Should I conjoin two AdjP or two Adj inside one AdjP?

Question

In an NP like "stronger but smaller storm", when drawing a tree, how do I know if I should conjoin two AdjP or two Adj inside one AdjP?

I'm working in the framework of Carnie's book (3rd edition).

He has in particular these rules:

XP -> XP conj XP

as well as

X -> X conj X

(where X is a category). So it's not clear which one to use in this case.

Here are all rules he gives:

He also gives this example:

but hastens to point out the following:

I don't really understand what he means by "the category of red and blue is ambiguous between a head word and a phrase" and whether similar reasoning applies to "stronger but smaller".

Answer 1

I'm afraid I don't know Carnie's book (so this could be a comment, but it's too long).

But my suggestion is: look at all the syntactic rules, and see what distinguishes an AP from an A. If the only rule involving them is AP → A, for example, then it doesn't matter which one you choose: the tree will be effectively the same either way. But there's probably more to it than that.

In particular, look at how his theory treats adverbs. If the phrase were "(very small) and (very strong)", how would that change the tree? What if it were "incredibly (small and strong)"? Does either of those require one construction over the other?

Answer 2

So far as I know, English has no meaningful distinction between categories that depends on whether an expression has a single word or has several words. Of course, the English language has a distinction between the words "phrase" and "word", but that's English, not English grammar.

On the other hand, you do very often see linguists distinguishing a grammatical type with name ending in "P" from one not ending in "P" depending on how many words an expression has. But this is always wrong and results from confusion between English grammar and the English language.

An NP can have a Determiner, though it needn't, but a N cannot -- it has nothing to do with how many words there are, per se, though of course if there is both an explicit N and an explicit Det, an expression has multiple words and is a NP rather than an N, but this is because there is a Det, not because there are two or more words.

A good rule about coordinating conjunctions is that the grammatical categories coordinated are always the same, but it will take you only a few seconds of experimentation to discover that a single word and a multiword expression can perfectly well be coordinated with "and". The number of words is not what matters.

Another good general rule, in my opinion, is that modification does not change category. I derive this from my reading of McCawley's The Syntactic Phenomena of English, though I don't know that M. actually states this as a general principle. This is in direct conflict with the (incorrect) notion that category is connected with the number of words in an expression. Following this principle, "very warm" must be an A (not AP), since "warm" is an A and "very" modifies "warm".

There is a categorial difference between A (adjective) and AP (adjective phrase) that is connected with whether a modifier comes before or after a N that it modifies, AP tending to come after. But this is a difference in complexity, not number of words. Compare

a bought chicken
a ready to eat chicken
a chicken bought ready to eat
*a bought ready to eat chicken

You may find a stylistic difference that seems to depend on number of words -- compare "a forest dark and dank" with ?"a forest dank" or ?"a forest very dark", where a postnominal modifer is used for poetic effect.