TAG derivation as monotonic C-command

The TAG adjunction operation operates by splitting a tree at one node, which we will call the adjunction site. In the resulting structure, the sub-trees above and below the adjunction site are separated by, and connected with, the auxiliary tree used in the composition. As the adjunction site is thus split into two nodes, with a copy in each subtree, a natural way of formalizing the adjunction operation posits that each potential adjunction site is in fact represented by two distinct nodes. In the FTAG formalism (Vijay-Shanker, 1988) each potential adjunction site is associated with two feature structures , one for each copy. As an alternative to this operationally deened rewriting view of adjunction, Vijay-Shanker (1992) suggests that TAG derivations instead be viewed as a monotonic growth of structural assertions that characterize the structures being composed. This proposal rests crucially on the assumption that the elementary trees are characterized in terms of a domination relation among nodes, and that each potential adjunction site is represented by two nodes standing in a domination relation. Under this proposal, the structures and in Figure 1 would be used to derive long-distance wh-movement. To adjoin into , the root and foot nodes of are identiied with the two C 0 nodes standing in a domination relation in (represented by the dotted line). This domination relation still holds after adjunction, as do all the other domination relations stated in deening and. (In sentences in which there is no adjoining at the C 0 node, e.g., `I wonder what Mary saw,' these C 0 nodes could collapse, preserving domination under the assumption that it is a reeexive relation.) Domination has also been argued to play a role in multi-component structures, where there is assumed to be a domination relationship between a frontier node of one component and the root of the other. While the use of domination relationships is attractive in allowing us to view TAG derivations as Thanks to Tony Kroch, Seth Kulick, and two anonymous reviewers for helpful comments and discussion. We gratefully acknowledge the nancial support of NSF grants SBR-97-10247 and SBR-97-10411. monotonic additions to a set of domination relations, the linguistic motivation for such domination statements among duplicated nodes is not very clear. Instead , from the point of view of the grammar, what seems to be crucial in deening the relevant portion of the structure of is not that there should …