A Few Misleading Terms and Curious Notions Popular in the Translation Industry

I have written quite a few posts on my blog about a number of misleading terms and notions that have been custom-made for the general public and for translators in particular by the giant PR machine of what has become known in the twenty-first century the “translation industry”. Some turned out to be quite popular, some seem to have mostly fallen on deaf ears.

The very term “translation industry” is misleading because no distinction is made with this term between translators and interpreters, namely between the persons who provide translating and interpreting services, and brokers who buy these services from translators and interpreters in order to sell them to companies and individuals, i.e. the actual “translation industry”.

That is also why I try to discourage translators from using the term “language services provider”, or LSP, in online discussions among translators. This acronym, still a complete mystery to people who are not working in or for the industry, appears to have been designed by industry experts to simply make the profession of translator disappear into the numerous hungry mouths of translation agencies who would love to swallow us translators up by creating a handy abbreviation.

And it seems to be working because this is what is now in fact already happening with the help of our so-called professional associations, most of which work for the “translation industry” these days instead of working for translators.

If translators accept terms in their naiveté that have been recently invented and coined, especially for us by the “translation industry”, they may be galloping toward extinction within a few short decades, or perhaps even sooner.

The “translation industry” would love to turn translators into subservient, obedient drones who would be grateful to the mighty and magnificent industry for any kind of work at any level of compensation. And although this goal has been on the industry’s agenda now for at least two decades, not a word has been written about this taboo subject so far for example in the ATA (American Translators Association) Chronicle, which proudly and somewhat ironically calls itself “The Voice of Interpreters and Translators”.

As far as the ATA is concerned, we are all one big happy family because both translators and translation agencies are “stakeholders” pursuing the same noble goals. To say that our interests are in fact often opposite would be blasphemy, which is why no criticism of the industry is allowed on the pages of the ATA Chronicle.

Fortunately, we have blogs and social media where a lot of information is available for those of us who want to understand what is happening and how and why things are changing for translators.

If most younger translators don’t know at this point that some 15 years ago, there were no “LSPs” or “language services providers”, only translation agencies and translators, it is quite possible that 15 years from now most younger translators will not know that the notion of “post-editing of machine translation” was at one point a practice that was disparaged by most translators because it was literally killing them, both intellectually and financially.

After more than half a century of impressive, albeit very slow and only incremental progress, the fact remains that machine translation, no matter how much editing is wasted on it by pitiful persons formerly called translators and recently renamed “post-processors”, is an exercise in futility and will always result in slightly less repulsive garbage instead of resulting in real translations equivalent to the translation humans do.

Machine translation kills the spirit of human communication. It does so by default because machines do not have the capability to understand the concept of communication among humans and an algorithm will never be a suitable substitute for human thinking.

And because machines are not very likely to grow a brain inorganically, on silicon, the “translation industry” needs to appropriate the brains of persons formerly called translators whose formidable task would be to revive the spirit of human communication that has been mercilessly put to death by an unthinking machine.

Google Translate has attempted to circumvent the carnage that “rules-based machine translation” can inflict upon translation of communication among humans by identifying previous human translations that are very close or almost identical to new texts that need to be translated.

It is a very clever idea and this approach indeed works much better than the older approach that is based on marrying dictionary definitions of words with rules and myriads of exceptions in the grammar of various languages, because such an incredible witches’ brew is created from these rules and exceptions that only a human brain can make sense of it.

However, the problem with the approach of Google Translate is that a translation that was previously supplied by human translators and that seems to be almost identical to another document in a foreign language may in fact be completely inaccurate, although it may seem to be perfect.

If for example the closest translation of a text about a real estate transaction says, “The property on 1234 Sunny Lane is definitely not worth one million dollars”, reflecting the considered opinion of an expert on real estate values from the previous year, but the current expert opinion of the same person says, “The property on 1234 Lane is definitely worth one million dollars”, Google Translate could easily substitute the old expert opinion for a new translation … because it is the closest existing human translation and a new translation has not been provided by a human brain yet. And it may appear to be a perfect translation.

I happen to know that this kind of thing happens all the time with Google Translate.

Very often when I print out a machine translation of a patent I am translating for a client, the machine translation is impressive and looks almost like a real human translation, while it seems to contain only a few blemishes here and there.

But then, when I compare it to the text of the patent application that I am translating, it sometimes has for example a different number of claims and other differences, which means that it is not really the actual translation of the text of the patent application, but instead only something that is very similar.

But very similar is not the same thing as … the same thing.

To file a machine translation of a patent that looks perfect, but says something different from the original document could lead to disaster. To rely in court on a machine translation that looks perfect, but does not correspond to the original document, would also likely result in disaster.

Let me try another dramatic example.

If an intercepted command from Army Headquarters issued by a general from an enemy’s underground bunker from ten minutes ago was translated by a human as, “We must not launch a nuclear strike at this time”, but since no human is available on the spot, a Google Translate substitute is used instead of a human translation for a new command intercepted that only a minute ago said, “We must now launch a nuclear strike…” there could be a minor problem if the machine translation was used to take or not take action, instead of a real translation.

Nevertheless, similar is good enough for the “translation industry” and the industry simply loves the concept of selling machine translations, “post-edited” or not, because it could be an extremely profitable line in the “translation business”.

If the industry could make its concept of post-editing of machine output by humans really work en masse, translators could be reclassified as workers who are simply processing and replacing words in the same manner as the many thousands of “freelancers” called “Mechanical Turks” who work for a dollar or two an hour.

The concept of mechanical Turks describes humans who are often located in impoverished countries and who work for large corporations such as Amazon or Microsoft in part-time jobs as I wrote in a post titled “How Many Translating Turkers Are Hidden Inside a Box of Language Tools?” eighteen months ago.

The difference between the concept of the “crowdsourcing mechanical Turk workplace”, to quote Wikipedia, and the concept of a crowdsourcing workplace for translators who would be replacing words or even sentences at a very low rate, hopefully for free for the sheer fun of it, for the mighty industry, is that people who look for example for mismatched numbers or mismatched colors in an Amazon order do not need to have any substantial knowledge of anything.

They only need to have a pulse and a human brain that works reasonably well.

But translators looking for mismatched words in machine translations need to know at least two languages as well as a lot about the materials that need to be translated. In fact they need to know so much that only those who have not only a very good knowledge of more than one language, but also a degree in a certain field of human knowledge, are likely to produce good work.

Good translators of literary works need to have a flair for creative writing. Good legal translators usually need to have at least some legal education and good patent translators usually have technical education. It is possible to become a good legal or technical translator without specialized education, but it usually takes several decades of hard work to overcome the handicap of lack of education.

But educated and specialized, highly experienced translators, that is not exactly what the modern “translation industry” wants or is interested in. That was so twentieth century!

The modern version of the “translation industry” in the twenty-first century is much more interested in figuring out how to get newbie translators to post-edit machine translations.

After all, how hard could it be to replace a few misplaced words by a few other words that would sound better?