Thanks for the warning. My general method is to take the XML text into BBEdit and use find/change to strategically insert tabs and line breaks. Then paste the whole thing into Excel, which creates a column of open tags, a column of text for translation, and a column of closing tags. The translator works with the middle column. When I get the Excel file back from the translator, I save as tabbed text, remove the tabs in BBEdit and have a translated XML file. It works pretty well, except sometimes there's an invisible, unselectable character between > and < sometimes, which foils my find/change because it is not “found” so doesn't match my search string. However, the sample I was working with was not so long that it wasn't worth the time to manually fix those instances. I imported the translated XML file back into CS4 and it worked perfectly- 50 pages of English became 50 pages of French, fully laid out, in three seconds! We were amazed.
I have actually done some tinkering with Story Tweaker, as part of the methodology search for this very project. The translators were very hesitant to check it out, though.
Thanks Johan for the WordPress link! I will check it out now.
-Glen