Coerced Relative Clauses in Korean

In this paper, I use the following abbreviations: acc(usative case), (imperfective) adni(ominal), (modal) adnm (ominal ending), (perfective) adnp(ominal ending), c(o)mp(lementizer), dec(larative mood), imp(er)f(ective as- pect), n(or)m(ina)l(izer), nom(inativecase), p(a)st, pl(ural), top(ic). The morpheme nun occurs with an eventive verb and expresses imperfectiveness of the event it denotes. If (u)n is used with an eventive verb, it expresses the perfectiveness of the event that the verb denotes. But it can occur with a stative verb with no meaning of perfectiveness. In addition to the two, the adnominal ending -(u)l often expresses a modal meaning, but there are cases it does not have any meaning of modality: e.g. … (u)l ttay when ….In this paper, I define a coerced relative clause (= CRC) as an adnominal clause which denotes a property of situations that is converted to a property of things when the clause combines with a head noun that denotes a property of things. The property of situations is always determined by the topmost clause of the CRC. This definition makes the range of CRCs clear, compared with previous analyses, but it makes the range wider than previous analyses of the constructions. In filling the missing links between a property of situations and a property of things, I adopt the Generative Lexicon (= GL) Theory. I show that the analysis of CRCs should be pragmatic and the ways of interpreting CRCs are diverse, on the one hand, but that it should be restricted systematically by the meaning specifications of expressions in the GL, on the other. I also show that the GL Theory, in the current form of the Theory, is not sufficient to fill in all missing meaning components necessary in interpreting CRCs. In this respect, we need a more flexible tool than the GL Theory, or extend the GL Theory so that additional meaning components can be included in the meaning specifications of expressions