Predicative (postnominal) adjectives

As for the set of [+p] adjectives, I claim that the appropriate representation for this type of relation is similar to the one found in (secondary) predication within VP, namely, when a predicate merges above the maximal lexical projection of N.

In standard approaches to (secondary) predication this relation has been termed an adjunction relation. However, I will assume that there exists a distinction between true adjunction or Pair-Merge and Merge for predication purposes between AP and NP. In predication – as has been classically argued (Williams 1980) – predicative adjectives select for the category they modify. A (mutual) c/m-command relation should hold for the predication relation to be established. The question is then how to express c/m-command and what basic relation is expressed by c/m-command.

Under the hypothesis of minimal search and restriction on bare phrase structure (Chomsky 2001b), we may contend that a head noun – and its maximal projection – can be extended through (a series of) second or subsequent merges. The unit so merged would be, strictly speaking, a special kind of Spec or, in my descriptive terms, an Adjunct/Spec. By definition, Specifiers must satisfy the Extended Projection Principle (EPP) feature of the head (internal Merge) or be semantically selected by the head (external Merge). In certain cases, they have to undergo Agree (where a head contains a probe seeking a goal with matching features). In the case of predicative adjectives they are not selected by N (instead the adjective would select for N, in a certain sense), and the relation between nouns and predicative adjectives in DP is not one of Agree. I stipulate that the relation between N and the predicative adjective is Concord (Carstens 2000). Concord is assumed to take place when pairing among features is required, as a part of Merge, but there is no matching of features with the resulting pied-piping and deletion. I will not take a position as to whether Concord delimits a second type of specifier, but there are independent reasons to assert that the relation between AP and NP in is closer to the Spec–head relation than to the adjunct–XP relation. As we will see immediately below, the weak constraints on the ordering among restrictive adjectives also appear to provide some support for this hypothesis.

If these assumptions are tenable, predicative adjectives are then merged in DP as (multiple) specifier-like elements. In other words, I assume that the syntactic relation between N and the predicative adjectives modifying it is similar to the one that is established in secondary predication, namely, the AP merges higher than NP within the same maximal projection and the configuration does not preclude m-command as would be the case if we were dealing with an adjunction relation. At the SEM module this way of combining the adjective with the noun will be read, then, as the intersection between the denotations of N and A. Merge of predicative adjectives in DP will not be subject, on this view, to any other restrictions, and lexical incompatibilities between N and A will produce deviance and not ungrammaticality.