The pronoun it serves as an optional reduction of the fact. It can stand directly before that-clauses in sentences with factive verbs:

Bill resents it that people are always comparing him to Mozart

They didn’t mind it that a crowd was beginning to gather in the street.

Although the difference is a delicate one, and not always clearcut, most speakers find it unacceptable in the comparable non-factive cases:

*Bill claims it that people are always comparing him to Mozart

*They supposed it that a crowd was beginning to gather in the street.

This it, a reduced form of the fact, should be distinguished from the expletive it, a semantically empty prop which is automatically introduced in the place of extraposed complements in sentences like

It seems that both queens are trying to wriggle out of their commitments

It is obvious that Muriel has lost her marbles.

Rosenbaum (1967) tried to identify the two and to derive both from an it which he postulated in the deep structure of all noun clauses. This was in our opinion a mistake. In the first place, the two it’s have different distributions. Expletive it comes in regardless of whether a factive or non-factive clause is extraposed, and does not appear to be related to the lexical noun fact, as factive it is.

The relationship of factive it to the lexical noun fact, and its distinction from expletive it, is brought out rather clearly by a number of transformational processes. For example, the presence of factive it blocks the formation of relative clauses just as the lexical noun fact does:

*This is the book which you reported it that John plagiarized

*This is the book which you reported the fact that John plagiarized

This is the book which you reported that John plagiarized.

But expletive it differs in permitting relativization:

That’s the one thing which it is obvious that he hadn’t expected

*That’s the one thing which the fact is obvious that he hadn’t expected.

As Ross (1966) has shown, facts like these create seemingly insoluble problems for a system like Rosenbaum’s, in which factive and expletive it are derived from the same source. We have not proposed an alternative in anything like sufficient detail, but it is fairly clear that a system of rules constructed along the general lines informally sketched out here, which makes exactly the required syntactic distinction, will not have inherent difficulties in dealing with these facts.

Direct comparison of factive it and expletive it shows the expected semantic difference. The comparison can be carried out with the verbs which are indifferent as to factivity:

I had expected that there would be a big turnout (but only three people came)

I had expected it that there would be a big turnout (but this is ridiculous—get more chairs).

The second sentence, with it, suggests that the expectation was fulfilled, whereas the first is neutral in that respect. On the other hand, expletive it adds no factive meaning, and the following sentence is ambiguous as between the factive and non-factive interpretation:

It was expected that there would be a big turnout.

This analysis makes the prediction that cases of it which cannot be derived from fact will present no obstacle to relativization. This is indeed the case:

Goldbach’s conjecture, which I take it that you all know. . .

The report, which I will personally see to it that you get first thing in the morning. . .

This secret, which I would hate it if anyone ever revealed. . .

On the other hand, it is not too clear where these it’s do come from. Perhaps their source is the ‘vacuous extraposition’ postulated by Rosenbaum (1967).¹

The deep structures which we have posited for the two types of complements also explain the way in which they get pronominalized. In general, both factive and non-factive clauses take the pro-form it:

John supposed that Bill had done it, and Mary supposed it, too

John regretted that Bill had done it, and Mary regretted it, too.

But the two differ in that only non-factive clauses are pronominalized by so:

John supposed that Bill had done it, and Mary supposed so, too

*John regretted that Bill had done it, and Mary regretted so, too.

These facts can be explained on the basis of the fairly plausible assumptions that it is the pro-form of noun phrases, and so is the pro-form of sentences. Referring back to the deep structures, we see that the only node which exhaustively dominates factive complements is the node NP. For this reason the only pro-form for them is the pro-form for noun phrases, namely, it. But non-factive complements are exhaustively dominated by two nodes: NP and S. Accordingly, two pro-forms are available: the pro-form for noun phrases, it, and the pro-form for sentences, so.

¹ Dean (1967) has presented evidence from German and English that extraposition is the general source of expletive pronouns.