Tuesday, April 28, 2009
Translation management software (TM) use during the initial translation not only helps translators (including myself) work more efficiently, but also ensures a more consistent translation. However, the potential of TM software can only be fully leveraged if the original (source) text's terminology and structure are consistent, as well.
Translation management software matches text fragments (sentences, strings or entire paragraphs) with previously translated text. Unlike the search functionality in standard word processing programs, however, TM software allows for small variations in the text (different word endings, one or two different words in a sentence, etc.) Consider this example:
Sentence 1: "The dog in the house is brown and small."
Sentence 2a: "The cat in the house is brown and small."
Sentence 2b: "There is a small, brown cat in the house."
TM software will record sentence 1. If it later encounters sentence 2a, it will present sentence 1 as a "fuzzy" (inexact) match. The translator then only has to change "dog" to "cat" and can skip the rest of the sentence. But if sentence 2b is used instead , the software will not know that the two sentences are similar and will therefore not present the first sentence as a possible option. The translator must now translate the entire sentence over again.
This is particularly important if various documents relating to the same product are translated at different points in time. If sentence 1 occurs in document 1, and sentence 2a or 2b occurs in document 2, the translator may not remember sentence 1 when he or she translates sentence 2 weeks after working on document 1. While TM software permits the creation of custom glossaries, these will usually only contain key terms, not variations on minor word choices. While the translator may not remember sentence 1, the software will if sentence 2a is used, no matter how long ago document 1 was translated. The translator can then ensure a consistent translation for sentence 2.
Such use of TM software only works, however, if the translation memory from document 1 is used when document 2 is translated. If the same person translates both documents, he or she will likely re-use the translation memory. If a different person handles document 2, he or she may not have access to the translation memory used for document 1 -- indeed he or she may not even know it exists. This requires the same translator to be available for both documents. The best way to ensure that is to let your translator know that document 2 will exist and when to expect the source text. The translator then can schedule his or her other projects accordingly and ensure they are available to work on document 2.
To ensure an effective, consistent translation of your source documents, therefore, ensure consistency in both translators and source text.