Processing of ACD gives no evidence for QR

Abstract

Hackl, Koster-Hale and Varvoutis (2012) argue for the existence of QR on the basis of the processing of ACD. But they failed to include a crucial control. Once included,one can show that the effect they demonstrate comes not from QR but comes instead from a pressure to insert word like also or same where the reported events are 'the same'. We further show that the asymmetry between every and the is due to the fact that use of every in the relevant cases makes it easier to establish a causal connection between the events, removing the pressure to insert also or same.Finally we show that - for their 'Large Ellipsis'condition in their Experiment 2- their hypothesis does not account for their reported data(at least not without additional assumptions), but ours does.