The trouble with translation memory programs October 1, 2009
Posted by Jill (@bonnjill) in Business practices, Tools, Translation.trackback
Here we go again…
There is a very interesting article in Ezine, Translation Memory Programs Causing Problems For the Translation Industry and Also For End-Users by John Hadfield, that is arguing that the use of translation memories is in fact slowing down translators and driving the price of translation up as a result. He starts off by claiming, “Under this system, the translator often has more work to do as a result of using a TM, but gets paid much less for a particular translation than he or she would have been paid before the introduction of TM’s.” I don’t know about that, but I do know that I have to really pay attention to every segment (because not all matches are true matches) and spend a lot of time figuring out how best to make the translation units flow together and not sound translated (a final run-through after cleaning up the document usually helps tremendously). He also states that many translators dictate their translations. I don’t know if many translators do this, but I do know a couple translators who dictate their work. He makes some very interesting points including:
…The most glaring result of this problem is that all translators have been forced to increase their standard price per word over the last few years in order to survive, so for documents which are almost totally non-repetitive (and where a TM is therefore useless), the translation agency or end-user ends up paying much more for its translations than it would have paid before the introduction of TM’s. However, that same customer still requires the translator to use a TM for its translation, even though it is obvious to all that the document concerned is not likely to show any repetition in any but a few random single or two-word phrases.
I don’t know if my raising my rates has anything to do with the use of translation memory. I have raised my prices to keep up with inflation. Also, as with any profession more experience should always be compensated with a higher salary, bonuses, etc. As freelancers we don’t have that luxury. I charge what the market will bear. I certainly don’t do it to survive because I am faced with discounts for repetitions and matches.
…Apart from the translation of manuals which use a great deal of repetition (such as workshop manuals, job code manuals, etc.) and certain standardized contracts and legal texts, statistical analysis of any large company’s or large translation agency’s translation work over a period of one year would very probably show that the compulsory use of TM’s, combined with the resulting increased prices per word from freelance translators (who perform by far the major portion of translations throughout the world), has finally resulted in the entire operation costing more to the end-user than it would have cost before the use of TM’s became general.
…There is also increasing evidence of a curious attitude prevalent amongst certain end customers and agencies in which the method of translation (i.e. the use of the TM system) seems almost to have become more important than the translation itself.
I’ll give him that. A common complaint I hear on all my translation forums is that many TMs contains mistranslations, sloppy work and out-and-out errors, which are then perpetuated in the company’s documentation for all eternity. Since we are not paid to correct the TMs and agencies are told by their end clients to not touch the 100% matches, these mistakes are usually not pointed out and the end client is usually blissfully unaware of the problem.
The author is proposing to require the agency’s translators to offer a reduced price per word for translations which do not require the use of a TM, and perhaps abandoning the use of TMs altogether. That is a very interesting suggestion, but I think I’ll stick with my word rates and my TEnT for now. It does save me time in texts that are repetitious, and I generally work with agencies that do not demand discounts for repetitions. I bought my TEnT to save me time and ensure greater consistency. And every once in a while I do get a plum job that is already in my TM, which saves me time and frees me up to take another job.
Translation memory is not a burden to be vilified, but I do think translation agencies and some TEnT developers should think long and hard about their practices. We pay for our tools – not the agency. Why should we be paid less to use them? That just doesn’t make sense. And don’t even get me started on selling certification classes to use the tools. Those certified users are now out in the cold and out several hundred dollars because the latest version is totally different from the one they were trained on.
I’m curious to hear what you all think. Does using a TEnT take you more time or less?
Musings from an overworked translator 
Awesome post! I think that the important thing is to keep both sides of the coin in mind when you use your TenT. I use either OmegaT or Wordfast for pretty much every job that I can; on the one hand, I think that I’m much less likely to skip a word/phrase/paragraph when the text is broken into segments, it’s easier to keep terminology consistent and I find that I work about 10% faster even if the document contains no repetitions, because I’m not constantly glancing back and forth between the source and target documents. On the other hand, I am really trying to focus on improving the quality of my writing and producing translations that sound like original English documents, and the TenT is not conducive to that because it forces you to focus too much on the segment and hand and not on the overall flow. As you mentioned, I think that doing your editing after you clean the document can really help.
I started my career in localisation in 1992, before the arrival of TM tools as we know them now. I remember very clearly how we dealt with updates to user manuals at the time. We made ‘diff docs’ for each file: versions of the updated file with change tracks showing which text had been deleted, added or moved. We then had to take the old version of the translated document and implement all of those changes manually, including all formatting etc. This was very time-consuming and error-prone.
When I first saw a Trados demonstration, probably around 1996, I was sceptical. But I became an enthousiastic user as soon as I saw how quickly and efficiently Trados handled those very same user manual updates.
In my line of work, with controlled language and lots of updates, TM tools definitely speed up the working process. Yes, much depends on the quality of the TM (GIGO – garbage in, garbage out), and checking matches of any kind does take time, but that is outweighed by the advantages of speed and consistency. End-clients are certainly not paying more than they would have without TMs!
I use Wordfast for the first draft of my translation. Like Corinne, I use it to make sure that I don’t skip sentences during translation, not for the TM itself. Once I have finished my first draft, I clean the document and then work solely on the English text, re-drafting it for a more concisely written translation with a better flow. I do a final check through against the original document before submitting it to the client. I would estimate that I spend 30% using the TM and 70% using Word and printed documents.
So do you get paid full word price?
Depending on the project, and assuming there are no client imperatives concerning the use of a specific TM, I use OmegaT but always do a final readthrough to have a good grasp of the text as a whole. I find that TenT do help me work faster in most cases for more technical translations.
For marketing documents, however, I generally do not use any TM, preferring the old method of comparing with previous translations for the same client when consistency is needed and laying stress on style, inventivity and catchy turns of phrase for the whole text. The same goes for academic translations, for instance: for some types of text, overall style and flow do matter much more and a translation per segment would yield a clumsy result and slow down the translation process unnecessarily.
Good post. Great comments.
TM looks to me more like a management instrument than a translation tool. Its successes, on the other hand, look suspiciously similar to a side effect.
One nit, though:
http://ezinearticles.com/?Translation-Memory-Programs-Causing-Problems-For-the-Translation-Industry-and-Also-For-End-Users
should be
http://ezinearticles.com/?Translation-Memory-Programs-Causing-Problems-For-the-Translation-Industry-and-Also-For-End-Users&id=2966446
otherwise the link is broken.
Your post is music to my ears! I do mostly creative, marketing-type translations which are completely non-repetitious and wholly inappropriate for TM, but that hasn’t stopped clients sometimes trying to make me use it.
I think the whole translation industry is becoming overly dependent on a tool which has its uses, but also major drawbacks.
Incidentally, I did a survey of 170 people at the ITI conference this year, and only around 40% of them used TM – so translators are resisting all the hype.
TMs are like automobiles. It’s the driver that decides which way to go, not the car. Likewise, translators have a choice to use or not use TMs depending on the type of translation task. I agree with translatormum’s comment because TMs, if used right, can save translators lots of time especially in large, technical projects that require collaboration with other translators.