processing text nodes one character at a time

I am transforming xml data into our type-setters proprietary data
format. I use XSLT 2.0 for the purpose. In special cases, I need to
process text nodes one character at a time: the string must be copied
to the output, but with a code (let's call it "x") inserted before
each character.

I think I need some combination of a for-loop and a substring-function
that selects the characters one by one (or maybe this can be solved
using a regex). I just can't get it right... Any suggestions?

Marie Bilde Rasmussen,
Gyldendal Publishers, Dictionary Department
Copenhagen, Denmark