12.07.2015 Views

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

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

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

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

106 ILIYANA KRAPOVAterms <strong>of</strong> binding relations. Complements <strong>of</strong>, for example, volitionals in MGrk(and Blg, as we shall see) allow but do not require <strong>the</strong>ir (null) subjunctive subjectto be coreferent with <strong>the</strong> matrix subject, thus voiding <strong>the</strong> familiar obviationeffect characteristic <strong>of</strong> Romance subjunctive clauses (Ib). On <strong>the</strong> o<strong>the</strong>r hand,verbs like try select a control complement in which <strong>the</strong> (null) embedded subjectis strictly anaphoric and may be interpreted only through a local antecedent in<strong>the</strong> matrix clause. 4This contrast in referential behavior, which is reminiscent <strong>of</strong> <strong>the</strong> distinctionbetween nonobligatory and obligatory control, as proposed by Williams (1980),is due, but only in part, to <strong>the</strong> individual lexicosemantic properties <strong>of</strong> <strong>the</strong>complement-selecting predicates. I will show that <strong>the</strong>re are syntactic aspects <strong>of</strong>control that cannot be accounted for by any <strong>the</strong>ory that considers control (andmood selection) to be a purely semantic matter (see Joseph 1992, Farkas 1992a,among o<strong>the</strong>rs). In this essay, I will try to attain a generalization regarding <strong>the</strong>syntactic conditions that require a particular null subject in a particular environmentand, ultimately, to establish a correlation between types <strong>of</strong> empty categories(ecs) and properties <strong>of</strong> clauses that license <strong>the</strong>se ecs.Varlokosta and Hornstein (1993) argue on <strong>the</strong> basis <strong>of</strong> an earlier proposalby latridou (1993) that MGrk subjunctives do not constitute a uniform class butshould ra<strong>the</strong>r be divided into two subsets, and that each subset can be identifiedthrough <strong>the</strong> type <strong>of</strong> null subject that it takes—pronominal pro versus anaphoricPRO. The discussion <strong>of</strong> <strong>the</strong> Bulgarian data (in section 2) will confirm <strong>the</strong> essentialcorrectness <strong>of</strong> this proposal. Once <strong>the</strong> distribution <strong>of</strong> pro and PRO in Big andMGrk subjunctive clauses is captured, it will be shown that <strong>the</strong> two types <strong>of</strong>contexts are mutually exclusive and can be defined through a correlation with <strong>the</strong>morphological content <strong>of</strong> subjunctive tense in terms <strong>of</strong> feature strength (section3). Thus, my analysis will lead to <strong>the</strong> conclusion that optional and obligatorycontrol <strong>of</strong> <strong>the</strong> null subjunctive subject in cases like (la) versus (2a) does notresult from properties intrinsic to pro or PRO" but is ra<strong>the</strong>r a consequence <strong>of</strong> <strong>the</strong>distribution <strong>of</strong> tense features and ultimately an instantiation <strong>of</strong> feature checkingin <strong>the</strong> relevant configurations (section 4).I will adopt <strong>the</strong> Minimalist approach <strong>of</strong> Chomsky (1993) with some insightsfrom Chomsky (1995.chap. 4) concerning <strong>the</strong> motivation and <strong>the</strong> conditions<strong>of</strong> (overt and covert) movement as driven by feature-checking considerations.I will assume <strong>the</strong> standard view that pro has nominative Case, which ischecked by <strong>the</strong> strong features <strong>of</strong> finite tense. With respect to PRO, I will adopt<strong>the</strong> Case-<strong>the</strong>oretic account <strong>of</strong> its distribution, as proposed by Chomsky andLasnik (1993), who argue that PRO is <strong>the</strong> minimal Case-marked DP that checksnull Case against a minimal Infl. "Minimal" Infl is identified with <strong>the</strong> weak feature<strong>of</strong> nonfinite tense (i.e., [-tense]) typically associated with control infinitivals,for instance, in English. For MGrk and Blg, however, which do not exhibitany finite/nonfinite distinctions in <strong>the</strong>ir subjunctive complements, <strong>the</strong>correct identification <strong>of</strong> tense features in terms <strong>of</strong> strength is still not fullyunderstood.I will assume that <strong>the</strong> structure that corresponds to MGrk and Blg subjunctivecomplements is as in (3). I will adopt <strong>the</strong> view (<strong>of</strong> Chomsky 1995.chap. 4)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!