Not only does Russell Smith not know when he's
wrong; he doesn't even seem to know when he's right. In his latest column ("Two small words, one large headache", Jan 4, 2007), he writes,
"The downside to this pontificating job is that when one makes a grammatical mistake, it makes one look particularly silly. And readers are likely to point it out with a pleasure verging on glee. (As well they should.) So, yes, it was inelegant of me to begin last week's column with the sentence, 'Readers write to me about perceived errors which I don't always see as errors myself.' Although many language authorities increasingly accept "which" in this role, strictly speaking the 'which' should be a 'that.'"
And when one accepts the criticism that a perfectly grammatical usage is a mistake, one looks even sillier.
The facts: (mostly from Merriam Webster's Usage Dictionary)
That was used to introduce relative clauses (both restrictive and non-) in early middle English, while
which began to be used as a relative pronoun only in the 1300s. Later, it gained currency and within 300 years there was little to distinguish the two. Thereafter,
that fell completely out of use in this context for a period, only to later return in its full role. This was followed by a gradual fall out of fashion in nonrestrictive relative clauses until it became quite rare in that usage by the early 20th century. Currently,
that is not used this way.
It was about then, perhaps slightly earlier, that usage writers, including the Fowler brothers, suggested that, because
that was specializing for restrictive relatives, perhaps,
which should be reserved for nonrestrictive relatives. Note, this was merely a fancy. There is no historical precedent and there is an overwhelming host of examples from the best writers of the time in contravention of this idea, a fact readily admitted to by Fowler. Even the usage writers who later mistook this suggestion as a rule ignore it, often within pages of asserting it.
Thus, it is neither true that "strictly speaking"
that is correct, nor that
which is increasingly accepted in this role. In fact, if anything, more and more people, influenced by Microsoft Word's grammar checker, have come to believe that
which is incorrect.
Some interesting divergences from the above follow:
That cannot be used when you wish to relativise the object of a preposition in either restrictive or non-restrictive cases unless you "strand" the preposition (which is perfectly grammatical).
The problems of
which I’m aware are of a more qualitative type.
*The problems of
that I’m aware are of a more qualitative type.
The problems, of
which I’m aware, are of a more qualitative type.
*The problems, of
that I’m aware, are of a more qualitative type.
And, from Fowler (1908), we get the following:
All
that I can do is useless.
*All
which I can do is useless.
That
which has gone before should be clear.
*That
that has gone before should be clear.
Smith also writes,
"and you see the easiest way to tell when you need one and not the other? It's the commas. 'Which' will always be preceded by a comma. . (sic) So there you go: Forget about restrictive and non-restrictive. If your second clause needs commas around it, then introduce it with which. If not, keep it as that. This is going to be the most useful rule you'll ever learn from me."
I certainly hope not. As we've seen, the rule is not true. And even if it were true, it is not useful. How is one to know whether commas should be added? Take for example, the change to FIFA's Laws of the Game, where some well-meaning person, likely at the suggestion of MS Word, has changed a perfectly acceptable sentence to one that is inane, to wit: " A tackle, which endangers the safety of an opponent, must be sanctioned as serious foul play." (
Source)