Mark Johnson (Mark.Johnson@xerox.fr)
Mon, 4 Sep 1995 17:13:44 +0200
Surely phrase-structure, like all of the other constructs posited in linguistics (e.g., complex feature bundles, derivations, etc.) is a theoretical construct, and so is justified by the role it plays in an explanatory theory. As far as I can tell, it plays a crucial role in theories like HPSG, LFG, etc., and so is justified within these paradigms. It also seems to play an important role in much psycholinguistic work, although it is not clear how closely the psycholinguistic concept of phrase structure is related to any of the linguistic ones. Bob's important point is that in the newer, dynamic, ``resource-based'' linguistic theories (e.g., Lambek Categorial Grammar, minimalist transformational grammar, etc.) the concept of phrase-structure plays only a minimal role (e.g., no constraints are placed on it directly, there is no need to talk about it at all). This is in marked contrast to what I would call the classical, static, ``structuralist'' theories such as GPSG, HPSG and early 1980s transformational grammar (a.k.a. ``G&B''), in which a phrase structure (or perhaps several) was reified, and required to satisfy a multitude of constraints. Clearly in these older theories phrase-structure plays a crucial role. Personally, I think that it is premature for us to imagine we have the ``correct'' theoretical concepts for describing language. I think that the best that we can hope for is to collect a set of implicational statements about the relationship between various linguistic concepts. For example, it seems clear that if we adopt a static, structuralist, conception and insist that all relationships are strictly local within the ``phrase structure'', then we must also adopt something like the SLASH-passing treatment of long-distance dependencies. Similar statements can be made about e.g., the various ways of analysing object-raising and the implications that these have for the treatment of case-marking and semantic interpretation. Mark Johnson
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:33:03 PST