Why a rigid approach to translation memory could be problematic

It might seem like a good idea to put your 100 percent translation memory matches in a figurative “don’t touch” box. Linguists and project managers wouldn’t spend time on those segments in the review stage of the translation process, theoretically cutting down on costs.

Yes, the whole idea behind translation memory is to increase your cost savings over time. But shelving those 100 percent matches can expose you to undesirable consequences. In fact, locking down those segments could add up to extra costs and quality concerns.

Let’s look at a few reasons why lockdown can potentially be a huge letdown.

1.  Language is always evolving

The nature of language, much like technology, is growth and change. For that reason, terminology should be updated regularly to make sure it continues to resonate with audiences. If part of your translation memory is locked down, you risk missing the target with your messaging—especially in the case of marketing content. Product descriptions on a website that remain the same year after year can become dated, turn off consumers and erode sales revenue.

2.  Possible rework costs

If your translation memory remains fixed in time, your translation quality can take a big hit. Or it might entail rework if the translation ends up missing the mark—costing you more.

3.  Expenses from importing

If you want to import translation memory from other vendors, you’ll probably face added costs to review and cleanse that resource.

4.  Hidden context means quality hindrances

Because part of the multilingual assets are essentially hidden from view, linguists can’t get a full picture of a document. They won’t be able to get a complete sense of the style or flow of the past translations, effectively threatening the quality level.

5.  Wrong segmentation, low quality

It depends on how the source content is formatted, but it’s possible that segments can break and lead to inaccurate matching.

For example, let’s pretend there are two separate segments:

  • Content
  • Distribution

In the locked-down translation memory, the phrase “content distribution” has been mistakenly broken into two segments. It is to be translated into Spanish.

Since the Spanish language entails a different word order than English, a linguist would put the Spanish equivalent of “content” into the second segment and “distribution” into the first. That means each word will be erroneously translated as the other one in the translation memory. For the next project, when “content” appears by itself and is translated as a 100 percent match, the resulting translation will actually turn out to be “distribution.”

That’s a serious error that can lead to customer confusion, re-translation and its associated extra costs, and lost productivity time.

6.  Mistakes gone undiscovered

It’s possible that when creating the original content, the linguists had little or no context. Referencing the additional context can lead them to see that the original translation needs to be improved or fixed.

7.  Gender differences in language

Gender used in segments might need to be updated if the initial sentences warrant it. But this necessary change can’t happen if 100 percent matches are untouchable. Again, this lets quality issues creep in.

Let’s say we have a sentence for translation that reads: Sales managers can access it on the intranet. The word “it” would have to be translated differently for German and Spanish because of gender differences. Not doing so makes for an incorrect translation.

The apparent savings can cost you

It’s understandable as to why you’d want to take your 100 percent matches out of the equation to save money. But keep in mind that doing so can actually entail extra costs, more work and quality concerns.

If you’d like more information or specific guidance, please let us know. We’re always more than happy to help with any and all translation needs.

Nathan Woods

About the author

ABC Translations