Movement to Spec-TP
This section takes up the question of whether this non-discourse-oriented preverbal A position is actually an immediate constituent of a projection of Tense/Aspect. First we examine evidence from υP adverb placement which demonstrates that the landing site of the displaced internal argument is outside of υP. This analysis is further supported by genitive of negation facts and evidence from binding theory. The latter provides additional evidence that the landing site of the displaced internal argument in accusative unaccusative constructions is an A-position, hence within the complement of C.
Given that υP adverbs occur at the left periphery of υP, the contrast between the (a) and (b) examples in (1–2) shows that the dislocated accusative internal argument occurs outside the υP projection:
(1)
(2)
The (b) examples show that when the dislocated accusative occurs within the υP projection (i.e., following the υP adverb), the sentences become significantly less acceptable.
Given that υP is a complement of the functional head Negation, the analysis proposed for (38–39) is further supported by scope of negation facts. This is based on the contrast between regular accusative constructions and the accusative Unaccusative, as illustrated in (3) and (4).
(3)
(4)
In (40a) the structurally Case-marked internal argument appears in the GEN when it is within the scope of sentential negation. If this internal argument occurs as ACC instead of GEN, the result is significantly degraded for most speakers. We assume that the strong preference for the genitive of negation is conditioned by the presence of ni odin “not a single”, presumably concatenated within the object NP. The same preference holds when the internal argument is scrambled to clause initial position (see (3b)). In marked contrast, in the case of the accusative unaccusatives in (4), ACC on the preverbal dislocated internal argument is preferable under neutral discourse to the GenNeg, which is degraded.
The fact that GenNeg on the displaced preverbal argument of the accusative unaccusative is distinctly less preferable than ACC is consistent with our analysis that it occurs in Spec-TP. The contrast with a scrambled object suggests that for GenNeg the scrambled object behaves as it it remains in complement position—possibly by reconstruction. Thus the GenNeg facts provide further evidence for A-position properties of the displaced argument in the accusative unaccusative—i.e., it does not exhibit reconstructive behavior.
Anaphor binding, as illustrated in (5), provides further evidence that the displaced argument in accusative unaccusative constructions occupies an A-position.
(5)
In (5a) the displaced argument appears to bind the anaphor in VP. Given that the antecedent of an anaphor must be in an A-position, (5a) alone would appear to establish that the movement in accusative unaccusative constructions must be to an A-position. It is possible that the anaphor is bound by the trace of the displaced argument, which asymmetrically c commands it. Nonetheless, (5b) provides clear evidence that the displaced INST argument is in fact in an A-position. If it were in a non-A-position, then under reconstruction the anaphor would be properly bound. Example (5b) shows that the argument that remains in VP cannot bind an anaphor within the displaced constituent (which we assume occupies an A-position).
With respect to anaphor binding, the accusative unaccusative construction again contrasts with the short scrambling examples. Thus compare (5) with (6).
(6)
The example in (6a) shows that even though Mašu i Ivana “Maša and Ivan” c commands the anaphor drug druga “each other,” proper binding does not occur between the two. However in (6b), the nominative argument (presumably in an A-position) can bind the reciprocal anaphor even though it does not overtly c-command the anaphor. The binding in (6b) appears to reflect reconstructive behavior on the part of the scrambled argument. If so, then reconstruction in (6a) would yield a structure in which the accusative argument does not c-command the anaphor, thereby accounting for the deviance of the example. The reconstructive behavior in (6) identifies the scrambled position as distinct from a canonical A-position. The lack of reconstruction in (5b) supports our analysis of the landing site as an A-position.
To summarize the results of this subsection, we have shown on the basis of adverb and scope facts (with respect to Negation) that the moved constituent in accusative unaccusatives lands in a site higher than υP (and NegP). The fact that this moved constituent can appear non-D-linked in its preverbal position strongly suggests that movement does not occur to positions generally associated with topicalization (e.g. Spec CP and XP adjoined to TP). Further binding facts indicate that this movement targets an A-position. Locating this movement between υP and C implicates the Tense/Aspect system. This is consistent with our analysis that movement in the case of Russian and Ukrainian accusative unaccusatives is driven by T’s EPP requirement.
