Wednesday 24 September 2014

On a question that (it) is hard to answer

A correspondent writes to ask about the use of it in relative clauses, in such sentences as the following (taken from Fowler and also a modern textbook). He finds its use unidiomatic in examples (3) and (5) in particular. Is the it omissible, he asks?

(1) This was a conference which it was my duty to attend.
(2) The debate on the bill produced a tangle of arguments which it required all Mr. Chamberlain's skill to untie.
(3) This is a thing which it is easy to say.
(4) The heaving and turbulent centuries which at one time it was the fashion to characterize the 'Dark Ages' have long had a peculiar fascination for historians.
(5) That is a question which it is very hard to answer.

These are quite complex syntactically, as they all have a nonfinite clause inside a cleft construction inside a relative clause. To see what's going on we need to simplify. Let's get rid of the clefting first.

(1) My duty was to attend (the conference).
(2) Mr Chamberlain's skill was required to untie (the tangle of arguments).

The semantic links are clear: duty goes primarily with attend, not conference. Skill goes primarily with untie, not arguments. If the it were omitted in the original examples, the force of the relative pronoun would be to point the listener/reader semantically backwards, towards the head noun. Duty would now seem to go with conference, and skill with arguments. The it restores the right semantic connection. So, in short: we avoid a potential ambiguity - though the fact that there's so much usage variation (the it often being omitted) suggests that it isn't one that causes much communicative difficulty.

The ambiguity is there in (3) and (5), but the shortness of the sentences, along with the clear meaning of the elements, makes the presence of it less needed - which is why my correspondent has noticed it when it's inserted.

It is easy to say (this thing).
It is hard to answer (the question).

It's obvious that things don't do the speaking or that questions don't do the answering, so semantically there's no need to reinforce the point when the clefts are restored.

This is a thing which is easy to say.
That is a question which is very hard to answer

Only someone ignoring the semantics would say there's a genuine ambiguity here. But traditional grammarians, obsessed with making a rule work in all cases, did regularly ignore semantics. And anyone following those rules will insist on inserting the it in these cases, probably on the grounds that it helps avoid a possible momentary distraction. From a psycholinguistic point of view, there may be a point here, but it's hardly one that's likely to cause communicative interference. I doubt whether most people would ever even notice that an it was omitted in (3) and (5). And some, such as my correspondent, evidently find the usage with it intrusive.

(4) is a special case, as it's a badly constructed sentence, which could do with being rephrased anyway! Try reducing the sentence to its basic form and you'll see what I mean.


Phillip Minden said...

"This was a conference which it was my duty to attend." To attend was may duty. My duty wasn't a conference.

"This is a thing which it is easy to say." A thing is actually what is easy to say.

John Cowan said...

For me at least, the matter is far more cut and dried. (1) and (4) are grammatical with it and ungrammatical without, and (2), (3), and (5) are ungrammatical with it and grammatical without it. As written, (2), (3), and (5) are every bit as bad as This was a man who he was very tall; in other worse, they have resumptive subject pronouns in relative clauses.

There is a problem with (4), which is that characterize for me demands as as its complement; to keep the existing structure, I would change the verb to call.

DC said...

Lipman: not sure what point you;re making.

John Cowan: I think judgments will vary depending on whether one is encountering these constructions in writing or in speech.

Unknown said...

I’m just a learner but Philip’s –(1) correct, (3) wrong– and John’s comments sound perfectly logical to me. They even match the “underlying logic” of the equivalent Spanish structures. (I admit, though, that speech doesn’t have to be logical to make sense).

David Crosbie said...

Yes, it does seem to boil down to taste. I find it in (3) and (5) uncongenial, but I wouldn't go as far as John in calling them ungrammatical. I'm happier with the it in (2).

I agree, though, that the absence of as after characterize is a serious blemish.

There seems to be a difference in the type of relative clause — often difficult to pin down when the antecedent is grammatically indefinite.

• The antecedent in (4) is semantically unique and grammatically definite.

• The antecedents in (1), (3) and (5) strike me as generic: the sort of conference, the sort of thing, the sort of question.

• The antecedent in (2) strikes me as indefinite but unique: a specific outcome which may be characterised as 'a tangle of arguments'. Certainly not (well, not for me) 'the sort of tangle of arguments'.

I've never forgotten a distinction made by Pit Corder (I don't know whether it was his invention) between

She wants to marry a Norwegian. (She has a thing about Vikings.)


She wants to marry a Norwegian. (He's a nice lad from Oslo.)

