Exceptional Case Marking Verbs

Tree family: TXnx0Vs1 Exceptional Case Marking verbs are those which assign accusative case to the subject of the sentential complement. This is in contrast to verbs in the Tnx0Vnx1s2 family (section 6.6), which assign accusative case to an NP which is not part of the sentential complement. The subject of an ECM infinitive complement is assigned accusative case is a manner analogous to that of a subject in a for-to construction, as described in section 8.5. As in the for-to case, the ECM verb assigns accusative case into the subject of the lower infinitive, and so the infinitive uses the to which has no value for <assign-case> and has <assign-comp>=for/ecm. The ECM verb has <assign-comp>=ecm and <assign-case>=acc on its foot. The former allows the <assign-comp> features of the ECM verb and the to tree to unify, and so be used together, and the latter assigns the accusative case to the lower subject. Figure 8.5 shows the declarative tree for the tree for the TXnx0Vs1 family, in this case anchored by expects. Figure 8.6 shows a parse for Van expects Bob to talk

   

Figure: ECM tree: $\beta$Xnx0Vs1

  

Figure 8.6: Sample ECM parse

The ECM and for-to cases are analogous in how they are used together with the correct infinitival to to assign accusative case to the subject of the lower infinitive. However, they are different in that for is blocked along with other complementizers in subject extraction contexts, as discussed in section 8.6, as in ((43)), while subject extraction is compatible with ECM cases, as in ((44)).

(42)0(42

(43)

$\ast$What child did the giraffe ask for to leave?  (43)0(43

(44)

Who did Bill expect to eat beans? 

Sentence ((43)) is ruled out by the <assign-comp>= inf nil/ind nil/ecm feature on the subject extraction tree for ask, since the <assign-comp>=for feature from the for tree will fail to unify. However, ((44)) will be allowed since <assign-comp>=ecm feature on the expect tree will unify with the foot of the ECM verb tree. The use of features allows the ECM and for-to constructions to act the same for exceptional case assignment, while also being distinguished for that-trace violations. Verbs that take bare infinitives, as in ((45)), are also treated as ECM verbs, the only difference being that their foot feature has <mode>=base instead of <mode>=inf. Since the complement does not have to, there is no question of using the to tree for allowing accusative case to be assigned. Instead, verbs with <mode>=base allow either accusative or nominative case to be assigned to the subject, and the foot of the ECM bare infinitive tree forces it to be accusative by its <assign-case>=acc value at its foot node unifies with the <assign-case>=nom/acc value of the bare infinitive clause.

(44)0(44

(45)

Bob sees the harmonica fall. 

The trees in the TXnx0Vs1 family are generally parallel to those in the Tnx0Vs1 family, except for the <assign-case> and <assign-comp> values on the foot nodes. However, the TXnx0Vs1 family also includes a tree for the passive, which of course is not included in the Tnx0Vs1 family. Unlike all the other trees in the TXnx0Vs1 family, the passive tree is not rooted in S, and is instead a VP auxiliary tree. Since the subject of the infinitive is not thematically selected by the ECM verb, it is not part of the ECM verb's tree, and so it cannot be part of the passive tree. Therefore, the passive acts as a raising verb (see section 9.3). For example, to derive ((47)), the tree in Figure 8.7 would adjoin into a derivation for Bob to talk at the VP node (and the <mode>=passive feature, not shown, forces the auxiliary to adjoin in, as for other passives, as described in chapter 12).

(45)0(45

(46)

Van expects Bob to talk.  (46)0(46

(47)

Bob was expected to talk. 

   

Figure 8.7: ECM passive

It has long been noted that passives of both full and bare infinitive ECM constructions are full infinitives, as in ((47)) and ((49)).

(47)0(47

(48)

Bob sees the harmonica fall.  (48)0(48

(49)

The harmonica was seen to fall.  (49)0(49

(50)

$\ast$The harmonica was seen fall. 

Under the TAG ECM analysis, this fact is easy to implement. The foot node of the ECM passive tree is simply set to have <mode>=inf, which prevents the derivation of ((50)). Therefore, for all the other trees in the family, to foot nodes are set to have <mode>=base or <mode>=inf depending on whether it is a bare infinitive or not. These foot nodes are all S nodes. The VP foot node of the passive tree, however, has <mode>=inf regardless.