I've run into "semantic head parsing" and "syntactic head parsing" and while I think I have a feel for the difference, I was wondering if anyone could give a more concrete definition or reference to read up about the difference.

In particular I'm curious about the characteristics of parsing sentences with each method. What benefits does each one have? What aspects of the sentence do they highlight?

  • Do you have any references wherein you have seen the terms? – Aaron Oct 4 '11 at 18:38
  • I know what "semantic head" and "syntactic head" are. The difference is relevant in parsing (or representation) with some grammars like HPSG. But I have never come across the terms "semantic head parsing" and "syntactic head parsing". A quick google reveals that there are no such terms (they lead back only to this question). Given what I know about syntactic and semantic heads, your last three questions cannot be answered. What exactly do you have in mind? – prash Oct 4 '11 at 20:18
  • @prash I'm looking for the papers now, but you're right, I saw the terms in the context of HSPG. – Sara Oct 5 '11 at 1:09

When you write down the semantic representation of a sentence (lambda calculus) you see that the determiner "consumes" the noun. Hence, it is the semantic head of a noun phrase. SEOP has a comprehensive overview of the phenomena that led to this, but Wikipedia is not bad either.

In syntax, the noun is considered the head of the noun phrase because it is the more "contentful" part, in the way people understand sentences.

In HPSG, the CAT derivation is somewhat like the regular phrase-structure derivation, with heads having a correspondence to phrase-structure (PS) heads. (Though, AFAIK, there is no need for the concept of "head" in PS.*) The CONT derivation is somewhat like the derivation in Montague semantics.

These are not two different types of parsing in HPSG (or anywhere else, AFAIK), but two (of the many) aspects of parsing with a rich grammar like HPSG.

Benefits: I don't think you can compare one with the other for benefits... if you go through the references I gave for quantifiers, you'll see that the mathematics would become very cumbersome and ugly (or maybe even impossible... I don't know) if you try to make the semantic heads correspond with PS heads.

* Though some parsing techniques find it useful to have heads of a PS marked.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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