Whenever we consider a freelance translator for establishing a relationship, the first thing we want to know is whether a candidate is proficient with any translation memory software. Although other factors such as experience or specialization are of paramount importance, too, this question comes first because our translation process always includes revision by a second linguist and automated quality assurance, which makes it absolutely essential to receive a translation in some kind a bilingual format. If we were just looking for a translator to do a one-time job, a TM tool wouldn’t be that critical, but it’s our policy to build long-term relationships where a more sophisticated and consistent approach to translation management is called for. An additional benefit of proficiency with TM tools is that an individual who uses a TM tool routinely is often more professional and disciplined in other aspects of their work as well.
Now, although a computer-assisted translation (CAT) tool is primarily designed to increase productivity and quality of translation, it’s our experience that freelance translators aren’t really racing to adopt this technology. Some just don’t use such tools while others—probably, finding themselves under increasing pressure to use one—do indicate some tool on a résumé, but when push comes to shove, it turns out they have little or no experience with it.
Truth be told, I simply don’t get this. I do admit that some of the reasons behind the resistance are valid, but what I don’t understand is why focus on problems? Don’t the pros far outweigh the cons? Personally, I believe that just a single benefit of making your translation much safer by using a TM is already a sufficient reason to employ a CAT tool since it protects you from accidentally doing something unwise. For example, when I was just starting out in this industry, it took me just one month to find out that working without a TM is extremely unsafe—and I learned this the hard way. Together with a colleague, we took up an urgent 5,000-word translation project and translated like crazy for 12 hours straight from a PDF directly into a Word document. Just when we began seeing the end in sight, I accidentally deleted the entire text in the file, closed the file, and saved it, effectively destroying a day’s worth of work. After we had contacted the client to apologize and ask for a deadline extension, I made a promise to myself to never translate anything beyond a few sentences without a CAT tool.
But let’s look first at the arguments I somewhat agree with so that an as avid supporter of translation memory, I don’t seem too critical from the outset.
Potentially Valid Arguments against TM
“I don’t use a TM because breaking a text into segments destroys the natural flow of a text”
I think this is the most important and valid objection. Within a TM tool, you deal with separate, typically sentence-based segments rather than a text as a whole. In a compelling blog post, German to English translator John Bunch distinguishes between the best translators who are writers using a source text as a starting point to create a new text and inexperienced translators who are just decoders rephrasing the original text with the words of another language. Sentence-based segmentation can impede a creative approach by forcing a translator to focus on translating separate pieces and lose sight of a text as a whole, thus turning into a “decoder.” As a result, a translation may fail to look as an integrated whole crafted by a skillful writer, but rather appear as a set of disparate sentences with little or no logical or linguistic coherence between them.
I completely agree with this objection and instead of arguing, I want to offer a few ways to mitigate this issue. First and foremost, a translator must be aware of this challenge and competent enough to tackle it appropriately. This means that if you are a translator, you must build your skills in this area and if you are a translation buyer, you need to look for a highly competent vendor. It does sounds common-sense, but we all know that what’s common sense is not always common practice, so this is another opportunity for me to reinforce that a translator must be professional and responsible, which I constantly do on this blog.
Second, a TM tool usually provides one or more internal features to surmount this challenge as well. For example, you can always enable a paragraph-based segmentation instead of a sentence-based one. By having an entire paragraph in a segment, it’ll be easier for you to make a translation more coherent. Another proven method is switching to another view that displays the translated text only. By doing so, you deal with your translation as an integrated whole without getting distracted by the original. This allows proofreading the entire text for broken logical links between sentences. We do this all the time, and it works.
Finally, I believe this issue will become generally less important as translators increasingly work on texts where segmentation is irrelevant. For example, about 50% of our workload are texts with little or no coherence between segments such as software strings, user manuals, competency descriptions, just to name three. With these types of content, it doesn’t really matter if a text is segmented by sentences or not segmented at all because you end up translating separate sentences anyway—irregardless of whether you use a TM tool or not.
Obviously, my opinion is heavily biased because I’ve been using these tools throughout my entire career as a translation professional. So please go ahead and correct me if I’m wrong with any of the assumptions.