Advances in the Syntax of DPs - Structure, agreement, and case

50 Steven Franks

(55) *[CP C [TP PRODAT TINF [VP PRODATV [AP PRODAT grustnomuDAT]]]]

Under my account this follows automatically from the hypothesis that the SD only

arises through direct assignment, since ordinary adjectives cannot receive case in this

way. Also important is the rejection of PROdat as a possible source of case features,

otherwise one might expect agreement in structure (55).

6.2.2 Why semipredicatives are not default instrumental

Similarly, one needs to prevent instrumental case from being assigned to the semi-

predicatives (in their ‘alone’ and ‘by oneself ’ meanings). Here my speculations adapt

the structural account of Madariaga (2006). If instrumental is due to Pred, then, elabo-

rating on (25b), Pred must select a small clause of the AP (or NP) type, but not of the

QP (or DP) type.^29 This is schematized in (56a), with the SC represented as functional

head F (just to give it a name). This structure however raises the question of how AP

(and NP) receive instrumental case.^30 Assuming APs agree and Pred values instru-

mental, we need to modify (56a) as in (56b).

(56) a. [PredP PRO Pred [FP F [AP/NP/*QP/*DP]]]

b. [PredP Pred [FP PROINST F [AP/NP/*QP/*DP]]]

If so, AP would agree locally in case with PROINST, whereas NP would be assigned

instrumental (by the Pred probe, through Multiple Agree, just like the post-copular

nominative). This is depicted in (57):

(57) a. [PredP Pred [FP PROINST F [APINST]]]

b. [PredP Pred [FP PROINST F [NPINST]]]

One conceptual issue with this analysis is that it resurrects cased PRO, in the famil-

iar sense of a category which cannot be overt but which can be agreed with. Given

the well-studied Icelandic facts, this may be inevitable. Moreover, given the contrast

between predicate adjectives and nouns in Polish (51), it may be advantageous to treat

instrumental on adjectives and nouns differentially.

29. Copular NPs thus receive case from Pred, whereas DPs receive case through multiple
    agree. See Perelstvaig (2007) for extensive arguments distinguishing post-copular NPs, which
    are instrumental, from post-copular DPs, which are nominative.


30. If the PredP approach proves untenable, “instrumental as default” can always be invoked,
    so long as it can be ensured that the semipredicatives never happen to receive default case.