Jost Zetzsche, a leading proponent of translation environment tools (TeNTs), wrote an enlightening piece about his experience in Accurapid's Translation Journal last month.
As his personal translation memory grew and grew, Jost found its usefulness "seemed to decline rather than improve. Too much time had passed between the earlier projects and the current ones (...), language had changed and my skill levels had, too." Gradually, he found he was spending "a lot of time deleting or wading through useless suggestions from the TM." With the newer TeNTs offering subsegment as well as segment matching, the problems were getting even worse.
His conclusion? Gigantic vaults of bilingual data are fantastic as reference materials, "but I am just not sure about their value as translation memory data in the classic sense." My conclusion: this is the year to purge the dross building up in our own bilingual resources, such as wobbly source texts and translations the client didn't like (yes, this happens!). Our resources may be carefully vetted, but they're never vetted carefully enough!
And hats off to Jost Zetzsche for his realistic analysis in a world obsessed by the power of the crowd and the value of the terabyte.
No comments:
Post a Comment