Geoff Pullum (gkp@ling.ucsc.edu)
Tue, 5 Sep 95 11:21:58 PDT
> From: Mark Johnson <Mark.Johnson@xerox.fr> > Date: Mon, 4 Sep 1995 17:13:44 +0200 > > . . . . . . . . . . . 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). Many readers may have experienced something of a shock on discovering that apparently there was a "newer, dynamic, ``resource-based''" linguistic theory called "minimalist transformational grammar". Newer? Everyone wants newer. Dynamic? How can dynamism be bad. Resource-based? Sorry, I mean "resource-based"? (Note the use of Berwick Quotes, placed around one's own technical usages in order to imply that they have come from some wider community out there where they have a known meaning.) Where might one find out about this new development? What are these new categorial-like results about phrase structure at which Johnson's sales-brochure language hints? A few moments later, the same readers may have had another, more depressing thought: could Mark really be referring to the unintelligble mush in the "minimalist program" paper in the 1993 volume, _The View From Building 20_? Could he really be suggesting that we are supposed to see in that embarrassing jumble of syntactobabble a new, dynamic, ``resource-based'' linguistic theory with definite properties and consequences, a theory that we can compare with the kind of frameworks Bob Carpenter is talking about? Apparently so. Unable to offer product lines that actually do anything, MIT is reduced to having its sales force and its in-the-field apologists simply make claims about its theories in the hope that no one will be able to tell the vaporware from the real stuff. The "minimalist program" has clearly been designed with so little actual content that it unifies with anything. That way, whenever anything (be it Lambek Categorial Grammar or HPSG or Windows95) works or looks good or has some prestige, the sales force can just drop a phrase mentioning minimalism in the same context, secure in the knowledge that no one can possibly prove the "minimalist program" to have distinct consequences or properties. Yet deniability is also guaranteed by the contentlessness of the actual presentations. Amusingly, on the actual point in question here --- whether there are PS-rule-like constraints on phrase structure --- Mark's claim, if one defocusses one's intellect sufficiently to see it as sort of having a kind of a consequence under some interpretations, seems to be closer to a falsehood than a truth. Since the "minimalist program" continues to espouse the usual mumbo-jumbo about "some version of X-bar theory", and since this mumbo-jumbo is supposed to mean that UG guarantees only the following sort of phrase structure configuration: X2 / \ (Y2) X1 / \ X0 (Z2) (where X, Y, and Z are lexical categories and parenthesized labels mark optional constituents), that means that only the following sorts of local trees are allowed: X2 X2 X1 X1 / \ | / \ | Y2 X1 X1 X0 Z2 X0 Now we know not only that there are (in effect) phrase structure rules in the "minimalist program" but also which ones they are, and exactly how many UG provides. If you expand out to rules that mention specific categories, then if the number of lexical (zero-bar-level) categories is K, there are 2(K^2 + K) rules. If the lexical categories are just {N, V, A, P, C, Agr, T, Neg} then that's 2(8^2 + 8) = 144 universal PS rule-like constraints on phrase structure. The only reason this does not suffice to show that Mark posted an untruth is that he follows Chomsky in hedging everything with cop-out words that can be used later to yield deniability. In the case of the quote from Mark above, the cop-out is easy to spot: > . . . . . . . . . . . . . . . . . no constraints are placed on > it directly, there is no need to talk about it at all). ^^^^^^^^ ^^^^^^^^^^^^^ There might be indirect constraints of some kind (perhaps anything imposed by UG could be called indirect); and the phrase structure might be in there in some sense without the theory having to "talk about it" (switch the issue from a metaphysical to a metalinguistic one). That should allow some wiggle room. Bob Carpenter has actually raised certain questions about phrase structure constraints and empty categories in a way that might permit contentful discussion of them. I hope Mark Johnson's remarks do not return these questions to the limbo of confusion where the MIT sales force seems to prefer them to remain. +---------------------------------------------------------------------------+ | G e o f f r e y K. P u l l u m * gkp@ling.ucsc.edu | | Stevenson College, University of California, Santa Cruz, California 95064 | | (408)459-4705 * Messages (408)459-2555/2905 * Fax (408)459-3334 | +---------------------------------------------------------------------------+
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:33:03 PST