Copyright © ProZ.com, 1999-2013. All rights reserved.
Customers with documents undergoing regular changes are a very attractive business: they come back for more. In such cases you have an outdated version of the document available, in both languages, both source and target. You also have the corrected version of the document in the source language and the order to create its translation.
The article shows a straightforward way to proceed in such cases. It describes the procedure using TRADOS on Word files. The method, however, is applicable to any combination of program with a translation memory and an editor that allows the user to embed corrections.
In some cases we may have the original as well as target version available, as a bilingual doc file and/or translation memory. In this case the new order just means you will first pretranslate the new version and then go through the changes. However, having the new version in the explicit form of "old + changes" throws in a wrench first and second adds a definite benefit or two to the job.
In real life it may not always be that easy to start off. One of my real cases involved a PDF file containing the translation of the outdated version and a DOC file with corrections to the original. In this case I had to recreate the outdated status first. I followed the following procedure:
Load the file with corrections into Word.
Reject all corrections and store the file as myfile_source_old. I used the name of the project for myfile and stored the file in a different subdirectory to avoid any future misunderstandings.
Convert PDF target file to Word-readable form and name it myfile_target_old.
I use nitro PDF (http://www.nitropdf.com) for anything to do with PDF files, so exporting was no problem. When worse comes to worst, you may be forced to do OCR on PDF documents, consisting of scanned printouts. By the way, this is the very reason why I always ask to see examples of PDF files, before committing myself: a word scanned costs me double or triple of a word read)
Use WinAlign to align myfile_source_old and myfile_target_old. Export the resulting file pair from WinAlign.
Create myfile translation memory using the exported file.
Check the result by pretranslating a copy of the myfile_source_old: it should result in (or be close enough to) myfile_target_old.
Updating old to new
At this point we have a translation memory available reflecting the previous situation. Also, we have the new version of the document with changes to it. We proceed as follows:
Accept all changes.
Store the file as myfile_source_new
Exit Word to pretranslate the created file.
Use Word to translate the new and/or changed segments.
If you expect to have more of the same coming, it may make sense to introduce an attribute field in the translation memory, specifying the version of the document, that the segments introduced pertain to.
Having myfile_target_old and myfile_target_new allows you a nice feedback: you can ask Word to look for changes between the two files and create a new file ("old with new changes") that should look pretty much like the marked-up file the customer sent you to translate. Comparing the two can help you find inconsistencies and spare you later corrections.
Reading one more time through the article I got a sinking feeling, that I am proudly explaining the design and workings of a wheel. What else is the translation memory if not keeping the known stuff in mind and adding the previously unknown? In other words, what's new?
For me, the fact the documents with marked-up changes fit so nicely into the process, was a real aha. And, dear reader, I hope it is an aha for you as well.