| Children have been found to acquire passives of "actional" verbs prior to passives of "nonactional" verbs. This has come to be known as the Maratsos Effect (ME) and has been widely replicated (Sudhalter & Braine 1985, Gordon & Chafetz 1990, Fox & Grodzinsky 1998, Hirsch & Wexler 2006).;I present two experiments that further investigate the ME. Researchers have not always been careful about using linguistic diagnostics for categorizing a verb as "actional" or "nonactional". In the experiments reported here, verbs are split into three categories based on linguistic diagnostics for eventivity and agentivity. In Exp. 1, eventive agentive verbs (paint, fix, and wash) are tested against eventive nonagentive verbs (forget, find, and spot). In Exp. 2, the same eventive nonagentive verbs are tested against noneventive nonagentive verbs (hate, know, and love). I find evidence for a three-way distinction in the English-acquiring child's acquisition path, rather than the two-day distinction which has classically been reported as the ME. Children learn the passives of eventive agentive verbs before the passives of eventive nonagentive verbs, which are in turn learned before the passives of noneventive nonagentive verbs.;I then explore possible accounts of this three-way distinction. I explore the intuition that children initially posit a structure for the eventive nonagentive verbs and the noneventive nonagentive verbs where both arguments of the eventive nonagentive and noneventive nonagentive verbs are projected internally to the VP. If one adopts Collins' (2005) analysis of the passive, then an account of the acquisition facts emerges on the basis of this intuition. However, this intuition cannot account for the facts if one adopts either Bruening's (2013) analysis of the passive or Legate's (2014) analysis of the passive. |
