The drawbacks of translation memory tools

This summer, I’ve had a number of conversations with fellow freelance translators about some of the negative aspects of translation memory tools (CAT tools, TenTs, you know what I mean!). While I use and enjoy OmegaT and Wordfast, I agree that TM tools are not right for every translator or for every translation.

TM tools have some obvious selling points, primarily that they allow us to work faster and more consistently. I’ll never forget the first job on which I used Wordfast: it was a computer skills manual, and the final chapter consisted of fill-in-the-blank questions that were drawn word for word from the previous chapters. Using Wordfast, I think it took me an hour to translate that last chapter, which was about 3,000 words long. Without a TM tool, that task would have required an excruciating amount of copying and pasting from the material I had already translated. Right there, I felt that Wordfast was worth every cent I had paid for it.

I also think that TM tools have some ergonomic advantages. If you’re working between two documents, it’s very tiring on your eyes to be constantly glancing back and forth between them as you lose and find your place over and over again. Less typing is obviously better for your hands, and the colored text fields that many TM tools use are much easier on your eyes than a glaring white screen with black type. I have a tendency to read too fast, and I like the fact that the TM segments force me to look at only one sentence at a time; I find that when using OmegaT or Wordfast, I can also work about 10% faster even if the document does not have many repetitions.

Some of the disadvantages of TM tools are obvious, and some are more subtle. It’s easy to see that match propagation (automatically applying a given translation throughout your document or project) also means error propagation. It’s horrifying enough to receive a reference document from a client that includes a serious translation error, but it’s far worse to think that that error has been enshrined in the client’s TM database and provided to every translator that the client works with.

Over time, and after attending Translate in the Catskills and other writing-focused workshops for translators, I’ve also come to believe that TM tools often impede good writing. The segment-by-segment approach has its advantages, but it definitely does not encourage smooth writing that flows throughout the document since you’re working on only one segment at a time. In some TM tools I’ve demo’d (not OmegaT or Wordfast), your translated segments  even disappear from the screen once you approve them, which makes it almost impossible to create a text that has a cohesive tone. TM tools also make it difficult to reverse the order of words or sentences, since you then end up with matches that don’t match up. The TM tool has no way of knowing that you decided to restructure a paragraph for stylistic reasons, it simply matches your source and target sentences and assumes that they mean the same thing.

Since I’ve spent this year focusing on the quality of my writing, I’ve changed my workflow for jobs on which I use a TM tool. Instead of proofreading directly in the tool’s interface, I clean up/compile the translation, then proofread from that target-only document, then input my changes into the TM tool interface. This takes longer than proofing directly in the tool, but I’ve found that when I’m proofing mostly for style, tone and flow, it’s nearly impossible to work in one segment at a time.

Any other TM pros/cons/best practices out there?

14 Responses to “The drawbacks of translation memory tools”
  1. Jill (@bonnjill) August 20, 2010
  2. Corinne McKay August 20, 2010
  3. Audrey August 20, 2010
  4. Caitilin Walsh August 20, 2010
  5. Val Ivonica August 20, 2010
  6. Tess August 20, 2010
  7. Kevin Lossner August 20, 2010
  8. Riccardo August 23, 2010
  9. Karen Tkaczyk August 26, 2010
  10. patenttranslator September 12, 2010
    • Corinne McKay September 14, 2010
  11. Amy N September 16, 2010
  12. Martin March 22, 2011
  13. Penny March 21, 2018

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.