Diatheses and VoiceP in Korean

Korean voice morphology reflects two distinct modules, which are differentiated by the [±NAT] feature on the root. The [+NAT] module is characterized by irregular, less productive suffixes and a higher degree of lexical conditioning, while the [-NAT] module uses systematic, highly productive suffixes that transparently mark diatheses. Despite varying surface morphology, these modules share the common structure VoiceP[±D], which is responsible for argument manipulation across the system. The differences in surface morphology between them emerge during Vocabulary Insertion: [+NAT] items rely on specific, lexically conditioned rules, resulting in irregular allomorphy and reduced productivity. On the other hand, the [-NAT] module follows general and productive rules and produces regular morphological patterns. Nevertheless, the [±NAT] distinction is superficial and hides the unified morphosyntactic structure underlying diatheses. Adopting Distributed Morphology, this analysis highlights how the uniform syntactic role of Voice interacts with lexical conditioning, thereby accounting for allomorphy, productivity and syncretism in Korean voice. This perspective deepens the theoretical understanding of the morphology-syntax interface.