ECM

In the section on resultatives, we saw that resultatives are always taken to modify objects--and we saw that in order to state the conditions on resultatives properly, we needed a notion of 'object' which need not necessarily refer to an NP that is selected by the verb. So, for example, the resultative in (1) modifies an NP that is selected by the verb, but the resultative in (2) modifies an NP that is not:

(1) I painted the table red.
(2) We laughed ourselves sick.

In (2), the object of the verb, ourselves, doesn't seem to be selected by the verb; that's why (3) is odd:

(3) *We laughed ourselves.

In this section we'll develop a similar kind of argument for movement, based on another construction in which an NP is associated with a verb that does not select it. The construction in question is called Exceptional Case Marking (or ECM), and it's exemplified in (4) (compare (5), with a similar meaning):

(4) I believe him to be a liar.
(5) I believe he is a liar.

What's interesting about (4) is that him is acting in some ways like it's an object--it shows up as him rather than he, for example. But it clearly isn't selected by believe. The phrase selected by believe is typically the person or thing that the subject thinks is right, as in (5) (where believe selects the whole sentence he is a liar), or (6) below:

(6) I believe him.

We can also see that believe isn't selecting he in (4) by considering examples like (7):

(7) I believe it to be obvious that Syntax is fun.

In (7), we can tell that the NP following believe isn't selected by believe, since that NP is in fact an expletive--it's just there to satisfy EPP in the clause following believe. Expletives only show up when a subject position isn't filled by something that's been selected, so the fact that we're seeing one here shows that believe isn't selecting the NP after it (in fact, in (7), nothing is selecting the phrase which ends up in that position).

Now that we know all that about ECM verbs, we can use them as an argument for movement. Remember, again, the two theories we're trying to decide between, about why subjects of passives seem to violate our rules about selection by appearing far away from the verb that presumably selects them:

Relax the Projection Principle. For instance, the Projection Principle could say something like "the phrase selected by the verb is its complement under certain conditions, but is the specifier of TP under other conditions."
Leave the Projection Principle as stated, and posit a type of operation that moves the object of the verb, in this kind of case, from its original position as the sister of the verb up to the specifier of TP.

Now, consider an example like (8):

(8) He is believed to be a liar.

In (8), what we've done is start with an ECM example, something like the one up above in (4), and then we've made believe passive. We've established that in (4), him is not getting selected by believe. So "relaxing the Projection Principle" won't help us here; if the result of passivization were just to cause the selection relation that normally holds between the verb and its sister to hold between the verb and the specifier of TP instead, then the result in (8) would be very surprising, since he is not in fact selected by the verb here.

On the other hand, a movement account could deal with (8); all we have to do is say that he has moved, in (8), from its position as the object of believe to the specifier of TP.

Back to A-movement.