Problems with a Movement Analysis of Right Node Raising in Tagalog

Bradley Larson

Sabbagh (2008) makes an empirical argument concerning the derivation of right node raising (RNR) sentences in Tagalog. He notes that all and only the syntactic constituents that can undergo typical leftward wh-movement can serve as the shared element, or target, of a RNR sentence. Therefore, he argues, a movement analysis would be the most plausible for Tagalog RNR. However, when investigated further, the parallelism does not hold. A greater variety of elements can serve as the RNR target than can be wh-moved, contrary to what Sabbagh claims. The fact that elements that cannot undergo A-bar movement can still act as RNR targets suggests that RNR in Tagalog is not derived via A-bar movement. The instances where RNR is not allowed can be explained by the interplay between morphologically realized specificity marking of Tagalog arguments and crosslinguistic information structure requirements of RNR sentences. The position immediately prior to the gaps in RNR must represent discourse-salient new information (as in Hartmann 2000, Ha 2007). The illicit RNR sentences that Sabbagh produces fail to be acceptable because the constituents in this position are marked with a ‘‘specific’’ determiner (Schachter and Otanes 1972, Kroeger 1993), which, being specific, requires a ‘‘previously established discourse referent’’ (Enc ̧ 1991:8). That is, the pregap expressions must simultaneously represent new information and refer to an established referent, a difficult feat. Discourse-initially this is not possible for obvious reasons, and it is correctly predicted that RNR sentences with pregap, specific-marked expressions are judged unacceptable. However, in particular contexts, the otherwise unacceptable sentences are judged acceptable.