The Analysis of the Frequencies of Reflexives: A Corpus-Based Account

Abstract

Our purpose here has been to address the use and frequencies of universally attested local anaphors and long-distance anaphors within the corpus approach. First, both Seycong corpus and novel corpus clearly illustrate the fact that the Korean reflexive casin ‘self’ can be used more frequently and that it is one of the primary exponents of Korean reflexives with caki ‘self’. We have argued in this paper that this finding, of course, accords well with our claim that casin ‘self’ appears anywhere other reflexives appear, but other reflexives are barred from certain positions where casin ‘self’ can appear. There might be plenty of reasons why the Korean reflexive casin ‘self’ is the most preferred reflexive among Korean reflexives. The biggest reason of all may be because it admits both local binding and non-local binding. In addition, the phi-features of casin ‘self’ are defective in that they are not specified. Thus, casin ‘self’ can get its reference from the first person, the second person, and the third person. Second, we have observed earlier that possessive case-marked reflexives in Korean are preferable to other case-marked reflexives and that a reflexive is most likely to be bound by an NP in a sentence if that reflexive can be interpreted as a possessive cased-marked reflexive. Although it is not clear how this can be accounted for, there may be much relevance between reflexives and their case markers. Third, we have pointed out that the overall frequency of caki-casin ‘self-self’ is lower than that of caki ‘self’. We assume that this result is expected, given the assumption that only caki ‘self’ among Korean reflexives can refer to the hearer as its referent. That is, this can be reduced to the availability of the hearer’s accessibility to caki ‘self’. Finally, we have demonstrated that Condition A in Korean does not constrain the distribution of anaphors which are supposed to be local and that anaphors in Korean freely overlap in local domain or non-local domain. Consequently, our corpus analysis does not argue for the hypothesis that reflexives and pronouns must be in complementary distribution.