Can We Always Blame the Machine Translator?

Good translator or bad translator, that's the question.Machine translation is both an enemy and a friend; the bad ass and the best friend in the same package. At least this is the impression what you get from the online writings about automatic translation. The topic is still such a hot potato that people do have strong opinions and they are willing to share those ideas with the whole world.

This may be a bit too pointedly said to, but the split between defenders and opponents is quite clear. Most professional (human) translators or interpreters dislike (or even hate) every machine translator. They seem to think that the quality of any machine translated text is poor. For them machine translation is an enemy which is stealing their jobs. On the other side are all the instances who are searching for cheap translations. For them machine translation is a treasure which translates everything, anytime and for free. They are rationalising that “any translation is better than no translation, right?”

Surprisingly, in any situation when there is a translation error, it doesn’t matter anymore which side you belong to. Machine translation is (more or less) the one to blame for every mistake. Professional translators think that none of their colleagues are able to make such mistakes, and those who use machine translation and know its weaknesses think there is an universal correlation between low price and low quality.

However, the fact is that anyone can start a translation service and call oneself a translator. This doesn’t automatically mean that these translators would (intentionally) produce worse translations, but it does mean that their skills, motivation or work ethics might be different. Human beings also tend to make mistakes every now and then.

Still, machine translators get all the negative publicity over and over again. The situation is as bizarre as the following quote from TranSlate Translations -blog discussing about a serious translation mistake on a drug package:

“It was evident that this was an automatic translation that was never reviewed or corrected. – – – In addition, whatever software they have used, was just laying the translations literally and literally wrong. – – – Even Google got the whole translation 98% right!”

The author of the blog is basically saying that the awful Spanish translation must be an automatic translation and that the most used machine translator was able to translate it correctly. To me this statement sounds more than conflicting: A machine translator translated the text almost flawlessly, but still there have to be a machine translator behind the bad translation. I don’t see the rationale here.

Maybe the pharmaceutical lab didn’t use machine translation, maybe someone in the company did the translation, maybe they used the cheapest human translator they found, maybe the first draft ended up accidentally on the package, etc. There really are million other options besides the machine translation scenario.

It is true that machine translators make mistakes, but the point is that no one should be judged without evidence. It’s good to remember that there are even worse alternatives than a bad machine translator. Namely, a human being who thinks he knows the language after 2 months of language studies as a hobby and one week vacation in the country. Right?

Free Trial: Try Multilizer PDF Translator for free!

Localize Your Software with MultilizerUnderstand foreign PDF documents quickly and easily by translating them with automatic PDF Translator. Download Free Trial!

5 thoughts on “Can We Always Blame the Machine Translator?”

  1. Good post, Iina.

    In my view, the answer is NO. We can’t always blame the MT.

    That is something that we wanted to make clear in the same post at TranSlate Translations, when it says that we didn’t “want to paint the automatic translation or wholesale translations as the bad guys”. We think that most machine translations are as good as their operator. MT has come a long way and they are getting smarter and more intuitive, yet at the end, they still require the knowledge of an experienced operator to take advantage of all their capabilities to their maximum potential and to infuse a bit of the quarks of the human language. Especially while localizing a translation.

    Now, looking at the progression of the translated words on that medicine label, was that we arrived to the conclusion that this was an automatic translation performed by an operator without the basic knowledge of Spanish. Most likely, they’ve used one of the custom enterprise software that nobody had bothered to customize. Of course there were other possible scenarios, some even too ridiculous to fathom.

    So chin up, MT is here to stay and getting better everyday. Even experienced translators use MT in one way or another. My point truly revolved around the irresponsibility of the pharmaceutical.

    As you can tell, I appreciate the opportunity of a nice debate…

    Cheers

  2. Thank you, Hermes, for your good comment.

    I truly agree with you that the pharmaceutical acted very irresponsible no matter which translation method they used.

    Machines are just machines and someone have to be the operator, just like you said. I think there’s something wrong if we try to humanise machine translators. The fact is that machine translators’ work is based (more or less) on translation memories, statistics and data analysis; they can’t think like we humans do. This means also that machine translators can’t act independently in a way that they would be responsible for any of their actions. I hope that the general attitude towards machine translation would change, and we all could see automatic translators as a tool which can help us in achieving our goals. It is a tool, and not an enemy.

  3. Machine translation is great but it’s never as trustable as the human translation! It’s true that some time translation memory just make translator become too much auto confident about what they are doing and not always they make their revision as they should.

Comments are closed.