Colloquium: The Architecture of Right Dislocation in Korean and Japanese

November 5, 3:30-5pm, Cobb 201
James Yoon, University of Illinois at Urbana-Champaign

Right Dislocation (RD, or Postposing, as it is sometimes called) is attested in strictly head-final languages like Korean and Japanese. The traditional idea within generative grammar concerning RD was that it is just another manifestation of free word order that characterizes these languages and should therefore be analyzed in a similar manner, i.e., by rightward movement/scrambling (Choe 1988; Simon 1989; Ko and Choi 2009, for a recent defense of this analysis).

However, most recent analyses of RD take a different view. In recent work, RD is viewed as a bi-clausal construction where the right dislocated constituent (Appendix) is an element of the second clause that—(i) is coindexed with a coreferential constituent in the preceding clause, (ii) undergoes leftward movement, and (iii) remains after ellipsis of the second clause (Endo 1996; Tanaka 2001; Yamashita 2008; Chung 2008, among others). The bi-clausal analysis of RD is similar to recent proposals for other elliptical constructions such as sluicing and fragment answers (Merchant 2001, 2004) and split questions (Arregi 2007).

In this talk we defend an analysis of RD that shares certain properties of the reigning bi-clausal analyses. However, we differ from previous accounts in how the Appendix is licensed. In our analysis, the Appendix is not licensed by movement—leftward or rightward, but is base-generated. And while we take the Appendix in RD to be a fragment, we do not generate it by deletion (following movement). Instead, the Appendix locally licenses a base-generated null predicate. The partial feature structure associated with the Appendix and the null predicate it licenses is unified with that of the preceding clause, yielding the interpretation of RD (Sells 1999). Since the interpretive connection between the Appendix and the preceding clause is established by coindexation of predicates and feature unification, the coindexation of the Appendix with a correspondent in the preceding clause is downplayed. The proposed analysis can thus be naturally extended to RDs where the Appendix does not have a correspondent in the preceding clause.

We shall also discuss the (‘parametric’) properties of the grammar of languages like Korean and Japanese that are implicated in licensing the kind of ellipsis we claim is attested in RD.