Jan 25, 2009

Learning basic TM system concepts with little risk: OmegaT

Marc Prior, in various discussions on- and offline, has made some interesting points about TM tools. He is the project coordinator for OmegaT, a cross-platform, Java-based freeware tool, as well as a German/English translator like me with many years of experience. Recently he commented that these freeware tools, while perhaps not the ultimate solution for a translator, do in fact provide a means of learning about basic concepts which are important for all TM tools with no risk. While I certainly did not disagree with this statement when I read it, I also did not reflect on it and give it the weight which it deserved.

I work in a relative high-stress, high-production environment with a strong demand for my services that leads me to say "no" to project inquiries so often that some days I feel like a parrot. It's hard to find the time to work on every interesting thing which is offered, so ergonomics and efficiency are extremely important to me, and if I know that a tool like MemoQ or Déjà Vu X will make a translator 5 to 30% (allowing for varying situations) more efficient than some other tool, I will recommend them strongly and point out the ROI and how much money is lost through opportunity costs inherent in making other choices. All that is true, but...

... there is a more basic issue which Marc pointed out, which I also experienced, but it was so long ago that I had simply forgotten about it.

For someone accustomed to translation by simply typing in a text editor, dictating or overwriting text in some other environment, the basic concepts of CAT tools can take quite some getting used to. Concepts like segmentation, fuzzy matching, concordances, translation memories versus terminogies and more are not simple at first. Even though I picked up the theory quickly enough due to decades of IT experience, in practical work it took a while for some of it to sink in. Only after I thoroughly botched one of my first jobs with Trados by treating close fuzzy matches at 100% matches (missing the cue of the yellow field color vs. green field color) did I learn how critical it is to rely on information in the portion of the window where differences are highlighted.

All too often people buy software because they think they should, but they don't have enough basic familiarity with the ideas to evaluate what will really work best. They might follow the debates of the pundits and decide based on reason or fear to purchase a particular tool, but too often that tool languishes unused or is never used to its full potential. I've been using Déjà Vu for 9 years now, and I am still learning important ways to improve the effectiveness of my work with it, and the same is true of every other tool I use, such as Trados or Star Transit.

So while I still believe that a translator is better off getting a good commercial tool with the ability to process a wide variety of formats, including formats from SDL Trados jobs, I will give Marc the point and the match on this argument. If you understand so little about the basic concepts relevant to CAT tools that you might not get full value out of a 30 or 60 day trial period, OmegaT is a good tool to start with and figure out what people are talking about. I'm not particularly thrilled about the current interface of OmegaT, but it's not awful, and specific issues I have with it like the lack of tag protection (although there is, thankfully, a function to verify tags) and the apparent inability to have divergent target translations for the same source text (a real killer for me) do not nullify its value as a learning tool with professional potential. Setting up projects and adding resources to them are quite straightforward. If I were teaching a class on the basic concepts of translation memory tools and didn't want to hassle with demo versions for which participants had failed to get an evaluation key (this happened to me once - what a pain!), then I would probably use OmegaT.

However, I would be very careful doing professional work with the tool. There are many people who do, but most of them are well aware of the limitations and compensate for them in appropriate ways. I am not a particular fan of the Open Source ideologies, but I do appreciate projects with community spirit and value for education. This is one of them.

No comments:

Post a Comment

Notice to spammers: your locations are being traced and fed to the recreational target list for my new line of chemical weapon drones :-)