Cleaving idioms with right-node-raising

Abstract

This paper examines right-node-raising (RNR) with idiom chunks. RNR sentences allow idiomatic interpretation when they contain the whole idiom chunk within the pivot (i.e., the shared element) (e.g., Jessica believed, but Zac doubted, that Justin popped the question.), but those containing only a part of the idiom within the pivot do not (e.g., #John kicked, and Mary filled, the bucket.). Given this, Woo (2015) argues for a multidominance approach (cf. Wilder 1999) to RNR in that the multiply dominated pivot must not be partially shared for idiomatic interpretation. However, we report that even if the pivot contains the whole idiom part, the issue of missing idiomatic interpretation in RNR still lingers (e.g., #We played a party game, and they used an ice hammer, to break the ice.). In order to deal with this problem, multidominance, movement, or PF deletion analyses must resort to an extra interpretive parallelism according to which a pivot cannot be used in two different senses simultaneously. From this perspective, we argue that an LF copying approach can explain the idiomaticity in RNR without extra proviso since under this analysis, it is not necessary to postulate a separate LF constraint of interpretive symmetry. We extend our analysis to Korean (and Japanese) data pertaining to RNR with idiomatic or polysemous expressions. We thus conclude that lexical mismatches and interpretive mismatches in English and Korean RNR are solid evidence of interpretive identity in RNR.