Published online by Cambridge University Press: 01 January 2026
An analysis of antecedent mismatch effects under ellipsis is proposed to explain why some cases of verb phrase ellipsis exhibit a sizeable penalty when the elided target is not structurally matched to its antecedent, while other cases show little or no penalty at all. The proposal attributes the penalty in the former case to an information-structural constraint governing contrastive topics, and it is argued that previous accounts have misattributed that penalty to a licensing constraint on ellipsis. Results from four experiments (three off-line acceptability, one on-line self-paced reading) confirm that the relative size of the mismatch penalty can be reliably predicted based on the information structure of the clause containing the ellipsis and that acceptability differences associated with information structure are observable even in the absence of ellipsis.
This research was supported by NIH grant 5-T32-DC0041 to the Center for Research in Language, University of California, San Diego. I am grateful to research assistants Jeff Derrenberger, Emma Hendricks, Kelly Mak, Tessa Opperman, and Andrew Strabone for help with data collection and to Corey Cusimano for editing. Thanks for valuable discussion go to Roger Levy, Hannah Rohde, Cynthia Kilpatrick, Ivano Caponigro, Robert Kluender, Klinton Bicknell, Chris Barkley, Philip Hofmeister, and audiences at WCCFL 2008 and CUSP 2009, both hosted by the University of California, Los Angeles. A special debt is owed to Andrew Kehler for comments on earlier drafts and for many thoughtful discussions.