Let us imagine there to exist, for instance, a DP. We always observe the only head D on the left or the right of the phrase (depending upon the direction of branching). But every constituent naturally have two borders which we normally indicate by brackets. Why aren't they detected? If they were expressed morphologically as heads and terminators how should we mark them in drawing the tree? As being category-markers, they would pertain to the same category thus causing us to assume the existence of discontinuous morphems (instances of the head), right? If so, how should we draw it?

1 Answer 1


In Syntactic Structures, Chomsky remarked that natural language needs a grammar more powerful than regular grammar (type 3), such as context free phrase structure (type 2), in order to keep track of the pairing of such phrase introducers as "if" and the corresponding "then" clauses which go with them.

I don't recall C's exact example, but consider the language generated by this context free phrase structure grammar:

 S -> it rains  
 S -> if S then S  
 S -> {john, mary} leaves

which pairs up each "if" with the correct following "then" ("correct", because in the tree structure, the "if" and corresponding "then" are in the same S). E.g. "if it rains then john leaves", "if if mary leaves then it rains then john leaves", "if if if it rains then it rains then mary leaves then john leaves".

There is, of course, an obvious problem with the facts of English, here.

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.