An experimental investigation of P/Case‑drop in Korean gapping

This study examines two syntactic analyses of P/Case-drop in Korean gapping (aka right-node-raising or right-peripheral ellipsis): LF copying (Abe & Hoshi 1997) and PF deletion (Kim 1997). We employ two online acceptability rating experiments to investigate to what extent the distribution of P/Case-drop is controlled by grammatical and extra-grammatical constraints. The experimental findings suggest that (a) linear non-parallelism elicits a processing cost for gapping and (b) P/Case-drop is a costly operation, which results in higher frequencies of PP fragments (in relation to NP fragments) and Case-marked NP fragments (in relation to Case-less NP fragments). We argue that the parallelism effect follows from the parser’s general preference to keep the structure of each conjunct maximally parallel in a coordination structure (Kim et al. 2020). Given this, we conclude that P/Case-drop phenomena in Korean gapping are better explained by a PF deletion analysis, supplemented with extra deletion (An 2016, 2019; Erschler 2022) and ellipsis parallelism (Frazier, Munn & Clifton 2000; Kehler 2000; Frazier & Clifton 2001; Carlson 2002), rather than by an LF copying analysis.