The limits of principle C reconstruction in multi-gap dependencies
Principle C reconstruction supposedly reveals the syntactic link between the filler and each gap in across-the-board and parasitic gap dependencies. However, the reported outcomes of the diagnostic conflict, giving rise to various syntactic `tools’ to derive multi-gap dependencies. I present a series of coreference judgment experiments reproducing patterns that were previously taken to indicate an asymmetry between the two gaps of each construction. Crucially, inter-individual variability shows that this asymmetry does not result from an absence of reconstruction to the non-initial gap, but rather from a decrease of the already compromised robustness of the effect with increased linear distance. Since these data are thus orthogonal to the underlying syntax of ATB and parasitic gaps, claims relying primarily on them lose empirical support.