Posted by: patenttranslator | November 15, 2017

In Its Ignorance and Greed, the Translation Industry is Mistaking a Tool for the Product

Relative newcomers to what is now called the translation industry may not realize that what used to be thought of simply as “translation” or the “translation business” rather than the translation industry has been turned upside down and inside out in the last two decades, especially after the 20th century turned into the 21st century.

By relative newcomers I mean people who have been translating for a living for less than two or three decades, because that is how far back in time one would need to go to be able to make a meaningful comparison between what freelance translation used to be in the pre-industry period, and what it is now.

That is what I have been trying to do in many of my posts, although some people might say that what I am trying to talk about in posts in which I doggedly criticize the current practices of the translation industry is pretty useless and mostly a waste of time because I am fighting an 800 pound gorilla. And up to a point, I would have to agree with them.

My posts are not going to make any difference to how the current translation industry model sees the role of translators. The industry sees us as unimportant and interchangeable busy bees whose job is to keep bringing more and more sweet honey in the form of translations to the industry, which is in the role of a beekeeper who consumes the honey.

The more honey we bring to the beekeepers, the better for them, and any method the industry can think of and that can be used to achieve the goal of “higher productivity” is thus obviously legitimate, especially since all of the new tricks in the industry’s book are in the end obediently legitimized by translators’ associations (such as the ATA, but not only the ATA).

The illegitimate methods popular in the current version of the translation industry include industry tools such as so-called confidentiality agreements, now often stuffed with illegal clauses, which include illegal and disgusting conditions, such as that the industry has the right to install spying software on our computers to take advantage of technology that makes it possible to better control the busy bees, just in case the bees might be getting improper ideas.

There are many other immoral and illegal tools the translation industry uses to keep more of that sweet honey away from hungry, hard-working bees, with a greater share for itself, even if it means the bees doing the work starve to death. These new tools include “full and fuzzy matches”, i.e. full and partial repetitions, for which only partial or no compensation is to be provided to translators who do the actual work based on fuzzy thinking prevalent in certain segments of the translation industry.

Another “tool” the translation industry is currently pushing is post-editing of machine translations.

Although the translation industry is very excited about this new tool, most translators for some reason keep refusing to use it.

I see machine translation as a powerful and very useful tool, especially since it is available for free to translators and non-translators alike.

But the fact that the translation industry is incapable of understanding the difference between a tool (machine translation) and a product (actual translation) shows how little the industry understands translation, which is what it is supposed to be producing.

Translation as a product has been with us for many centuries, long before tools such as pens, typewriters, the internet, and machine translation were invented.

But up until now, the tools that can be used for writing (and translation is just a special subcategory of writing), have not been mistaken for the actual product, i.e. writing or translation. It took the chutzpaw, so typical of the modern translation industry, to attempt to erase the difference between a tool like machine translation and a product of the human mind called translation.

New tools are created all the time. Especially in the last few decades, many powerful new tools have been created that are very useful for translators.

For the first 10 years in my freelance translating career, I was working without the tool called the internet, although I can no longer imagine working without it.

For the first 15 years in my freelance translating career, I was working without the tool called machine translation, which is alternately available and not available to me.

I don’t see the fact that translators use this tool as a dirty secret that should be hidden from prying eyes.

As I wrote in my last post, I recently read a corporate blog in which the author of the blog post said that post-editing of machine translations is “something that everybody is doing … the dirty secret that nobody wants to talk about.”

It is true, as the post says, that everybody is using machine translations. The translation industry is using it, and individual translators are using it too, including myself. But it is not true that, as the post says, everybody “is doing some sort of post-editing” of machine translations.

Translators for the most part stopped using dictionaries in the form of expensive and heavy books that are already obsolete by the time translators pay for them, and replaced these books with online databases and machine translation. Although I have more than a hundred specialized dictionaries, I generally use them only a few times a day, and I have not bought a new one in more than a decade.

The fact that translators have been using machine translation programs instead of dictionaries for some time is not exactly a secret, let alone a dirty secret. Many posts on my blog are dedicated precisely to this subject.

The fact that the translation industry claims that everybody is “doing post-editing of machine translations” and considers this claim a dirty secret is another piece of evidence as to how little the translation industry understands its own product, i.e. translation.

I do not doubt that the translation industry does a lot of post-editing of machine translation by using translators, probably newbies or “bilinguals” who are preferably located in countries where labor is very cheap, who edit the machine translation detritus for the equivalent of 1 cent per word or something like that.

The industry either does not understand, or if it understands it, simply does not give a damn that editing machine translation is in fact an impossibility. As long as the text to be translated is at least moderately complicated, it needs to be retranslated to remove major mistakes that machine translation creates, regardless of how good the software might be.

Machine-translated texts can be used by a translator as an excellent source of information. But because editing such a source of information is so laborious and time-consuming, and no amount of editing is likely to remove the mistakes that mechanical processing that any type of machine translation is based on, I doubt there are many actual translators whose work includes post-editing of the product of a software package.

Machines will never be able to understand the source text, only humans can do that. Would it make sense to post-edit a poorly written book? I don’t think so. Although some parts of the plot might be reused, it makes much more sense to write a new book, and the same is true for translation.

No software or computer is capable of what is a simple act for a human mind, but an impossibility for artificial intelligence: actually understanding anything other than how to carry out instructions based on mathematical formulas.