TonyTheProf said...

The traditional grammarians certainly erected a barrier to the removal of "it" in (3) and (5). But their influence has been waning, and I would suggest that the forms of grammatical structure may differ depending on whether the words are written or spoken. Speech tends towards redundancy hence the removal of what is superfluous, but there is evidently a lag between what is written and what is spoken. I'd suggest that the spoken leads the way in linguistic change, while the written catches up, but that gap has been shortening. As transcribed speech and reported speech grows, with less editorial corrections to grammar than would have been the case, say 100 years go, the lag between changes in speech and writing have diminished.

DC said...

Yes, as I suggested in my earlier comment, the speech/writing distinction is central. One could even re-analyse some of these examples as blends of competing structures, a common feature of spoken language.

Phillip Minden said...

I think it's the other way around, nice as it may be to stage the honest, simple speaker against the evil grammarian as if it were 1974. :-)

I think the awkward construction with the gratuitous pronoun in (3) and (5) was introduced not by grammarians but hypercorrectly by speakers or writers in order to sound more educated.

DC said...

But, as my corespondent mentions, it was Henry Fowler who drew so much attention to the issue.

Phillip Minden said...

I had a quick look at Fowler and at the depths of the internet, and Fowler seems to say exactly what I felt, while the internet indicates examples (3) and (5) are from a 1977 textbook by a German.

DC said...

That's correct, and it was the German textbook examples that motivated my correspondent in the first place.

David Crosbie said...

Is anybody happy with modifications of (3) and (5) so that it follows a plural antecedent?

These are things which it is easy to say.

Those are questions which it is very hard to answer.

Barry Cusack said...

This is a pretty puzzle!
DC says:
I doubt whether most people would ever even notice that an it was omitted in (3) and (5). And some, such as my correspondent, evidently find the usage with it intrusive.

The reason for this is not hard to see. The use of it in (3) and (5) arises from the elements into which the complex sentence can be broken down:
A. This is a thing.
B. It is easy to say this thing.
These two clearly can be linked to yield:
C This is a thing which it is easy to say.
B. here is a cleft sentence – alternatively described as being an example of “preparatory or anticipatory” it. This results, as DC says, in: a nonfinite clause inside a cleft construction inside a relative clause i.e. the form of (3) and (5), and C above.

A different case is these two elements – one of which is different grammatically, though only very slightly in meaning:
A. This is a thing.
D This thing is easy to say.
These two can be linked to yield:
E This is a thing which is easy to say.

E is by far the most unproblematic form, and I would guess, far more used than C, which is the form of (3) and (5) in DC’s list. Therefore it is easily recognized as obviously correct. However, (3) is A and B linked, not A and D linked. And when we hear or see (3) or (5) we find it a little unfamiliar, we wonder what it is, and prefer the well-known structure, E, without it.
DC poses the question of his correspondent: Is the it omissible?
In (3) and (5) the answer is: there are two possible, different constructions, one with, one without, "it".

Barry Cusack said...

David Crosbie:

That’s better! With a plural subject, there is no ambiguity over what the problematic “it”
refers to. It clearly can’t now refer to the plural subject. Therefore it must represent the anticipatory “it” of one of the (notionally) constituent elements which have been joined to make the complex sentence.
And, of course, this “it” represents the anticipatory “it” regardless of whether the subject is singular or plural.
(For some reason, “Those are questions which it is very hard to answer” seems more obviously acceptable/congenial than the other.)

@BobKLite said...

Those two (with plural antecedents) are OK by me. (And - à propos of nothing in particular

b) I think it was one of the Prof's books that traces that 'by me' to the German bei mir.)

Unknown said...

So, would the expression “Those are questions which it is very hard to answer” be considered correct by an English examiner if a foreign student used it in a language proficiency test or something like that?

DC said...

I imagine it would depend on the examiners - how aware they are of the usage issues raised by a construction of this kind. I'm afraid I don't know what happens in actual practice.

Chris from Germany said...

I have asked two native speakers of English whether or not they think the pronoun "it" is required in the five sentences posted above.

First native speaker (UK):
"It" is not required in sentences (2), (3) and (5).
Sentences (1) and (4) are ungrammatical without the pronoun "it".

Second native speaker (US):
The sentences are all wrong; the pronoun "it" makes the sentences ungrammatical. Wow, I would definitely not have expected such a reply to my question. By the way, I still prefer the reply given by DC above; it makes perfect sense to me, and I wholeheartedly agree with his explanation of this "grammar oddity".