Why do phrases like "the car in Texas" break down into

(NP (Det the) (N car)
   (PP (P in) (NP (N Texas))))

Why is the prepositional phrase "in Texas" constituted of the preposition "in" and a single-child noun phrase, instead of the preposition "in" and the child noun itself "Texas"?

(NP (Det the) (N car)
   (PP (P in) (N Texas)))

What's the point of the seemingly-redundant noun phrase node in the syntax tree?

Why aren't phrase nodes with only one child reduced to that child directly?

1 Answer 1


By convention, NP is the phrasal projection of N, which can be an argument and is what P combines with to make a Prepositional Phrase. N by itself inside the noun phrase "the car" is not a complete argument.

You could alternatively adopt the convention that any X not dominated by a projection of itself is automatically considered a maximal projection (as Chomsky suggested in Bare Phrase Structure (BPS), 1995); in that case you could dispense with the NP node dominating N "Texas" and let P combine directly with the N, which by the BPS convention would also be equivalent to NP.

  • 1
    Thank you for the thorough answer. So it's a matter of convention?
    – minseong
    Nov 24, 2021 at 12:32
  • Yes, I think the usual convention is to graphically distinguish complete, phrasal NP from non-complete head N in the diagram, but if you adopt a convention whereby the symbol "N" can stand for both when there are no other items in the NP, that's perfectly coherent. Having distinct symbols for the head and the phrase is convenient if you are trying to depict head movement, for example, and there might be other situations where having two symbols is more perspicuous.
    – Psven
    Nov 24, 2021 at 13:06
  • Muysken and van Riemsdijk 1985 ''Projecting features and featuring projections" suggested replacing X and XP with X[±proj,±max], where [+proj] means it is a projection and [+max] means it doesn't project further. A head dominated by a projection of itself would be X[-proj,-max], an X' would be X[+proj,-max], and an XP projected from an X would be X[+proj,+max]. A non projecting N which is also an NP would then be N[-proj,+max]: [-proj] because it is not projected from a lower N, and [+max] because it doesn't project any higher. Chomsky 1995 suggested something similar with max and min instead.
    – Psven
    Nov 24, 2021 at 13:13
  • I'd say that the nominal is "car in Texas". The determiner "the" is of course external to the nominal, the NP being "the car in Texas".
    – BillJ
    Nov 24, 2021 at 17:21
  • Tree pruning is often necessary with projected nodes.
    – jlawler
    Nov 24, 2021 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.