Information Structural Effects on Ellipsis Processing

Laura Kertz

UCSD

I present results from a study examining the source of parallelism effects in ellipsis. Verb phrase ellipsis is a phenomenon in English where a verb phrase can be omitted if it is recoverable from context. Competing theories argue that the underlying licensing mechanism for ellipsis is either inherently syntactic or semantic, and these two approaches make conflicting predictions regarding the possibility of syntactic mismatch between antecedent and target pairs. As various researchers have pointed out, however, neither approach is capable of predicting the full range of ellipsis data, where mismatch is sometimes good, sometimes bad, and sometimes in between.

Findings from the first set of experiments in the current study identify an information structural confound present in previously reported data, where acceptability of mismatch is conditioned on the focus structure of the target clause. The second set of experiments show that apparent syntactic mismatch effects are better explained as an interaction between topic structure and focus and confirm a prediction that mismatch effects can be observed even in the absence of ellipsis. The final experiment in the study introduces online reading time evidence in support of this proposal and demonstrates that the information structural cues identified in the previous experiments disrupt ellipsis processing quite early--prior to the reader even encountering the ellipsis.