Discourse Coherence and Relativization in Korean

It is widely accepted that the split between restrictive and nonrestrictive relatives generally coincides with their asymmetric contribution to discourse: restrictive relatives contribute content that is part of the main assertion of the sentence, while nonrestrictive relatives contribute content that is backgrounded or supplementary to the content of the main clause. Recent studies (Huddleston & Pullum 2002:1033-1065; Potts 2005:93-94), however, have argued that this traditional dichotomy along model-theoretic lines is not successful, offering evidence that some cases of nonrestrictive relative clauses do have equality of informational status with their main clauses (see, for example, Huddleston & Pullum (2002:1064)). In line with these studies, the present work argues that the model-theoretic properties of a relative clause (restrictive vs. nonrestrictive) and the kind of content it contributes (at-issue vs. supplementary) do not always go hand in hand. In Korean, two different types of relative clauses can be identified on the basis of whether the nominal that functions as an argument in both the relative and the main clause (henceforth, anchor) is outside the relative or not. In Externally-Headed Relative Clauses (EHRCs) like (1) the anchor sakwa ‘apple’ occurs immediately after the relative clause and there is a missing element within the relative that is coreferential with the anchor. In Internally-Headed Relative Clauses (IHRCs) like (2), by contrast, the anchor is within the relative clause, while the “external head” position is occupied by a defective noun kes that obligatorily takes a verbal complement. (1) and (2) are adapted from M. Kim (2004:101).