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

110 ILIYANA KRAPOVAstrictly differentiated from that of Type I Ss, which presumably had the effect ofblocking the possibility for an obviative reading of the null subjunctive subjectin the latter.2.2. pro subjects in Type I subjunctivesNow, let us examine some evidence that will reinforce the proposed distinctionbetween the two types of subjunctive clauses in Big. The issue of systematicallydifferentiating between pro and PRO becomes important due to the fact that dacomplementsare finite with strong agreement, and therefore pro and PRO willnot differ in terms of 4> features, although, in theory, they will differ in terms ofCase.In this subsection, I will argue that the set of pronominal properties typicallyassociated with pro gives the basis for the proper identification of the nullsubject in subjunctives of Type I in Blg as pro (for MGrk see Terzi 1991, 1992;latridou 1993; Varlokosia and Hornstein 1993).First, there is no complementary distribution between null subjunctive subjectsand nominative DPs or overt pronouns, a behavior expected of pro and notof PRO. As the indices in (9) illustrate, the overt pronoun can refer to the matrixDP or to some other DP, salient in the discourse. The same indexing applies tothe ec, which points to the fact that it behaves like an empty pronominal byvirtue of its specific and free reference:(9) Ivan- iska [brat mu/lojyJecy: da sledva.]Ivan want-3SG brother his/he da study-3SG'Ivan wants his brother/him to go to college.'/'Ivan wants to go tocollege.'Further, (l0a) from Blg and its MGrk parallel (l0b) show that embeddedagreement can take any person or number value, irrespective of the agreementwithin the matrix clause. Again, this is to be expected, if the subjunctive subjectis pro and not PRO: like DPs or overt pronouns (as in (9)), pro has nominativeCase, which is checked against the embedded Infl (tense, as I have assumed).Besides, the pro subject can be referentially free because its content will alwaysbe identified by the finite agreement in the embedded clause. The latter thereforedefines a complete functional complex (in Chomsky's 1986 terminology):(10) a. Nadjavam se [pro da dojdes/dojdat.}hope-ISC da come-2SG/come-3PL'I hope that you/they come.'b. Elpizo [pro/na erdis/erdun.]hope-ISO na come-2SG/come-3PLsame as in (lOa)Consequently, (l0a) and (l0b) are opaque complements, as are their respectiveindicative counterparts in (1 la) and (1 Ib): in both languages the verb hope