Is cutting costs by completely ignoring perfect matches really a good idea?

CAT tool developers often highlight the capability of their tools to save on both time and cost. There is some truth to that, but it can’t be applied to every kind of text or data format. Some companies who use these tools have now made it their policy that translations from so-called “perfect matches” (depending on the tool also called context matches, ICE matches or 101% matches) can be used for each new text to be translated without prior revision. However, this is an error in judgment which can result in a massive decrease in quality – and we’d like to explain why.

First, let’s define perfect matches: these are superior 100% matches since not only is the phrase translated in the exact order and wording, it also relates to the same context of the document. This is supposed to guarantee more reliable translation results.

Using translations of individual phrases or sentences from earlier projects without prior revision can lead to the following problems:

1. The writing style of the documents which the translations are taken from often varies (e. g. a hip-sounding marketing text vs. plain technical documentation). This results in an inconsistent style mixture in the new document.
   
   
2. The terminology of the company may have changed in the meantime. As a result, a term can now have two or even three different translations or the translation of a UI term in a software tool is no longer consistent with the current version of the software if the translator doesn’t review the translations. Too many different terms can also confuse the reader.
   
   
3. The document which the suggested translations are taken from without further revision uses a different form of address than is required for the current translation (e. g. in German one text uses “Sie” and the other uses “du” to address the reader). If left unchecked, both forms of address are now used in the new translation, which is simply unacceptable.
   
   
4. Some problems may have occurred during the processing and segmentation of the documents, leading to some sentences being split into two parts. Due to the different grammar and syntax the original text and the translated text now no longer match content-wise. If, in addition, perfect matches are used without revision, this can lead to false translations in the final document. Example:
   Segment 1a: Delete                                        Segment 1a: Datei
   Segment 1b: file                                             Segment 1b: löschen
   
   The reverse word order results in the translation memory translating “Delete” with “Datei” and “file” with “löschen”. This may have been correct for the original document but in the next one this will likely cause a false translation.
   
   
5. The perfect matches contain errors – for example obsolete terminology, spelling mistakes, translations of terms that should have been left as they were, or falsely localized URLs.
   
   

In light of all these reasons we strongly advise against placing blind trust in perfect matches. They should always be reviewed in their specific context and adjusted accordingly, if necessary.

Last but not least: If your translator is not paid for perfect matches due to your decision and budgetary reasons, he is, judging from our own experience, likely to ignore them for (their own) budgetary reasons. As a result, the translator might ignore the context and you’ll receive a less consistent or accurate translation – meaning that while you might save some money, you’ll also be getting a product with diminished quality, i.e. you lose precisely the competitive advantage which you were hoping to acquire in this developing business.

Any questions?
Just call us

+49 89 67989548