In this paper we argue that the distributively interpreted RNRed shared material stems from joining together two separate parts of it which are generated at some point of derivation at the right-edge of each conjunct clause. The process of joining them together is subject to the identity condition. If both the first-conjunct and the second-conjunct right-edge elements have the same reference, one RNRed shared material appears at the right edge of the second-conjunct clause. However, if they have disjoint reference, they cannot be unified into one term. Rather, a coordinating conjunction is employed which combines them together. On the basis of the identity condition involved in the course of the process, we can account for Jacobson s (1999) effects. Before the first and the second right-edge elements are joined together, the pronouns inside them have to be licensed as bound variables in each conjunct. In that case they are legitimately unified into one single term. Otherwise, they cannot be, due to the identity condition. It is also proposed that the parallelism requirement for dependency relation comes into play in accounting for pronominal binding in the RNR construction. In this vein the RNR construction behaves in parallel fashion to the VP ellipsis construction.
2. Pronouns at right edges in the RNR construction
3. The Formation of RNR in English
4. Union in RNR and the Identity Condition