;   Medical Translation Insight: Improving TM maintenance - ForeignExchange Translations

Improving TM maintenance

Improving TM maintenanceTranslation memories are important in helping to maximize quality and efficiency in translation. Consequently, ensuring good practices in using and maintaining translation memories must be vital to any translation company.

One of the interesting aspects about this is that it's not just a one-time effort. At ForeignExchange, we regularly examine our internal processes and procedures in order to identify opportunities for enhancement. As an example, we recently reviewed our standard procedures of how we use and maintain our translation memories (TMs), and we are now working on implementing improvements to our TM maintenance process.

Previously, it was more or less standard procedure to simply create project TMs upon project shipment and close, and then merge these segments into our master TMs to coincide with all earlier segments. In this way, we could capture every translation done for a client, documented with the project number, stage of translation, and any other relevant identifying information. If updated unclean files were not available to do a cleanup, then a post-alignment was generally done to capture the final translations. In either case, any translations done would usually be present in the master TM. Maintaining a complete history of translations was the primary goal of this method.

After some analysis, and more than a few internal discussions, we realized that the significance of translation memories in our work merited a more practical approach to our general maintenance procedures. Keeping a history of all translations was fine, but it was quickly apparent that much more could be done to derive further benefit from our TMs and add value to our translation processes.

Among the improvements we are now pursuing is the maintenance of our master TMs. To offer some transparency, we plan to prioritize client-approved segments by overwriting non-approved segments rather than simply merging everything to keep a composite history of work. While this will certainly vary to an extent, depending on the client's requirements, we believe prioritizing these final approved segments will reduce the number of duplicate entries and inconsistent translations. This will, in turn, lead to higher-quality TMs and higher-quality medical translations.

A second major improvement involves the updating of bilingual files at every linguistic step. While post-aligning final files sounds good in theory, it often creates a backlog of work and cost at project-end, resulting in delays in updating TMs for future projects. Furthermore, aligned files pose their own set of risks apart from unclean files. As some of our clients have come to require a final bilingual output (which we create from updated unclean files), we are moving into processes which require the updating of these bilingual files concurrently with each linguistic step. This greatly reduces the turnaround time in updating our master TMs, and also provides us with a simple means of re-creating TMs should they ever become corrupted or need repair.

As with any process change, it will likely take some time to truly see the results of this effort. Improved methods of checking updates and validating correct entry of segments will also be implemented. But perhaps most importantly, by further educating ourselves in the various uses and functions of translation memories, we hope to provide greater benefit to vendor and client alike.

For more about ForeignExchange's TM practices, take a look at the following articles:


ForeignExchange Translations provides specialized medical translation services to pharmaceutical, biotechnology, and medical device companies.
 
 

2 Comments:

  1. Oliver Lawrence said...
    An interesting article, and hats off for a commitment to quality that, as a translator, I'm aware that sadly not all agencies share.

    A translation that is not approved in one context may potentially actually be a better translation when the same source segment occurs in other contexts - I would be interested in how you plan to get around this.
    peter wilms van kersbergen said...
    I salute the intent to improve. But from our experience, I also have the question how you plan to execute on maintaining TM content.
    At Medtronic we have a policy to only maintain TMs internally, and we require any vendor or freelancer to return translations, which are then validated by an internal linguist before they go back into our master TMs.
    We maintain separate TMs per different business unit, and per business unit we use sets of attributes to track different translations per different source context.
    That is, however, not always enough to keep the memory clean and up-to-date. How would you deal with this, I wonder? Any best pratices we can share?

Post a Comment





 

Services | Resources | Company | Contact Us | Blog | Home

(c) Copyright 2010, ForeignExchange Translations, Inc.