This study attempts to show that voice specification must be encoded in phrase structure as a voice feature in
v, and consequently passives/unaccusatives, as well as actives, must involve
v for voice specification. The fact that passives/unaccusatives and actives share some syntactic behaviors that result from the properties of
vP confirms the proposal. Although
v is necessarily involved for voice specification, the
v in actives and that in passives/unaccusatives differ in their feature compositions. While the
v in actives has a Case feature and a semantic feature for an external argument, the
v in passives/unaccusatives contains the EPP-feature instead. Furthermore, the
v in raising constructions lacks even the EPP-feature.
View full abstract