I know that my question may seem that of a delusional madman.
But I'm well aware of what CDATA is, I had just hoped to access it
the same way one can access data from a URL:

Well, you can. The CDATA contains a string. If this string is well-formed
XML, you can turn it into a tree by parsing it. Saxon 7.1 allows you to do
this using saxon:parse(). But it still begs the question: the purpose of
CDATA is to say "the stuff in here isn't markup", so if it is markup, why
are you putting it in CDATA?

Because I would like to parse my language for validity. Part of the
language definition allows arbitrary well-formed XML as data input -
so I guess my validity parse can only be "partial". If I use a DTD,
I need to escape the arbitrary XML (hence the CDATA). I'm looking into
the "lax" option of XML Schemas in an attempt to avoid this.