Fragmentary utterances refer to short answers smaller than grammatically complete sentence. Interestingly, the fragments convey the same propositional content that their full sentential counterparts do and have assertoric force. To capture this form-function mismatch, two types of analyses have been made so far: direct interpretation analyses and ellipsis analyses. The former assumes that fragments don't have hidden sentential structure and consist of non-sentential XPs. Hence, the expository burden from form-function mismatch is placed on syntax-semantics. The latter assumes that fragments have full sentential structure prior to ellipsis. Hence, compared with direct interpretation analyses, less simple syntactic structure is assumed. The ellipsis analyses have a crucial advantage such as preservation of usual syntax-semantic mapping: uniform factors in interpretation must stem from uniform syntactic source. This paper aims to show that the ellipsis analyses are more plausible in explaining grammatical properties of fragments. We claim that ellipsis approach is superior to alternatives in explaining both case-marked and non-case-marked fragments in Korean. Diverse evidence from case-matching, scope, P-stranding is presented to confirm our analysis.