This paper provides a new account for why unaccusative verbs are easier to process than unergative verbs in the reduced relative garden path construction, as demonstrated in Stevenson and Merlo . Reanalysis to the passivized reduced relative clause form requires the verb to be causative. Stevenson and Merlo  argued that unaccusatives are causativized in the lexicon, while unergatives are causativized in the syntax. This account argues instead that an independently attested co-occurrence restriction contributes to greater initial ambiguity in the unergative case; causative unergatives require an argument/directional attachment of prepositional phrase [Hoekstra, 1988, Levin and Rappaport-Hovav, 1995, Folli and Harley, 2006].
We implement the unergative-PP co-occurrence restriction in Minimalist Grammars [Stabler, 1997]. We model the contribution of prepositional phrase ambiguity to unergative reduced relative ambiguity with Entropy Reduction [Hale, 2003]. We obtain greater Entropy Reductions for the unergative condition, modeling that human comprehenders are more taxed by compounded ambiguity.
"Why Unaccusatives Have it Easy: Reduced Relative Garden Path Effects and Verb Type,"
University of Pennsylvania Working Papers in Linguistics:
1, Article 16.
Available at: http://repository.upenn.edu/pwpl/vol17/iss1/16