The beginning of the Wikipedia entry on hair:
(1) Hair is a filamentous biomaterial, that grows from follicles found in the dermis.
At first this looks like that as a non-restrictive relativizer, but in fact which is not really an improvement:
(2) Hair is a filamentous biomaterial, which grows from follicles found in the dermis.
The intended reading is surely restrictive — corresponding to either of the alternatives:
(3a) Hair is a filamentous biomaterial that grows from follicles found in the dermis.
(3b) Hair is a filamentous biomaterial which grows from follicles found in the dermis.
Instead, (1) has the that from (3a) together with the punctuation of (2). How could that happen?
Suppose you typed (3b) in a word processor that comes with a grammar checker like the one for Microsoft Word. You’d then be offered two choices: (2) and (3a). See Geoff Pullum’s discussion on Language Log of the grammar-checker-induced error
Check all boxes, which apply to what you are looking for.
as a “correction” of
Check all boxes which apply to what you are looking for.
(“fixing” the punctuation). In this case, the Microsoft grammar-checker offered a second alternative:
Check all boxes that apply to what you are looking for.
(“fixing” the choice of relativizer). So one way to get to (1) would be to combine the two fixes — the first incorrect, the second merely unnecessary — to get a deeply wrong result.
Non-restrictive that. Things that look like (1) but are intended to be understood non-restrictively are surprisingly common. Here’s Lee Siegel reviewing Joyce Carol Oates’s The Gravedigger’s Daughter (NYT Book Review 6/17/07, p. 18):
By that point, you might just forgive writing that has become so rushed it sometimes appears semiliterate. Oates’s maddening habit of using “that” instead of “which” in a nonrestrictive clause — “most of the papers continued to run Chet Gallagher’s column, that had won national awards” — is too ugly and incoherent to be an attempt at stylistic innovation. Such indolence is surprising in a writer of Oates’s caliber.
Probably not a consequence of grammar-checking, but possibly an unfortunate consequence of which-avoidance, induced by a misunderstanding of “Fowler’s Rule”, which insists on that (rather than which) in restrictives. Aggressive but misguided which-avoidance can lead to unfortunate results, as Robert Coren noted in a comment on one of my recent which postings:
Several decades ago, my workplace had a tech writer who was a serious which-hunter, and she had so terrorized at least one of the developers that he produced a draft that avoided “which” entirely, using “that” even in non-restrictive clauses, resulting in pretty much unreadable, or at least incomprehensible, prose.
I haven’t searched Joyce Carol Oates’s work for examples of non-restrictive that, but I suspect they aren’t as frequent as Siegel supposes — and some of them might have been the work of over-zealous editor.
Some background: by Fowler’s time, non-restrictive non-human relatives in standard English used which and not that. Fowler suggested that things might be better if restrictive non-human relatives in standard English were similarly uniquely marked. Nobody was suggesting messing with the non-restrictives, which had been fixed (in standard English) for some time. The restrictives were the problem.
The usual formulations of the “rule” nevertheless do mention non-restrictives, because that in non-restrictives occurs with some frequency in non-standard English. So half of the “rule” is actually a description of the standard language. But putting the two clauses together has led some people to a misunderstanding of writers (like me) who reject Fowler’s Rule for restrictives: if we say that one half of the rule is nonsense, why believe the second half? (Bob Lieblich has reported to me people who have reasoned this way, so that they find no problem with non-restrictive that and believe that Language Log postings on Fowler’s Rule support their position.)
A final note: Francis Christensen’s Notes Toward a New Rhetoric (3rd ed., 2007) has a chapter on “Restrictive and nonrestrictive modifiers again” which treats the issue as a matter of punctuation; his discussion goes well beyond relative clauses, taking in a collection of “bound” (restrictive) vs. “free” modifiers. Which vs. that comes into it only incidentally.