Posted by: patenttranslator | December 29, 2018

How to Train a Translator Like a Dog

The perfect dictatorship would have the appearance of a democracy, but would basically be a prison without walls in which the prisoners would not even dream of escaping.   Aldous Huxley

Those of us who have or at some point in time used to have a dog know that every dog needs to be trained. It’s best to start training them when they are still puppies, otherwise, it may be too late to do much about how they behave because, as the saying goes, you can’t teach an old dog new tricks.

Over the years, we had seven dogs in more than thirty years, not counting Cherrie-chan that we had when we lived in Japan, all of which my wife, who by now has been actually my ex-wife for seven months, got from the pound in San Francisco. Fortunately, they were all house trained, and the problems they had, like being aggressive and chewing furniture, which is how they ended up at the San Francisco dog pound, stemmed from the fact that the humans who owned them did not love them, or at least not enough to make a connection with the dog’s heart.

You can train even an old dog new tricks, but first you have to make a meaningful connection with the noble animal.

Lucy, a scary looking expert at love bombing.

The biggest problem with puppies is of course that you have to train them not to go to bathroom inside the house, generally by rubbing their nose in the mess they made in the room before they get the concept of what being housebroken means to a dog. They eventually all get the concept that this is a major no-no. Other problems with dogs, such as when they are too noisy and keep barking too much, or aggressive to other dogs or even people, must also be dealt with sometimes, but humans are generally more forgiving when it comes to these personal canine characteristics.

Dogs are masters at making sure that we allow them to behave more or less the way they themselves want to comport themselves rather than what we want from them by using various incredibly sly canine methods, including a tactic also known as “love bombing”.

We are exposed to canine “love bombing”, which is different from human “love bombing”, used both by men and women, but mostly and generally much more effectively by women, when the dog acts oh-so-very-happy when we come back home after having been gone even for only a few minutes, and starts doing things like wagging the tail, jumping around, or even jumping at us and licking our face. It’s a very, very effective tactic, and every dog knows that the tactic can be used to train humans by melting their hearts so that as a result, their humans will become obedient and mindful of perfectly legitimate canine wants and needs.

Just like dogs, translators can be trained too to behave exactly the way their owners want and need them to behave. Fortunately, virtually all translators are house-broken, but even if they were not, given that they work from remote locations through internet, their owners who run translation agencies in “translation industry” 2.0, would probably not give a shit.

The demands that the owners translators in the “translation industry” have on us also are very different from those of rightful human owners of dogs. And needless to say, unlike with dogs, the best and only true friends of humans, there is no love lost, or pretense of love on either side, either among translators or among their owners in translation agencies, which nowadays prefer to call themselves “LSP”, or Language Service Providers (ha, ha, ha).

Dogs allow us to train them because they know their owners love them. It is much harder for translators to be trained by their owners in the “translation industry” because based on how they are being treated, translators know that their owners despise them.

The tools that “LSPs” need to train their translator-slaves are also very different from tools used by humans who try to train their scary beasts and cute doggies to behave properly.

It should be mentioned that the training of translators by their owners in the “translation industry” as if they were dogs has a relatively short history.

Before approximately Anno Domini 2000, translators were not really considered by their owners to be kind of less than human, partly because prior to the current, corporatized and dehumanized form of the “translation industry”, translation agencies were usually run by former translators who liked the management of translation projects more than translating, partly because they understood what translation is about, although mostly because they knew that there is much more money in management of translation projects than in actual translating.

The modern translation agency owners, who Jesus would probably call money changers, are almost always completely monolingual, don’t really know anything about translation, they could not translate anything if their life depended on it … and they are proud about it.

They are just salesmen who would be just as happy selling cigars, or refrigerators, or cars instead of selling translations if they could make more money doing that. But there is good money to be made in translation, you can sell it without knowing much about it, and unlike in most other enterprises, you don’t really need much of a capital to start your translation business.

One of the first tools that the translator owners in the “translation industry” used to train their translators to show them how worthless they are were new  “NDAs” or “Non-Disclosure Agreements” that started being used to beat obedience into translators by the industry around the year 2,000.

These new “NDAs” were very long, unfair and extremely demeaning to translators, as I wrote in several posts on this blog, such as How the So-Called Translation Industry Turned “Confidentiality Agreements” into Declarations of Acceptance of Servitude, or One Possible Strategy for Dealing with Unfair and Demeaning “Non-Disclosure Agreements” Pushed Recently by Some Translation Agencies.

Many of the clauses in these agreements were illegal, but translators were too scared not to sign them because they needed translation work, even if generally poorly paid, so instead of protesting or finding their own customers, they just shut up and signed.

Approximately at the same time, the agencies, especially the big ones, started lowering the rates that they were paying for translating. This was made easier by the fact that many new translators living in countries with a much lower labor costs than for instance in Western Europe or North America entered the competitive field of supplying cheap translations from third world countries.

The latest innovation in this trend to outsource labor overseas, imitating the trends of large corporations in other industries, is the development of so -called back offices in other countries where labor is very cheap, which in turn may be working as subcontractors for other back offices located in yet other countries where labor is even cheaper.

Agreements to work for their bosses anonymously, the customers can’t tell from which country or continent a translation that they think originates in United States, France, or England really came.

While the prices of translation charged to the customers did not really change that much, the rates paid for the translations to far-flung, inexperienced newbie translators were slashed to the bone, which worked very well for the bottom line of the “translation industry”.

