This week's book giveaway is in the OCMJEA forum. We're giving away four copies of OCM Java EE 6 Enterprise Architect Exam Guide and have Paul Allen & Joseph Bambara on-line! See this thread for details.
I don't see why not. I assume your problem lies in breaking the (badly-designed) text nodes into pieces based on the semicolon delimiters?
What I would do is first to count the number of pieces in both of the nodes. If those numbers aren't the same then I would exit with an error message.
Then I would do a loop where K runs from 1 to N, where N was the number of pieces, extract the Kth piece from each of the text node, and output a suitable <LINE> element with that data.
If your problem is now that you don't know how to do any of those loops in XSLT, the way to do a loop in XSLT is usually via a recursive template. For example here's pseudocode for counting the number of pieces of a semicolon-delimited string:
 First you should not use tag name starting with xml or any combination of upper/lower case of the 3 letters. It is reserved by the recommendation. However, most/all parsers are very pardonning in this regard. But you should still try not to name tag that way which is not asking too much.
 I guess you should do that with xslt 1.0 if you are novice in xslt. The breaking string up into tokens are done through some recursive call of template to itself. I can show you how.
It is as simple if you are not familiar with the idiosyncrasy of functional programming. If you have doubts, you should break it up studying its components yourself.