Usage probability and subject–object asymmetries in Korean case ellipsis: Experiments with subject case ellipsis

Case ellipsis for subjects and objects in Korean exhibits several clear asymmetries that have not received a unified explanation. This paper provides a new, probability-based analysis of variable case marking that can account for three types of subject–object asymmetries noted in the literature in terms of asymmetries in the usage probability of the properties of argument NPs in their syntactic or discourse context. This account captures the key generalizations underlying the asymmetries that case ellipsis for wh-word subjects, subjects in OSV sentences and non-specific subjects is unacceptable, whereas case ellipsis for objects with similar properties is acceptable; it also explains why sentences with a subject NP not marked for case that have been predicted to be syntactically ill-formed by previous syntactic accounts are judged acceptable when the subject represents expected, predictable information in context. These results provide strong support for the view that native speakers’ knowledge of grammar includes not only some degree of knowledge of probabilistic information but also access to fine-grained predictability and probabilities.

Non-canonical word order and subject-object asymmetry in Korean case ellipsis

The dispreference for subject case ellipsis in OSV sentences has been analyzed as resulting from a violation of a structural requirement on the position of bare subject NPs (Ahn and Cho 2006a, 2006b, 2007). In this study, we present evidence from an acceptability rating experiment demonstrating that OSV sentences containing a case-ellipsed subject exhibit acceptability patterns different from ungrammatical sentences violating a core syntactic principle on case assignment and that these sentences are judged acceptable when the subject refers to expected, predictable information in context. This evidence supports the conclusion that the dispreference for subject case ellipsis in OSV sentences is due to violations of probabilistic constraints that favor case marking for rare types of subjects and such violations can be remedied by non-syntactic information.