The rates paid to translators were further reduced by the invention of wage theft enabling programs such as Trados, which can be used to remove from the word count translated words that translators are not paid for, as I write for instance in There Is Really Nothing Fuzzy About the Logic Or the Concept of Fuzzy Matches or in Is Trados Co-Responsible for the Falling Rates in the Translation Industry?

The “translation industry” is still busy innovating away along these lines, which is to stay trying to come up with new ways to increase its profits by paying translators less and less for their work.

Yet another innovation of the “translation industry” is post-processing of machine translations created by machine translation programs by pitiful human automatons, who are no longer called translators, but go instead by the generic title of “post-processors”.

I have been writing about this latest gold rush concept of the “translation industry” also in many posts on my blog, such as in Post-Processing of Machine Translations, the New Money Maker of “the Translation Industry” or in There Are Many Ways to Commit Suicide, But Post-Processing of Machine Translations Would Be a Horrible Way To Go.

Of course, these automatons, formerly called translators, who are paid next to nothing and usually reside again in countries with very cheap cost of labor, will not catch most of the mistakes because they generally have no qualifications and are paid almost nothing for what would be very hard work – if it was possible to do it well. But since doing something like that well would mean retranslating the whole damn thing by an actual qualified and experienced human translator, this would beat the purpose of the entire exercise since it would be very time-consuming and much too expensive.

Here I do find one interesting similarity between training a puppy to become house-broken and training a lowly post-processor to catch machine translation mistakes. You have to rub the nose of the post-processor in the shit and piss left in the text by the machine translation program, just like you have to rub the nose of puppies in their own shit and piss that they keep leaving on the carpet before becoming house-broken.

But the big difference is that while virtually all puppies eventually figure out that if they wait until they can go outside, their sensitive noses will not be subjected to the horrible smell of their own excrement, human post-processors of the detritus left after pre-processing by machine translation programs will have their noses rubbed in it and will be exposed to the cruel and unusual punishment that is their job until they find a better job or until they die, whichever comes first.

Happy New Year 2019!



  1. Just this past year, I had a small translator-run agency client in Japan suddenly change the estimation method for Japanese-English projects from a character count to the Trados word count. I pushed back with a detailed analysis showing that the word count was substantially lower, unfair, and not something we had ever used or discussed. They finally caved and went with the character count, but I haven’t worked with them since. They used to be a good client, but I suspect they are having financial issues to blatantly start gouging translators like that.


  2. Everybody else in the “translation industry” is doing nasty things like this to make a buck, so even small agencies run by relatively honest people will try to do the same thing, even if it is blatantly dishonest. I had a small agency – a husband and wife team – do the same thing to me to me about 10 years go, namely try to change the word count from the actual word count to Trados count.

    In the end, the guy paid me. But I never worked for him again. It’s too bad because we were friends and I used to enjoy talking to him. But I would not touch a job from him now, or even just talk to him, because I lost the respect I used to have for him.

    It’s best to work mostly for direct clients because they are less likely to be dishonest in this manner – if they are not just a middleman aiming for the highest possible margin – and I work for agencies only occasionally on smaller jobs, so that I can always tell them no!

    And if they can’t take a no for an answer, I can just drop them without worrying about it.

    Then again, I am semi-retired and don’t really need to work anymore. So that helps a lot.


  3. A friend who has translated Japanese patents for two or three law firms since 2000 was just told one client will now use a translation agency instead of him, even though told his work had been excellent. He estimates that he will lose 60% of his income from January and so has quit translation after doing very well and will likely go back to school at age 45 since his masters degree is in the humanities.

    In 2010, I told him that I thought this would happen around 2020 as MT continued to improve, and it will happen in early 2019 as he winds down. The good news is that he is financially secure and is looking forward to his new path.


  4. You are so smart, Jeb, always predicting everything so correctly.

    You should be a politician or something.


  5. Well, your predictions are doing well to as far as we know. You said in 2001 that translators still have centuries left of translations to do. I noticed around 2013 or so that you scaled that back a bit and said 20 years left. Then you trimmed the projection a little more and said “at least until my retirement.” Well done.

    Happy New Year!

    P.S. I think you will enjoy the medical advances that more powerful computers will bring in the 2020s.


  6. The problem is, Jeb, you don’t want to understand what I am saying if it does not fit your tunnel vision.

    All of the quotes above are compatible with my old quote from 2001, namely that translators will be translating for at least a few more centuries, more likely forever.

    Incidentally, as I am already retired, I don’t really need to make money because my “fixed income” is more than sufficient to cover all of my expenses now that I am financially responsible only for myself. When I was still working full time, I was financially responsible for four people: myself, my wife, and my two sons, so I was basically a slave.

    But I ain’t no slave no more!!!

    Do you have a good plan for your retirement?

    I still work because my old customers keep sending me interesting work and I like to work and have additional money. I don’t work for agencies, I use some computer tools, mainly machine translation, but these tools are not and most likely will never be a replacement for real translation originating in human brain.

    Happy New Year to you as well!


  7. What is my tunnel vision? You still think there will be a need for technical/news translators in 500 years, let alone 5 years? I guess so, as long as the MT editors work for three cents a word.

    Oh, congratulations on your new found freedom.

    We’ll soon see that retirement will be about a hundred year phenomena. There was little retirement before 1930 and there won’t be retirement after 2030 as A.I. continues to become exponentially more powerful in all areas in the 2020s. There isn’t a bright red line but nobody under 50 needs to worry about retirement.


    • There will be retirement after 2030 as long as I am alive. Enjoy your permanent non-retirement, if that’s possible.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: