On Golems, Dinosaurs and Hamsters As Brainwashing of Newbies in Translation Industry Reaches New Heights

I noticed on my WordPress dashboard that viewers were being referred to my silly blog for several days from the following URL: newclasses.nyu.edu/portal/tool/4e266304-bddf-403b-9e94-427c8e59bf9f/discussionForum/message/dfViewThread. But since I am not a registered NYU student, the link does not work for me.

So I did not find out what this link was about, although I suspected that some language teacher was using again something that I wrote in a language class at the New York University. It happened already at least once before, as I found out from a student in the French class at NYU who lived here in Chesapeake when she called me a few years ago. We met for a nice chat over coffee, which eventually resulted in regular meetings of a small group of local translators, unfortunately defunct at this point.

But yesterday I received a long comment on my blog which seems to explain the referrals to my blog from a list of new classes at New York University.

I found the comment, which contains a number of grammatical errors, while the style is not exactly something to write home about either, very interesting and refreshing.

Here it is (emphasis mine):

I attended a panel discussion at NYU, about MT and CAT tools. Here is the report I forwarded to my instructor, who is also worried about the intrusion of the “Machine Monster”.

Dear xxxx,

There’s good news and bad news! The good news is that the translations industry is thriving. The bad news is that you have to become very good friends with the translation Golem in order to thrive!

There were six panelists: four representatives from agencies, one from the UN and a dinosaur by the name of XXXX, who teaches French in the NYU translation program, and who is vehemently opposed to machine translation and even CAT tools. She started the discussion, by telling the audience that her training and decades of experience allow her to work just as fast, if not faster, than she would with the aid off CAT tools. According to her, transitioning from a typewriter to a word processing program was the most fundamental and beneficial change in her professional life, but also the last one. Her delivery was the quintessential New Yorker mixture of ranting and whining. Then we heard from the opposite end of the spectrum, a PM from an agency that uses machine translation almost exclusively for technical translation, which is then post edited to varying degrees, based on the needs and requirements of the customer. The quality of machine translation varies depending on both the language pairs and the subject matter / field. For example it works great for Brazilian Portuguese, well for German and Scandinavian languages and badly for Hungarian, Greek etc. It works great for automotive translation, but not really well for patent translation, since it describes elaborate processes. So there is hope!

The main message that I took away from the discussion is that there is stratification taking place, or has already taken place, within the translation industry. LSP’s are now able to deliver different products based on the needs of the customer. Someone who wants to put an ad on social media, that people may look at one, is less concerned about the quality of the translation than someone defending a patent in court. Another point that was brought up is that we have moved into an age of mediocrity, where the proper use of language is becoming less and less important. Therefore some customers question why they should pay lots of money for a grammatically perfect translation, if the target audience doesn’t know the difference. Harsh, but true! But then, there is still a market for good translations done by professionally trained translators, which the author, of the piece that you sent me, also points out.

Other highlights of the discussion:

The tech guru, from one of the agencies, made this statement that Google Translate is the best thing that ever happened to translators. The reason is that many people think ‘great, now I can do my own translations’; but then, when they see the garbage that comes out, realize that Google Translate is not really that great!

The thing that was brought up over an over is the fact that there is a lower expectation in the target language. People are just not concerned with proper usage and grammar anymore.

Agencies not only match their translators with projects based on their language pairs and specialities, but also on the CAT tools they use, because clients oftentimes demand the use of specific tools.

Larger projects, that in the past were given to one translator, are now given to several translators accompanied with very specific software that ensures that they are all ‘on the same page’, even monitoring their speed, work habits, etc.

This commercialization of skills is not only taking place in our industry, but in many others as well. I.e. people now longer pay attorneys, doctors etc. for advise that they can get online, or from another less expensive source.

I am definitely on the side of the dinosaur here, which is of course not surprising because I am a dinosaur myself. I don’t use CATs because like many translators, possibly the majority of those who do not have to work mostly for translation agencies, I find them useless for my purposes, as well as counterproductive.

I use MT only as a context-based online dictionary because I think that editing of garbage can only result in garbage that does not stink quite as bad. That must be the dinosaur in me too.

I absolutely agree with the commenter’s conclusion that stratification is taking place in what is called translation industry. Translation agencies have realized that it is possible to make money by selling “edited”, and sometime even unedited, machine pseudo-translations to customers who as the commenter put it “question why they should pay lots of money for a grammatically perfect translation”.

But they may not realize that the hamster-edited pseudo-translations are not really just somewhat grammatically imperfect translations. Syntax and grammatical errors can be relatively easily fixed even by a hamster who does not know the source language very well as long as he or she speaks English. The problem is, the “edited” product of the Golem monster (Machine pseudo-Translation) may contain mistranslations that a hamster, even a well trained hamster who is a whiz at using search engines and computer-assisted tools, will not be able to catch.

What is the correct ratio of translation versus mistranslation that would justify a price reduction of, say, 40%? Would a 40% discount be a fair tradeoff for a translated document that contains only 10% to 20% of mistranslations? But what if only 5% of the document is mistranslated, except that the mistranslated section happens to be the most important part of the document? Would that not then make the entire translation one huge typo?

I disagree with some of the other conclusions of the commenter, although these may have been the conclusions of the presenters at the NYU newbie translator indoctrination class.

It is not the case that people no longer have to pay for lawyers or medical doctors because they can have their questions about their medical status or legal problems answered for free by using an Internet search machine.

Some of these questions can be answered in this manner without having to consult a medical or legal professional. But in fact, lawyers and doctors are still charging just as much, or more, for their expertise as they did before the Internet started impacting their business about two decades ago.

Some people will use machine pseudo-translation to translate all kinds of documents to and from different languages without giving it a second thought. And some customers will fall for the salesmanship of translation agency gurus offering to “dramatically reduce the cost of translation with new language technology”, meaning mostly with computer-assisted translation and machine pseudo-translation edited by newbie translators turned into editing hamsters.

Machine pseudo-translation is for example the perfect solution for determining which documents need to be translated and which documents are not very relevant in patent translation, which is my line of work.

But for critically important texts, documents that may not contain errors and mistranslation, such as for example patents, or poorly worded, ineffective or counterproductive texts, such as financial prospectuses or advertising texts, the mess created by the machine pseudo-translation Golem, even if it is edited by human hamsters skillfully controlled by machine translation agencies, will simply not do.

For documents that matter, the clients still need a good, old-fashioned dinosaur who has university training and decades of experience, not hamsters who are forced with “language tools” to run faster and faster on the hamster wheel of our glorious translation industry.

According to the legend of Golem, a monster made of clay who was created by Rabbi Loew to protect Jews in Prague against pogroms, the monster went at one point berserk and started destroying everything around him when Rabbi Loew forgot to remove the shem, a magical stone through which the Rabbi was able to control Golem, from his mouth.

We can remove the sham of machine pseudo-translation from our thinking as a concept of “a language technology tool” that will solve all of our translation needs only if we realize that just like the Golem of Rabbi Loewi, whose grave in Prague I remember fondly because I had a date there with a rather interesting girl many years ago, the MT Golem can be very useful to us, but only if we know how to use it properly.

It can also be even more destructive than Rabbi Loew’s Golem of Prague in the 16th century if we try to use it for tasks that can be accomplished only by a highly educated and well trained human translator.

Or, in other words, for something like that you will need a dinosaur.