Advice sought re WYSIWYG end-user XML edit facility

Hi

This is perhaps straying a little from the strict parameters of the XSL
list, but I would appreciate any advice from the great and the good (that
means all of you!) on this question. If that advice happens to include a
pointer to a more appropriate forum for asking the question, then thats
fine too.

* Background

I have a pretty simple DTD/schema for marked up text documents (simple as in
a few dozen elements). The documents are small, but the schema includes an
element which allows any document to incorporate transparent links to other
documents, such that when the document is displayed all linked content
appears in place (i.e. these are not hyperlinks  which are also supported,
of course).

Our technical authors are using Structured Framemaker to produce these
documents, which requires FM to call XML stylesheets on open and save: the
open XSL pulls in all included content, and the save XSL slices and dices it
back into its component pieces. The author thinks that they are editing a
single large document, and the Framemaker template we are using gives them a
WYSIWYG view of the document that pretty closely mirrors how the resultant
pages are displayed in our end-user app (using a browser component  theres
a separate XML stylesheet to generate HTML, and a CSS to modify the
presentation).

This is all working (although theres been the odd appeal to this list for
help in achieving some of the trickier bits)!

* What Im looking for now

I want to provide a facility whereby end-users will be able to edit some of
the documents while browsing them. Obviously there will be a few controls in
place (e.g. only allow editing of leaf documents) but at the moment I just
want to investigate how easily and affordably this could be done. The key
points are:
 - WYSIWYG (as per the existing XSL and CSS layers)
 - the editor needs to be schema/DTD aware
 - the end-user should not need to be au fait with the schema/DTD in order
to edit the document effectively
 - small footprint (Framemaker would be way, way, over the top, not to
mention the cost!)
 - um, yes, Ill mention cost after all: a reasonable deployment price 
there could be a large number of users eventually, if all goes well!

If anyone has any helpful ideas for me, Id be keen to read them.

Thanks

Cheers
Trevor