Colloquium: Lyn Frazier
On Tuesday, October 23rd at 4pm in SH 3.104, Lyn Frazier (University of Massachusetts Amherst) will be giving a talk in the GK colloquium.
Title: Processing ellipsis: The circumstances of repair
There is evidence from the processing of mismatch ellipsis that an antecedent for an elided constituent will be repaired when it does not syntactically match the elided constituent. Such repair operations do not take place indiscriminately but rather under particular circumstances including:
–when there is evidence for the repair
–when only a few repair operations are needed
–when the repair essentially reverses a natural speech error
Acceptability judgements are graded, inversely correlated with the degree of difficulty of the repair (Arregui et al., 2006, Frazier, 2013). One might take split antecedent ellipsis to be a proto-typical case of repair since no matching antecedent is available, i.e., by definition there is no surface conjoined VP to antecede the elided constituent. It will be argued, however, that standard cases of split antecedent ellipsis are actually cases of accommodation, not repair. Several studies show that split antecedent ellipsis does not exhibit the hallmarks of repair, for example the properties indicated above, but instead may be a case of ordinary accommodation. This raises the question of when the processor repairs an input and when it merely accommodates an antecedent of the appropriate type. It is suggested that the answer, at least in part, depends on whether the syntactic representation of the antecedent clause is in what we call Activated Syntactic Memory (ASM). Preliminary evidence suggests that the syntactic representation of the last (independent) clause is in ASM, explaining why standard cases of split antecedent ellipsis involve accommodation, not repair, as well as explaining the preferred antecedents of various VP ellipsis structures and possibly the loss of the correlation with repair difficulty once the antecedent is farther away and thus no longer in ASM.