In this dissertation, I examine the interpretation of NP-ellipsis (henceforth, NPE) in Korean. Some unexpected interpretive and grammatical contrasts arise in Korean between pro-forms such as "caki" "self" and "ku" "'3.S.M" (~"he") in object position and their null counterparts in NPE constructions. To be more specific, the NPs that are not expressed (or "elided" NPs) tend to be immune to some interpretive and grammatical constraints which are obligatorily imposed on their overt counterparts. To account for this puzzling phenomenon, I argue that the interpretive flexibility of the elided NPs can be reduced to the selective reproduction of relevant nominal features at LF. I propose what I call the "selective LF-copy approach" to the NPE construction, in which covert computation for NPE involves reproduction of minimally necessary features than total lexical contents of the antecedent NP. Assuming that NPE is a base-generated empty skeletal pro-form, I postulate a new feature system for nominal expressions that makes reference to the semantic referential feature [REF]. I implement a theory in which the interpretive flexibility of NPE is attributed to, first, the feature composition of antecedent nominal expressions and, second, how much of this composition is reproduced at the ellipsis site and then comes to be interpreted. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]