Unexpected behavior for image insertion using WordML to DocBook scripts

["Pike, Mary Ann-P63372" <MaryAnn.Pike@gdc4s.com>]

I have been playing with the WordML to DocBook conversion scripts found
on explain.com. They produce nice XML output, but I the images are not
converting to the structures that seem to be indicated by the
documentation.

In my Word file, I have a paragraph of type imageobject-imagedata that
contains only a file path to a graphic. The paragraph immmediately
preceeding that is of type mediaobject-title and contains what I intend
to be a title for the graphic. Given the documentation for the
conversion scripts, what I expect to see is a figure, containing a media
object that has a title and contains and image object that references
the graphic.

What I do see is a media object that has only a title, followed by a
media object that contains an image object with my graphic reference. No
figure object. I also don't get any figure objects for graphics that are
currently embedded in the Word document (I will be changing all of these
embedded graphics to file references). The script documentation seems to
indicate that figure objects are created automatically for graphics.

Can someone tell me:

What is the proper way to use mediaobject-title so that I get the title
and the graphic within the same media object?

Can I insert the figure-title item myself to create a captioned figure
containing the media object that follows it? When (if ever) is a figure
object created automatically?

Thanks for any suggestions,

Mary Ann Pike