I propose that the dative and locative case particles are not postpositions but morphological case markers, quite in contrast to the common view in the literature (Yang 1972, Cho and Sells 1995, and Suh 2013 among many others). I will show the difference between the nominative and the accusative and the dative and the locative in case drop, case stacking, and case spreading is attributed to the fact that the dative case and the locative case are inherent case, as compared with the nominative and the accusative, which are structural case par excellence. The present proposal has a nontrivial implication regarding the other case particles typically argued as postpositions in Korean.

This paper examines the exceptional behavior of P-stranding under Sluicing in some languages which do not allow P-stranding under regular whmovement (cf. Merchant (2001)). Spanish, Polish, Brazilian Portuguese and Indonesian, in contrast to Greek and Czech, are in principle a prepositionpied- piping language, but they allow P-stranding under Sluicing that arguably derives from cleft structure or via repair-by-ellipsis. Still critical to this process of P-stranding under Sluicing is the identity-in-ellipsis requirement that the morpho-syntactic form of the surviving wh-expression in the sluiced clause matches the one that its overt or covert correlate expression would potentially take if it underwent regular wh-movement. Extending this generalization to ‘Sluicing’ and ‘Fragmenting’ in Korean, we show that the somewhat unexpected postposition stranding and omission of the wh- and fragment expression in these constructions of Korean also derive from (pseudo-)cleft structure, meeting the identity/parallelism condition on ellipsis.