An empirical study of honorific mismatches in Korean

Abstract

While numerous studies have approached honorification mainly focusing on how honorific forms and features agree with each other in syntactic derivation, the present study focuses on mismatches between a referent and the verb that the referent depends on with respect to honorification. If Korean honorification is a syntactic phenomenon, the mismatches in honorification must be fully accounted for as syntactic agreement cannot be disobeyed in principle. However, either theoretical or empirical inquiry into the mismatches in honorification has not received due attention. Systematic theoretical research on the issue is scant and the empirical endeavors are even rarer. Therefore, the full range of acceptability variance for different honorific expressions remains unknown, which hinders a livelier discussion about honorification. In this context, the present work employs two widely supported methods of linguistic research, viz. acceptability judgment testing and corpus exploration, in order to account for cases of the honorific mismatches. First, the cross-validation reveals that there is no a posteriori evidence for believing that Korean honorification undergoes a syntactic agreement. The experimental results indicate that Korean native speakers largely allow the mismatches involving a referent in an honorific form and a verb in a non-honorific form. Furthermore, they reveal that this type of mismatch forms occurs almost twice as frequently as the forms in perfect agreement in the corpus. These results demonstrate that the syntactic agreement-based approach conflicts with the native speakers' acceptability judgments and the language usage in daily speech. Second, the present study empirically corroborates the evidence that Korean honorification is semantically/pragmatically controlled. On the one hand, it is borne out that the mismatches involving a referent in a non-honorific form and a verb in an honorific form are mostly disallowed. These mismatches are instances of semantic restriction (i.e., anomalies); the head lexically imposes a blocking constraint on its dependent (not vice versa) with respect to the meaning and marking of honorifications. On the other hand, the empirical findings support the validity of the pragmatic conception that speakers manipulate the attitudes toward a given discourse situation and thereby choose the most appropriate honorific form out of the different forms with different degrees of deference. Since the most salient entity to the speaker in discourse (mentally as well as physically) is the hearer, the hearer (if honorable) tends to receive the strongest deference in honorific expressions; it turns out that if the verb is honorifically marked almost two-thirds of the associated referents refer to the honorable hearer. In addition, the mismatches with a referent in a non-honorific form and a verb in an honorific form can appear when the speaker needs to modulate their attitude of deference to an extreme degree or the honorific feature of the referent is semantically underspecified.