Let's take the example 'A kiwi is [a type of bird]'.
Page 109 of this book https://faculty.mu.edu.sa/public/uploads/1367260110.5528Understanding%20Syntax.pdf sais that the head of a phrase:
A. Has the same distribution as the entire phrase
B. Is the one obligatory item in the phrase
That would mean that the head of [a type of bird] is 'bird' because we can say 'A kiwi is a bird' and not X'A kiwi is a type'.
This seems to be different to partitives, where 'some' does seem to be able to act as the head: 'Some of his remarks were quite flattering', 'Some were quite flattering' (some context retrieving needed?).
I guess the problem is that this goes against how we normally analyze these phrases. How can a word embedded in a PP which is embedded in the root NP be the head of the root NP? I suppose it would follow that of is a complement of kind and kind is an adjunct NP? And that this preposition (if it is even a preposition) isn't selecting a complement. Or is 'type' indeed the head? If so, what is the evidence that it is?