In 1997, when TMX, was first proposed, there was no real way for companies to utilize translation memory data created in one computer-assisted translation (CAT) tool as a resource for another CAT tool. Customers and users were, by and large, tied to the tools in which their TM data was stored, since changing tools would mean walking away from the accrued benefit of their TM investment. Although few companies at the time recognized the need for a TM exchange format, visionary thinking by the original OSCAR committee, and the dedication of Alan Melby and Yves Savourel, among others, led to the development of a standard that is now seen as critical for protecting an organization's localization investment.

As demonstrated by the LISA 2002 Translation Memory Survey, companies have increasingly large TM assets that need to be usable both today and in the future. TMX certification represents a real step in giving companies the assurance that today’s investment in language technology will be valuable in the foreseeable future. LISA wishes to congratulate OSCAR and SDL International on reaching the milestone of the first TMX certification.

Tests indicate that users can expect to retain between 70% and 99% of the value of their TM data when using TMX to move between tools. This degree of reusability ensures that companies will not be bound to a particular product and left with large amounts of useless legacy data if they decide to change tools or need to use different tools at the same time for related projects.

The potential financial losses now associated with the inability of translation memories to move across products and platforms will no longer be an impediment to companies making the investment to use them. Global Business can now ensure that their linguistically based Intellectual Property is safely under their control.

Despite this major achievement, work remains. Fortunately the issues that currently prevent 100% reusability between tools are understood. According to Paolo Vanni of J.D. Edwards, “The next big hurdle confronting TMX, and translation memory exchange in general, involves the establishment of standards and rules around issues such as tokenization, filtering, and segmentation, which today jointly account for most of the loss experienced with TMX.” Under Vanni’s leadership, OSCAR’s Segmentation working group is currently exploring ways to address these issues to make TMX even more useful in the future.

All in all, a lot of work has gone into TMX since it was first conceived in 1997. The success of TMX and, now, of the TMX certification program, point to the future and show the GILT community what can be achieved if it becomes involved in the development and implementation of standards.