This study focuses on subordinate clauses encoded by ‘-un kesi’ and ‘-un kesul’ (the ‘-un kesi’ construction and the ‘-un kesul’ construction). It describes evidential meaning and polysemous features that are found among related examples. More specifi ...
This study focuses on subordinate clauses encoded by ‘-un kesi’ and ‘-un kesul’ (the ‘-un kesi’ construction and the ‘-un kesul’ construction). It describes evidential meaning and polysemous features that are found among related examples. More specifically, the ‘-un kesi’ and the ‘-un kesul’ constructions that are used as an adverbial clause or an internally-headed relative clause (RC) are the major targets of this study. In order to examine the way that these constructions are in use, we collect naturally-occurring data through corpus searches. In addition to this, we used deep learning programs and heat maps to discover identificational features for complement clauses, internally-headed RCs, and adverbial clauses, the three subordinate clauses that are marked by ‘-un kesi’ and ‘-un kesul.’
The results of this study are as follows. First, internally-headed RCs are found both from spoken and written data and they are used for specific discourse functions. Second, resemblant constructions to internally-headed RCs have unique discourse functions and they show similarities to left-dislocation. Third, the formal restrictions in internally-headed RCs can be described in terms of a set of firsthand knowledge and response to it. Taking this approach, we can explain why only certain clause-linking meanings are found between internally-headed RCs and the main clauses. When the response is pragmatically expected, ‘time’ or ‘reason’ can be held between the two clauses linked by ‘-un kesul’, and when it is not, ‘time’, or ‘concessive’ meaning is to be held. Fourth, internally-headed RCs marked by ‘-un kesul’ and ‘-un kesi’ respectively seem to be at a different stage regarding grammaticalization. Internally-headed RCs marked ‘-un kesul’ showed less variation and less exception regarding restriction in its formation. Fifth, complement clauses, RCs, and adverbial clauses marked by ‘-un kesi’ and ‘-un kesul’ seem to form a network based on the formal and semantic similarities. To be more specific, these constructions seem to be related to the firsthand knowledge of a speaker or the subject of the main clause. We found that the evidential meaning observed from internally-headed RCs marked by ‘-un kesi’ and ‘-un kesul’ might be an outcome of the extension of a construction based on semantic similarities.
This study can contribute to existing studies as follows. First, this study used naturally-occurring data from spoken and written corpora to complement previous studies which have dealt with artificial data. Moreover, the current study examined resemblant constructions to internally-headed RCs and suggested that this construction shows a great deal of similarities to left-dislocation. By expanding examples of left-dislocation, this can enhance our understanding about the syntax and information structure of left-dislocation in Korean. In addition to this, this study shed light on examples of internally-headed RCs and adverbial clauses marked by ‘-un kesi’ and ‘-un kesul’ which represent firsthand knowledge. Specifically, this research suggested the possibility that the evidential meaning observed in internally-headed RCs might be identified as participatory, which has rarely been discussed in Korean linguistics so far. Lastly, This study paves the way for future research to sketch the meaning and function ‘-un kesi’ and ‘-un kesul’ represent through grammaticalization.