Recognition memory for active and passive sentences

Data from three experiments are reported in which recognition memory for active and passive sentences is compared. The results consistently show no difference, a result taken to invalidate syntactic memory hypotheses (e.g., the kernel plus tag and markedness hypotheses) which assume that the stored representation of passive sentences is more complex than that of active sentences. Previous reports of superior recall of active sentences are attributed to a reconstructive bias. One form of propositional hypothesis and a verbatim hypothesis are consistent with the data.This research was supported by Grant MH 19859 from NIMH to the first author, and by Rutgers University Research Council grants to both authors.