Multiple attributive adjectives with commaless prosody must
occur in different slots. The different slots are defined
by relative position and by semantic properties of words
occupying that slot. 

Multiple attributive adjectives with comma prosody occupy
the same slot (they occupy the same slot iff they have
comma prosody). Their order within the slot makes no
difference to the interpretation. Therefore there is no
semantic phrasing of multiple attributive adjectives in
the same slot. It seems therefore as though multiple 
attributive adjectives in the same slot are generated
not by adjunction recursion but by asyndetic coordination.

[I'm not 100% sure about the above, but am sureish.]

Attributive adjectives seem not to be adjuncts because
they are not recursible and because their order is fixed.

On the other hand, they're not like complements because
they are never subject to lexically specific constraints;
all common nouns have the same possibilites for attributive
adjectives [I haven't checked this].

Besides adjuncts, defined by recursibility, and complements,
defined by potentially being subject to lexically specific 
constraints, there is another, third type of dependency with
neither of these two properties. This third type includes
Subject and Extractee (= EWG's 'Visitor'; I call it 'Preject').

I conclude that the different attributive adjective slots
involve dependencies of this third type. It may or may not
be significant that all dependencies of this third type
are predependents, while all complements are postdependents
[with _Sophy_ in _Sophy's_ an unsolved problem].

Any comments?

--And.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%