Bridge Verbs

Verbs that select sentential complements restrict the <mode> and <comp> values for those complements. For verbs which allow long distance extraction from their sentential complements (bridge verbs), the S complement nodes are adjunction nodes. Figure 8.2 shows the declarative tree for a sentential complement-taking bridge verb, anchored by think.

{Sentential complement tree: $\beta$nx0Vs1

   

The need for an adjunction node rather than a substitution node at S₁ may not be obvious until one considers the derivation of sentences with long distance extractions. For example, the declarative in ((37)) is derived by adjoining the tree in Figure 8.3(b) to the S₁ node of the tree in Figure 8.3(a). Since there are no bottom features on S₁, the same final result could have been achieved with a substitution node at S₁.

(36)0(36

(37)

The emu thinks that the aardvark smells terrible . 

(a)		(b)

{Trees for The emu thinks that the aardvark smells terrible .

   

However, adjunction is crucial in deriving sentences with long distance extraction, as in sentences ((38)) and ((39)).

(37)0(37

(38)

Who does the emu think smells terrible ?  (38)0(38

(39)

Who did the elephant think the panda heard the emu say smells terrible ? 

The example in ((38)) is derived from the trees for who smells terrible? shown in Figure  8.4 and the emu thinks S shown in Figure 8.3(b), by adjoining the latter at the S_r node of the former.^8.8 This process is recursive, allowing sentences like ((39)). Such a representation has been shown by [#!kj85!#] to be well-suited for describing unbounded dependencies.

{Tree for Who smells terrible?

   

In English, a complementizer may not appear on a complement with an extracted subject (the `that-trace' configuration). This phenomenon is illustrated in ((40))-((42)):

(39)0(39

(40)

Which animal did the giraffe say that he likes ?  (40)0(40

(41)

Which animal did the giraffe say that likes him ?  (41)0(41

(42)

Which animal did the giraffe say likes him ? 

These sentences are derived in XTAG by adjoining the tree for did the giraffe say S at the S_r node of the tree for either which animal likes him (to yield sentence ((42))) or which animal he likes (to yield sentence ((40))). That-trace violations are blocked by the presence of the feature <assign-comp>=inf nil/ind nil/ecm feature on the bottom of the S_r node of trees with extracted subjects (W0), i.e. those used in sentences such as ((41)) and ((42)). If a complementizer tree, COMPs, adjoins to a subject extraction tree at S_r, its <assign-comp> = that/whether/for/if feature will clash and the derivation will fail. If there is no complementizer, there is no feature clash, and this will permit the derivation of sentences like ((42)), or of ECM constructions, in which case the ECM verb will have <assign-comp>=ecm (see section 8.6.4 for more discussion of the ECM case). Complementizers may adjoin normally to object extraction trees such as those used in sentence ((40)), and so object extraction trees have no value for the <assign-comp> feature.