially correct, or completely incorrect, and only a human translator will be able tell which is which is which.