Left/right-node-raising in Korean

Abstract In this experimental study, we explore the syntactic constructions of left-node-raising (LNR) and right-node-raising (RNR) in Korean, examining whether LNR functions as a mirror image of RNR, as suggested in previous work by Yatabe (2001) , Nakao ( 2009 ; 2010 ), and Chung (2010) . Our findings from formal experiments indicate the existence of a unique left/right-node-raising (LRNR) construction in Korean, which we propose as a true mirror of RNR. We analyze RNR and LRNR through the framework of multidominance theory, drawing specifically on Wilder’s ( 1999 ; 2008 ) in situ linearization approach and Citko’s (2005) movement-based linearization approach, which provide insights into how shared elements (i.e. pivots) in these constructions are interpreted syntactically and semantically across conjuncts. For the derivation of LNR, however, we propose an alternative approach, namely, a scrambling-plus- pro analysis. Our experiment centers on acceptability judgments of sentences featuring dependent-plural markers (DPMs) within pivots across these constructions, revealing that DPMs influence acceptability differently in each structure. Specifically, DPMs mildly impact the acceptability of RNR but more strongly affect LRNR, which we attribute to distinct linearization schemes in each construction. In contrast, the presence of DPMs within pivots severely degrades the acceptability of LNR. Our study concludes that RNR and LRNR are more effectively explained by multidominance theory with in situ and/or movement-based linearization, while LNR aligns better with the scrambling-plus- pro analysis due to its unique derivational properties. Thus, our findings support the claim by Kim et al. (2023) that LNR is not a direct mirror of RNR.