Home Languages Articles Links Downloads About Contact



Other Topics

bulletTranslation prices
bulletMachine translation
bulletHarry Potter
bulletTranslation quotes

Free website Translation Service



Memories for Sale

In something that might sound like an idea from Philip K. Dick's story We Can Remember It for you Wholesale, translation memories have are becoming a commodity unto themselves. How this emerging market will evolve and what translators and their industry can expect of it is an important question.

Translation memories, for those not in the know, are computer files containing bilingual databases of terminology, text strings, and even entire sentences produced based on translations of documents or terminologists compiling such resources. Any translator working in any environment can produce such files simply by using Machine-Assisted Translation (MAT) tools such as Trados or Catalyst. The important development is that these files and their contents are now available for purchase. In other words, you can buy the work, that is to say the translation memories, created by another translator and use them for your own translations.

This leads to all manner of issues. The first is quality. In other word, who you gonna trust? Since the businesses involved in making translation memories available for purchase do not perform rigorous quality checks, it's a caveat emptor market, with anyone considering such an acquisition having to rely on user reviews, if there are any, the reputation of the translation, if that can be found out, or their own check of the translation memory, which would be time-consuming and costly to perform. All translators have colleagues they know well and consider reliable sources of information for terminology, style, or subject matter analysis. Exchanging information in a variety of forms has been a common practice for decades, perhaps even millennia. But spending money for an unknown translator's memories raises quality issues have yet to be reliably addressed.

The second issue is relevance. Without actually seeing the content of a translation memory, a potential buyer cannot really know what is inside it, and so has to rely on descriptions from the seller. Even assuming the descriptions are thoroughly detailed and bereft of marketing hype or overt deception, there is only one way to be certain of the content: get a look at it. But doing that is potentially time-consuming, and more importantly, would allow buyers to get the information they want without paying for it. The problem of describing content exists in all information-based industries, but is particularly problematic here, since there is no standardized manner for the descriptions. A patent, for example, is a legal document, but its content is typically technical in nature. Further, a patent that could be described as technical might be about a new pharmaceutical, medical instrument, RAM chip, or semiconductor manufacturing process. In other words, without a precise, standardized way of describing the content of translation memories, the would-be buyer may be facing an unacceptable risk.

The third issue of importance is intellectual property rights. Who owns these memories? An in-house translator works under the same conditions that all knowledge workers do: their work and products derived from it is all owned by the company and cannot be used or sold for other purposes. Many in-house translators do however moonlight in the freelance world, using their knowledge and at times resources from their day job to complete other translation work in the evening or weekends. Translation memories produced in this process could arguably be considered, at least in part, as belonging to the employer rather than the employee, thus making sale of them illegal. The situation for full-time freelancers is even more complex. If a freelancer compiles a translation memory based on work done for several clients, with these clients variously contributing their own memories, terminology lists, glossaries, or reference material, then who owns the translation memory the translator creates? This is a sticky issue that will likely result in some interesting lawsuits in the near future, and for now must be worked out on a case-by-case basis.

The sharing of translation memories, whether as a gesture of cooperation or as a business activity whose end is profit, is in principle a great idea. Translators can benefit from others' efforts, avoiding reinventing the wheel while working, and the potential utility of MAT tools can be leveraged that much more. But what is in principle good is often in practice quite difficult. Until the issues of quality, relevance, and intellectual property are worked out, this market and its products are likely to remain small and offer little to translators or translation companies.

Back to Machine Translation. top