One step at a time: Representational overlap between active voice, be-passive, and get-passive forms in english

Abstract

The active voice and passive voice are complementary sentence forms that are available when describing a transitive event. In English, the latter has two variants: be-passive and get-passive. Numerous attempts have been made in the literature to represent the syntactic and semantic differences between these forms, while maintaining their shared features, yet theoretical accounts still differ. At the same time, empirical studies into structural choice have frequently investigated the use of passive voice versus active voice, while the distinction between get- versus be-passive has not received much attention. Here we investigate the degree of similarity between the three transitive variants (be-passive, get-passive, active voice), providing experimental evidence of their mental representations in relation to each other. We describe three experiments in which participants gave acceptability or naturalness ratings for sentences formed with either be-passive or get-passive, and containing one of several adjunct types. Participants were also free to provide an alternative way to phrase each, enabling us to consider whether there are differences in accessing alternatives. We observed overwhelming preferences for changing get-passives into be-passives, and for changing be-passives into active voice, but none for changing get-passives directly into active voice (despite active voice being the most preferred variant). This preference for changing get-passive into be-passive was observed even when a change into active voice was further facilitated by the availability of a 'ready-made' agent. These patterns of change are consistent with partial representational overlap along two dimensions described by Thompson et al. (2013): Patient Prominence and Patient Importance. Our findings also contribute to discussions of passive structure by revealing the relative closeness of the mental representations of these forms.