The problem of prenominal modifiers

The view sketched above can be extended to other postnominal modifiers including PPs, as in (1a), reduced relative clauses, as in (1b), and combinations of them as in (1c). The latter involve recursive DP shells and multiple raising to light heads, as shown in (2).

But what about prenominal modifiers, APs like those in (3), which are semantically equivalent to copular relative clauses? ¹

One possibility is base generation along the lines in (4).

On reflection, however, this idea is problematic. Given a Ë-role-based approach, in order to project AP in the site shown in (4), we would apparently have to allow for an optional oblique Ë-role between two obligatory roles in our hierarchy, as in (5).

Worse yet, given the wide range of modifiers available in the prenominal site, we would seem to have to allow for a very large number of optional oblique Ë-roles between our two obligatory ones (6). This looks unpromising.

The only plausible alternative we see is that prenominal position is not a base position for adjectives in English, but rather a derived one. That is, we are led to resurrect the view of early transformationalists: that intersective attributive APs originate in the position of RCs, and obtain their surface position by movement, along the lines shown in either (7a) or (7b).

However, this raises the natural question as to why restrictive adjectives must move from their base position. Why can’t they remain in postnominal position like PPs, finite and reduced relative clauses?

1 We are not suggesting, of course, that all prenominal modifiers are equivalent to relative clauses. In fact there are well-known differences between them. For more on this topic see Larson (1998) and Larson and Takahashi (2007).