Annotating Particle Realization and Ellipsis in Korean

We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annotated data include 100,128 Ecel (a spacebased word unit) in spoken and written corpora composed of four different genres in order to evaluate how register variation contributes to Korean particle ellipsis. Identifying the grammatical functions of particles and zero particles is critical for deriving a valid linguistic analysis of argument realization, semantic and discourse analysis, and computational processes of parsing. The primary challenge is to design a reliable scheme for classifying particles while making a clear distinction between ellipsis and non-occurrences. We determine in detail issues involving particle annotation and present solutions. In addition to providing a statistical analysis and outcomes, we briefly discuss linguistic factors involving particle ellipsis.