People have been hoping to find a quick and simple solution to all of their problems in technology for many centuries. Rudolf II, The Holy Roman Emperor (1552 – 1612), also called “The Mad Alchemist”, had his own alchemist’s lab and was a patron of a number of alchemists at his court in Prague some four centuries ago. He was convinced that it had to be possible to create a magical substance called “Philosopher’s Stone” (“Lapis Philosophorum”), a substance that would “transmute” base metals (such as lead) into noble metals (gold) and that it was also possible to develop a magical “Elixir of Life” which would confer on him and those who could afford it almost eternal youth and longevity.
Personally, I think that spending a lot of money on similarly insane undertakings in the realm of science makes much more sense than spending trillions on bombing and invading foreign countries, because something good may eventually come out of these crazy projects since alchemy eventually gave birth to what is now called chemistry.
But that’s just me. My government obviously disagrees with me.
Tourists from several continents still flock to a tiny street behind the St. Wenceslaus Church (chrám Svatého Víta) in Prague called “The Golden Lane” (“Zlatá ulička”) to take with their tiny smart phones and huge tablets pictures of tiny houses that were built for the intrepid alchemists (people were much smaller four centuries ago).
The almost religious faith in the magic of technology is evident also in modern “translation technology”.
People have been trying to discover a way to translate human thoughts expressed in words with machine translation (MT) since about 1945. Although MT became a very powerful tool, especially compared to the situation 70, 50, or even 10 years ago, MT still only translates words, while it has no clue (and never will) about the human thoughts behind these words.
Google Translate is the only MT tool that kind of makes sense, some of the time, sometime even most of the time, but there is a reason why it seems to be better than other approaches to machine translation. Google Translate is not based on trying to analyze with machines and software human thoughts in one language and then “transmute” them into another language based on rules expressed with algorithms. Instead, it attempts to instantaneously locate a translation (originally created by a human translator) of the same or similar human thought into the desired language. It is a clever approach, but even this approach can only work, to a limited extent, for texts that have already been translated.
Several other aspects of “translation technology” are also hailed as magical tools that are going to revolutionize human translation.
Although speech recognition has been around for several decades now, relatively few translators are in fact using it. I thought of using it myself, but since I translate mostly Japanese and German patents, I realized that it would probably not work for me.
Every translator has an idiosyncratic method and every translator’s method is different from what somebody else is using. When I translate a Japanese patent, I need to visually keep jumping from the beginning to the end of long sentences to locate a verb that is likely to be hiding there, and then I need to jump back to the beginning and insert the verb where it would belong in English, while making sure that this is the correct placement based on the meaning of the text. It is obviously very easy to misplace the verb when a long sentence contains several verbs, which would result in a mistranslation. I also have to keep in mind that the subject is often not expressed at all in a Japanese sentence and that it can often be found several sentences prior to the present one, sometime on previous page. For these reasons, I don’t think that speech recognition software such as “Dragon NaturallySpeaking” would work for my purposes. And since it cost 800 dollars, I never gave it a try (and never will, unless they lower the cost to 80 dollars).
There are also other quasi magical and more recent tools, including computer assisted translation tools (CATs), crowdsourcing of translations, which means throwing texts into clouds where thousands of invisible humans (who are usually not really translators) will be eagerly translating texts from a foreign language (for peanuts or for free), or human-machine localization of translations, which is a concept in which humans (who again are usually not really translators), “edit” machine translations so that they would finally start making sense.
I think that CATs will probably be around for a while, maybe for a long time, because unlike the other tools mentioned above, these tools are in fact useful to translators, albeit only to some translators and only for certain types of translations. I don’t use them, and probably never will, for reasons that I have tried to explain in many posts on my silly blog.
But the more I am trying to explain why I find CATs useless for patent translation, and in particular for the way I translate patents, the more I am being attacked as a luddite and ignoramus by CAT true believers who simply refuse to conceive of the possibility that some translators may choose not to partake of their beloved toolbox.
I have noticed that all of these new and wonderful tools have one thing in common and that they are sold to translators in the same way: If you, dear translator, start using our Dragon, our CATs, or finally learn how to quickly “edit” MT for us, this will make it possible to increase the efficiency of your primitive human translations, which means that you will make much more money because instead of translating two or three thousand words per day, you will be able to easily translate TEN THOUSAND WORDS A DAY!
The problem is, recent history has shown that instead of making much more money, many translators are making less money if they are using these tools, especially if they work for translation agencies who started requesting obligatory discounts for “repeated words” and paying full rate only for “new words”, based on clearly insane concepts called “full matches” and “fuzzy matches”.
You can’t transmute lead into gold, no matter what kind of “Philosopher Stone” technology you are trying to use, and if you triple your daily output of translated words with your magical Dragon or your magical CAT, the chances are that some of the words will be incorrect, since human brain is not really designed to catch in a single day all of the potential mistakes that may be hidden in a huge chunk of text containing ten thousand words.
Old-fashioned work that is based on a high level of concentration, combined with human knowledge that is based on education and many years of experience is still and always will be the best approach to translation, and that is what I intend to keep offering to my clients.
Which is not to say that it is not possible to turn straw into gold with all of the new and terminally cool translation “translation technology tools”.
All of these tools are now surrounded by so much hype and so many lies that we are entering Rumpelstiltskin territory.
The translations created with the help of these tools will not be exactly pure gold, and if wealth is created by spinning straw into gold with these totally cool technologies, based on recent history, it is clear that translators will be paid with straw, and translation agencies who can force translators to use their “preferred tools” will get to keep the real gold.