Monday, January 28, 2008

Myths and legends about translation tests

  • Be wary of translation tests: agencies may use them to piece together a translation for free.

    This, as far as I can find out, is one of the favorite urban legends of translation. In 24+ years of professional experience I've never seen any evidence that something like this ever happened. Even if it did, the agency in question would soon disappear: the quality of the resulting translation would be so bad and uneven that any customer would soon flee.

  • I don't want to do a free test: I'll send them some sample of previous work of mine, and they can evaluate that.

    You can try, but usually whomever sends out the test does it for a purpose: comparing candidate translators to one another - something that you cannot do with translation samples of different originals. Also, bear in mind that a well designed translation test does not only test the quality of the translation: it checks how well the candidates followed the instructions received.

  • I don't want to do a free test, they should pay for it!

    Go for it, if you can get it. The most likely outcome is that you'll just exclude yourself from the selection process.

    I used to work as a manager in the translation department of a large software company. To evaluate candidates for staff positions (well paid staff positions), we used tests to compare the quality of the translations done by the candidates, and how well such candidates would follow the instructions received.

    The first tests were unpaid. Those who refused to do them were just removing themselves from the selection. The company was not being stingy (we invited the candidates who passed the first screening to the company's HQ, all at the company's expense), we just were not interested in people that would not invest a couple of hours of their time to show they were interested in working for us.

Having said that, I recommend against doing free tests longer than reasonable (say, 250 to 500 words maximum). Probably an experienced translator could do away with tests, or maybe limit them to only very good prospects.

But, as I was saying, just refusing to do tests is a quick way to remove oneself from the selection process.


  1. I agree with most of your points, except that in point 1 the agency may have a project manager who would take care of bringing uniformity to the various pieces of translation. So, it's not that unlikely. Besides, the situation you don't acknowledge in point 1 recently happened in Brazil, and I heard of it in a discussion list for Portuguese translators.

  2. The operative words are "heard it": for me it remains a urban legend because I have never seen evidence of it happening - it is always hearsay. To be convinced I would need to see actual evidence: different tests for the same job, and then a translation that is made up of those tests (or close to it).

    Even if it happens, it must be exceedingly rare: yes, an agency could use a project manager to smooth out the various pieces of translation. In practice that would be difficult even if all the pieces were of acceptable quality. It is likely, though, that their quality would range from good translations to bad ones to "translations" done with Babelfish. Editing a patchwork translation like that, especially a long one, would be a real nightmare.

  3. I once received a translation test that was three pages long... Give me a break!!

  4. Three pages would probably be about 750 words or so... way too long, I agree.

  5. I haven't done a translation test in a long time. I've been paid to grade quite a few, which is always very interesting (they always run the gamut from great to terrible). I've been translating for 10+ years and have plenty of academic credentials and published books to show for myself. I also regularly turn away work these days. Any agency that wants to work with me can take a chance and pay me to do some short job for them, which they can then evaluate. There's really no need for a test! And there's certainly no need for those ridiculous application forms some agencies send out where they want you to list all the versions of all the software you support.

    The publisher of a magazine I worked for years and years ago once told me something very illuminating. He was looking at resumes for his replacement and he said he just threw out the resumes for people who entered their salary requirement. He said once you reach a certain level you don't put down what you earned at your last job and expect to get slightly more at the next job. You show what you can do and you demand to be compensated accordingly or you don't take the job. That's kind of how I feel about translation tests.

  6. I agree with almost everything in this blog-post and I’d like to add this: most serious translation clients (translation agencies or other) assess translation tests against a standard quality matrix. For this reason, translation tests have to be standard texts. For instance, at translation agency I work for, we have different translation tests for different subject matters but all “medical” translators translate same text for test and that standard test is rated against a quality matrix, enabling us to compare translators.

    If one can’t invest a few hours for a translation text, I (as a PM) wouldn’t be interested in collaboration anyhow. Speaking of time, who's going to respect PM's TIME? :D

    Thanks for good reading,

  7. The PM's and the translator's time equally deserve respect. This raises another question: if time is of the essence, wouldn't it make more sense to give small paid jobs (as opposed to test translations) to new translators?

    This would be great for the agency as well - instead of running around like a headless chicken, desperately trying to find the right translator for a large project due yesterday ordered by an important client, and taking the risk that the as yet unknown translator messes it all up, the PM would, by assigning small jobs to new translators, have more assurance that by the time the big project rolls in, they have an established relationship with a trustworthy translator. And this way, the translator gets paid for the "test" so he's happy, too. The bottom line is, everybody is more inclined to work together and the end client has much better chances at walking away with a quality translation - and coming back to the same agency the next time around.

    Another alternative not mentioned in Riccardo's post is paying for the test and then deducting the amount of the payment from the payment for the first job. Of course, this would have to be agreed in advance. Fait for all parties involved and no financial loss for the agency.

    I am sure we all know how many free translation tests it takes only to end up getting a measly job a year later, worth a whopping... $20!

  8. I am increasingly unwilling to perform tests other than, as you say, for very promising prospects. I just feel I have built up a sufficient reputation and customer base not to have to do "free work", for which I can ill afford the time. Perhaps it sounds arrogant, but I think I am in a position to pick and choose the jobs I do to some extent, and I think that has got to be a goal for the professional freelance translator.

  9. I am both an engineer and a translator. If, as engineer, someone asked me for a small free test project, I would laugh at his face...
    Very good blog, by the way.

  10. Riccardo, just found this great post and discussion.

    Test translations are one of those eternal topics. Fundamentally, I think your advice is right-on: avoid tests if you can but beware that this may remove your name from consideration.

    We recently wrote about our experiences with test translations at Medical Translation Insight.

  11. There is another point in doing test translation! You never know who is checking and evaluating your work. This job can be assigned to unfair editor who will intentionally discredit your work. This also happens sometimes.

  12. I agree with Alejandro, fantastic blog!

    I think tests are necessary, but what we simply have to be wary of is carrying out tests that are over, I would say, 300 words. If the correct text is chosen for the test, I believe it to be the best way to see a translator’s potential.

    It is not necessary to ask a singer to sing for an hour to see if they can sing or not, after singing one verse, it is clear. The same goes for translation.


Thank you for your comment!

Unfortunately, comment spam has grown to the point that all comments need to be moderated. All legitimate comments will be published as soon as possible.