Although the translation industry is very aggressively pushing the idea of post-editing of machine translations by slave-like quasi-human translators, and will be probably doing this for many years to come, I think that a strategy based on greed and misunderstanding of what translation actually means is bound to backfire.

The industry assumes that its clients are so dumb that they can’t tell the difference between what it calls “the dirty secret that nobody wants to talk about”, and the result of post-processing of machine translation, and a real translation, which can only be created by a real human translator, as opposed to what is created when an underpaid post-editor who tries to save what has been generated by a machine using software, so that the industry can then sell it as real translation.

For some purposes, such a product might be good enough, especially since it is significantly cheaper than real translation.

For some purposes, unedited machine translation is already good enough, especially since it is free.

But for most purposes, neither edited not unedited machine translation is good enough. One purpose for which machine translation cannot be used is patent translation, especially of patents for filing, which is mostly what I am doing now.

The industry’s clients are not happy with what the industry is trying to sell them. That is what I hear from recent clients, mostly patent law firms, who don’t understand why the quality of translations is suddenly so poor now, in part probably because they don’t know about a “dirty secret that nobody wants to talk about.”

The generally low quality of translations characteristic of the current form of the translation industry thus creates an opportunity for highly specialized and experienced translators who are able to offer real translation in specialized fields to clients who need it, but only if we are able to bypass the translation industry and connect directly with the actual users of our service.

Since I don’t work for the modern form of the translation industry, I am very fortunate to have been able to take advantage of this opportunity for quite some time.

I think that the fact that I have never been as busy as I am now is is not a coincidence. I think it is due, at least partly, to the fact that the translation industry is mistaking a tool – machine translation, for the product – real translation.

The more the translation industry continues trying to push its “dirty secrets” such as edited machine translations to sell them as real translations, the more work there will be for specialized translators who can provide what the clients really want.

In its ignorance and greed, the translation industry is thus creating new opportunities for those of us who refuse to work for it and who are able to sell our skills directly to the people who need them.

 

Advertisements

Responses

  1. Another excellent posting that says it as it is. It seems to me that the translation industry, if we want to call it that, is suffering in the same way most other industries are in this day and age, namely from the so-called scissor effect – this works in the way that the quality of a given service divides itself naturally into two categories, the bargain basement, where cheap is everything and never mind the quality (as in our beloved pound shops of all varieties) and the quality sector, where you have to pay top dollar to get the quality you need or want.

    I do not see translations being an exception from this and, like the author, am just glad that I specialised early enough to now be part of the quality sector, where I have also never been busier.

    An unfortunate development but, alas, a sign of your (greedy) times.

    Like

  2. Yes, it’s all part and parcel of the insane corporatization of the world.

    I don’t see good things happening in other industries either.

    A lot of really bad things happening at the same time is much more likely.

    I am an old man already, and what I see around me is spelled out in the title of this film:

    Like

  3. Steve, you put your finger on a crucial point in the use of MT by professional translators when you write: “As long as the text to be translated is at least moderately complicated, it needs to be retranslated to remove major mistakes that machine translation creates, regardless of how good the software might be.”
    In other words, if we use MT for reference or information, we then normally RETRANSLATE it. There is a big difference between RETRANSLATING and POST-EDITING.

    Liked by 1 person

    • Absolutely, Victor! I disagree, however, that there is any professional value in machine pseudo-translation (MpT), and I’m always a bit appalled when I hear or read of someone relying on it like a dictionary. I use it myself that way for Portuguese in a supermarket or DIY store to tell which is the beef and which is the pork or to see if I have the right word for “hammer” before I ask where to find it. But for professional work, I don’t need some context-free bullshit spewed out by an algorithm. That’s like the worst of the old, bad dictionaries that offer various words in a numbered list but no context to know what really fits. I also don’t consult my hardcopy dictionaries like I used to, but not because MpT offers me anything useful. The real joy of the modern information infrastructure is the ability to do proper corpus studies, to switch to an image search for your keywords and other tricks that help one to understand context and regional language issues better than most dictionaries ever did. Corpus linguistics is where the real professional action is; screw the lying profiteers and their MpT BS. It’s just another remix of the scams I’ve watch IT “experts” pull in some 45 years of IT consulting, development and use. When I hear the latest sales pitch for “neural” progress or whatever, I am always reminded of the Y2K scam that I was a part of, where I watched washed-up COBOL programmers have a last hurrah stuffing their bank accounts by playing on the fears of the technically ignorant. The machine translation gurus and circus barkers are largely of the same sort, with very few exceptions.

      Like

  4. Hi Kevin (long time no see).
    I like your bonfire for the straw men (i.e. the idea of translators RELYING ON the results of MT aka MpT and RELYING ON dictionary entries). Perhaps I could help by lighting the fire for your torch.
    And I, too, look for concepts in context (e.g. in legal texts from various jurisdictions, and sometimes in images for construction-related terms). Perhaps you have formalised these processes more (e.g. in “corpus linguistics” and “keyword image search”), but essentially our approaches (and our horror at the prospect of MT post-editing) are similar.

    Liked by 1 person

  5. “It is true, as the post says, that everybody is using machine translations.”

    Erm, not this translator 🙂

    “Would it make sense to post-edit a poorly written book?”

    I hate to break it to you, Steve, but this is done all the time 🙂 I have a friend who’s an editor, and she frequently chops up, moves around and even rewrites poor manuscripts to turn dross into something which is at least readable, and often rather good.

    Like


Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

Categories

%d bloggers like this: