Comparative Syntax of the Balkan Languages (Oxford ... - Cryptm.org

DIRECT OBJECT CLITIC DOUBLING 143here agree in 4> features (that is, in number, person, and gender, since baresingulars, like a-expressions, are not marked for morphological Case inAlbanian and Greek). 32The question arises as to why bare indefinites cannot be doubled orscrambled. I will approach this question by considering first why bare singularscannot be doubled or scrambled." To the extent that this question has beenaddressed at all, bare singulars have been treated as forming a complex predicatewith the clausal predicate (cf. Haiden 1996), that is, as incorporatingsemantically. While this might seem intuitively correct, the fact that count baresingulars need not be adjacent to the clausal predicate but may be moved toSpec of CP, as in (40), shows that this semantic incorporation does not result34from syntactic incorporation of the bare singular into V.(40) a. Fustan doja re bleja. Albdress wanted te buy'It was a dress that I wanted to buy.'b. Zeitung habe ich gestern gelesen. Germnewspaper have I yesterday read'It was a newspaper that I read yesterday.'I propose that the impossibility of doubling and scrambling bare singulars is dueto feature mismatch between the clitic head and the direct object bare singularwith respect to the D feature; while clitics carry a D feature (cf. Emonds, toappear, Uriagareka 1995), bare singulars are NPs that altogether lack a Dprojection. Clitics are listed in the lexicon as separate morphophonologicalunits. That clitics carry a D feature (alternatively: are specified in the lexicon aselements of category D*) or are underlying determiners (cf. Postal 1969, Raposo1997) is not surprising, in view of the fact that they originate from personal anddemonstrative pronouns that are prototypical D heads (cf., Abney 1987 andsubsequent literature). This means, among other things, that only DPs but notNPs may be doubled and scrambled, since the [-D] feature of the latter willclash with the [+D] feature on the clitic head, thus causing the derivation not toconverge. This reasoning, however, rests on the assumption that bare singularsare NPs that lack a D projection. This is problematic, as it seems to run counterto Longobardi's (1994) that only DPs but not NPs may function as arguments,his idea being that bare noun objects have a morphologically null D head.Therefore, the assumption that bare singulars are NPs and not DPs with a morphologicallynull D is in need of some justification. Is there any evidence thatlegitimizes the claim that bare singulars lack a D projection? In what follows, Iwill argue that there is.First, note that bare singulars occur only as predicate nominals and as directobjects. Crucially, they cannot occur as subjects." Further, bare singulars do notoccur as direct objects of just any predicate; they may occur as direct objects ofonly those predicates whose bare plural direct objects cannot get a genericinterpretation (in the sense of referential or kind denoting) but get only anexistential interpretation. ' This fact alone raises an important question, namely: