August 7, 2008
Recipe for spiraling translation quality
I lot of translation agencies make unrealistic promises when it comes to translation memory. Here's one:
Cost savings
You will never again pay for previously translated sentences – irrespective of how many times these sentences reappear in different documents we translate for you.
To a customer, that probably sounds great. But let's think about this for a minute. If you work for this company as a freelancer, under no circumstances will you be paid for a translation found in their translation memory. What happens if the translation needs to be changed to fit the context? Who pays for the time needed to maintain consistency between existing and new translation?
Two types of translators are going to accept that kind of work: translators who don't mind working for free, so they fix and refer to the translations anyway; and translators who don't care about quality, so they just skip over the already-translated stuff and wing it on the new material. Translators who work for free are usually scrubs who are desperate for work and mostly not very good translators. Translators who don't care about quality probably aren't the kind of people you want working on your $100,000 globalization project.
So even if you started out with a very good translation (which for Japanese to English is unfortunately not a very good bet), with each succeeding generation of revisions your translation is going to deteriorate in quality, until you might as well have paid $0.02 per word in the first place.
There are situations when I don't charge for 100% matches: when an entire file or section is already in the memory. But if there are mixed matching and non-matching segments, then I charge a nominal fee for context checking, consistency, etc.
Overly optimistic claims like the one above are basically a way to try and dress up bottom-feeding, crappy translation with slick terminology. The results are the same, but a few clients are going to get burned before they wise up. And unscrupulous translation agencies are probably banking on that.
Hi Ryan,
Thanks for this post, I’ve thought about this a few times when similar claims came up. The context issue is really important as you point out, especially if you are talking at the level of sentences.
Ryan,
Good post. Soon after CAT tools were recognized as a translator time-saving tool, they became an agency marketing tool.
But in my experience, the people writing these type of claims and those producing the translations are in separate departments. Thus the project manager’s budget is often cut before he or she even begins the job.
Then, the PM often doesn’t (for lack of time usually) properly assess the job before sending it to a translator says “don’t even look at 100% matches,” resulting in the situation you’re talking about.
An agency concerned with quality has to realize that CAT tools can help save time and money — and those time and money savings can be passed on to the client — but the time and budget needed to produce good translations cannot be measured strictly by a Trados analysis. I think more conservative price ranges should be quoted to take into account potential problems. Sometimes the TM is great.
But each job is different and TM quality varies. And promises like these often constrain good translators — and production departments — who want to produce quality.
Obviously, an agency that does not make these kind of promises will need to offer more client education because, just based on numbers, the agency quoting free 100% matches will win the bid.
Thanks for the great response, Glenn. I love to get the agency perspective on these things.
(Also sorry about this late reply — I get email notifications when I get comments, but this one seemed to slip by)