Reconstruction (A)symmetry between Parasitic Gap and Across-the-Board Constructions

Ross (1967) and Williams (1989/1990) propose that both gaps in parasitic gap (PG) and across-the-board (ATB) constructions are bound by the same antecedent. Recently, Hornstein and Nunes (2002) propose that these two constructions are derived through the same operation, sideward movement. Under the shared antecedent analyses, reconstruction asymmetry between PG and ATB constructions is problematic. In this paper, I propose that this difference results from the parallelism requirement in the ATB constructions. Due to the requirement, at LF, the restriction part of moved wh-phrases must remain in both conjuncts in ATB constructions. By contrast, in the PG construction, the restriction part doesn t have to remain in both the RG and PG position. Assuming theta-features are LF active features, I suggest that the restriction part remains in the real gap position at LF. For this reason, reconstruction of the overt wh-phrase seems to be possible only into the position of the real gap, and not to the position of the parasitic gap.