• Keine Ergebnisse gefunden

Let us start out by recalling the functional architecture that has been proposed for the nominal domain:

(31) [KP [DP [NumP [GenP [nP [√ ]]]]23

(31) represents only the relative ordering of the different Functional elements inside the DP layer. DEM is not represented in this structure, since we accept the assumption that it is not a FC. The possibility should be envisaged that DEM, like POSS, starts from a complement position. But we leave this question for investigation elsewhere.

23 In this paper we don’t analyse the point of the lexical category. For our purposes it is not crucial whether it is a noun or a bare root selected by a functional small n.

Whereas DP, NumP and some version of a Classifier or a GenP are generally assumed to be Functional categories, to our understanding there is no general agreement on the universality and status of KP. Moreover, differing interpretations have been given of that category (cf., for instance, Kayne (2002) vs. Ogawa (2001)).

Remember also our previous working hypothesis, restated as (32):

(32) The data in OR do not indicate an unstable state, in which two competing grammars can be used alternately by speakers, but rather the alternating nominals have distinct semantic and syntactic properties.

Given the systematic properties of the BNPs that coexist in time with full DPs, I will try to explore the properties of the extended nominal domain, in association with the position of the BNPs. I will argue that the determiner was merged into D exclusively in cases where a specific unique countable nominal had a discourse-linked value. The main point to be developed is that other functional categories available in the extended functional domain of the N were able to license a nominal.

4.1. An earlier analysis

In previous work24 we proposed a view of the functional nominal domain in which the different Functional Categories of the universal repertoire were activated bottom-up as morphological evidence acted as a cue. In this process, the grammaticalization of the most deeply inserted Functional Categories took place parallel to the insertion of new morphological material in the upper part of the DP. Now we intend to go one step further, by relating this development to two factors that were not taken into account in that previous work: the properties of K/D25, and the fact that the indefinite article was not fully grammaticalized as marker of indefiniteness, at least in the plural.

We will still maintain some of the ideas of our previous analysis and which followed some insights offered by Boucher (2005), which we will review now.

Following Dobrovie-Sorin (2001: 208), Delfitto & Schroten (1991), Schroten (1991;

2001) and others, Boucher argues that in languages without overt determiners, no functional projection D needs to be established in order to derive the argument status of nominals. In these languages, the referential or definite value of the noun can be checked against other existing functional projections. He adopts the principle spelled out in (33):

(33) The restricted Quantification Constraint (RQC):

The existential interpretation of DPs is only available if two ‘logically’ different elements are present, filling distinct syntactic positions and respectively providing the domain of quantification and the ‘quantifying in’ operator.

(Boucher, 2005: 97)

Boucher goes on to say that the assertion in (33) should be interpreted as in (34):

(34) Lexical items (N, V, A, P) are first order predicates having scope respectively over objects, situations, attributes. Each lexical head projects a ‘pure’, semantic

24 Bartra-Kaufmann (2007).

25We must leave for future research the decision about the exact nature of this uppermost projection. We maintain its ambiguous status, either as a D which hosts Case (Giusti, 2002) or as a K which hosts reference (Boucher, 2005).

projection and assigns (one or more) theta roles. All nouns predicate a quality […] of some referent, which we call R (Higginbotham, 1986), and place in Spec,NP.

The structure is represented in (35):

(35) [NP R [N’N… ]]

‘Semantic’ functional heads are operators that must bind an argument in the lexical projection they c-command. Specifically, D binds (R) in the NP projection. In the case of Bare NPs, other extended functional projections may do so in the absence of DP.

Restating Boucher’s claims, we can say that

(36) The universal condition on restricted quantification can be satisfied by NPs if they are dominated at least by NumP, which is included in the set of ‘operator’

heads.

Following (36), the structure in (37) would reflect the properties of a nominal like that seen in the English sentence in (38). The detailed structure is shown in (39):

(37) [NUMP [NUM’ [Ni-Num] [NPR [N’ti]]]]

(38) Beavers build dams

(39) [NUMP [NUM’ [beaveri-s][NPR [N’ beaveri ]]]]

These structures are not possible in Modern French, Modern Spanish or Catalan.

Boucher claims that the reason is that in these languages the number affix is not strong.

In Boucher’s analysis,

(40) For a BNP to be used in subject position, the noun must adjoin also to K0.

Though based on it, our earlier analysis was a bit more complicated than Boucher’s.

Briefly, taking as a crucial argument the fact that most BNPs found in OS and OC were not incorporated into a complex Prepositional or verbal component, our proposal stated:

(41) (a) From the inventory of Functional Categories of UG, particular grammars activate only those for which there is formal and morphological evidence.

(b) Numeral subscripts in a representation like [DP D3 [NumP Num2 [GenP Gen1 [NPN] ] ] ] indicate the order of “activation”.

(c) The external subject position requires at the least an active FC NumP.

In that analysis, the following structures were presented:

(42) (a) fazer ayuntamiento

(b) [VP [V’ [fazer [NumP [ayuntai-miento-NumØ [GenP ayuntai-miento [SN ti]]]]

(43) (a) moros le reçiben

(b) [NumP [moroi’-[Num -s [GenP mor i- -o [SN mori]]]]

4.2. Some shortcomings

This analysis had, however, several shortcomings, which we summarize in (44):

(44) (a) It failed to predict exactly in which cases one or another FC would be activated;

(b) There was no clear relation between the internal structure of the BNP/DP and external licensing in the sentence.

(c) It failed to explain how, for instance, Num could be responsible for licensing a singular BNP in Old Romance in some positions but not in others (contrary to the facts in English, or in Modern Romance, for instance).