The prose which introduces the notion of 'context-determined declarations'
suggests a mapping associated with each complex type which maps from
element information items (or their expanded names), via the particles
of the content model, to types.  That's a convenient concept.

But what is defined is only a subset of that mapping, omitting simple
types and elements that match wildcards.   There are technical reasons
for the omissions, but they mean that the concept actually defined is not
the concept described in the introductory prose and is not well captured
by the term 'context-determined type'.

Either a new term needs to be found, or the technical problems which
led to the counter-intuitive aspects of the definition need to be solved
in some other way.  Perhaps 'locally declared type' ?