Last month, I wrote about the unfounded “rule” limiting which to non-restrictive clauses and that to restrictive clauses. I hoped to show that restrictive which is common, standard, and unobjectionable, and has been for centuries. I’ve been updating the post with subsequent commentary from editors and linguists.
Restrictive which is ubiquitous, but non-restrictive that is very rare. The Merriam-Webster Dictionary of English Usage has several citations, including Shakespeare (“Fleance his son, that keeps him company”) and Oliver Goldsmith (“Age, that lessens the enjoyment of life, increases our desire of living”). The same source says the construction is used mostly by poets.
On Language Log in 2005, Geoffrey Pullum posted the following data from the Longman Grammar of Spoken and Written English (Note: integrated relative = restrictive; supplementary relative = non-restrictive).
The remarkable bottom-row figures will have caught your eye. Pullum describes finding a non-restrictive that as “like spotting the syntactic analog of an ivory-billed woodpecker”.
So you’ll understand why I felt a moment of excitement when I read what I thought was a genuine example in the wild: specifically, in The Blue Flower by Penelope Fitzgerald (chapter 34, “The Garden-House”):
Then as he dried himself he repeated, ‘Never does the heart sigh in vain, Justen,’ and she scarcely knew whether to be unhappy or not. In her mouth was something bitter, that tasted like the waters of death.
Reading it again, though, I decided it was more likely a restrictive clause, albeit one set off by an unusual comma. The comma supplies a pause, but it probably doesn’t mark a supplementary relative clause. That is, Fitzgerald’s line is equivalent to this:
In her mouth was something bitter that tasted like the waters of death. [restrictive]
and not this:
In her mouth was something bitter, which tasted like the waters of death. [non-restrictive]
What do you think?
Update:
My excitement has diminished as I’ve become more accustomed to seeing the construction. I recently read Anne Enright’s novel The Gathering, which has several instances of non-restrictive that. Here are two of them:
Ada bringing us for red lemonade into a pub, that had a black roof with huge letters of white written across it.
If Ada believed in anything she believed in this persistence, that other people might call the soul.
And another, this one from Obstinate Uncle Otis, a short story by Robert Arthur:
Maybe he thought he c’d ignore that lightning, like he ignores Willoughby’s barn across the road, or Marble Hill, that his cousin Seth lawed away from him so that now he won’t admit there is any such hill.
A recent post on Language Log has an example from a comic strip, while Alex Segal, in a comment, shares several examples of non-restrictive that and discusses their distribution and grammaticality.
Sentence first 
I believe it is simply a slightly elided version of ‘In her mouth was something bitter, *something* that tasted like the waters of death’.
Thanks, Harry. I considered that possibility, but wanted to see what other readers would think. If that is what’s going on here, I can understand why she would omit the second something for stylistic reasons, but it generates an odd little ambiguity.
I’d say it’s either a strangely punctuated restrictive clause or some sort of elliptical construction, as Harry suggested. It doesn’t look like a real nonrestrictive that to me.
I find both theories equally plausible and so (like the good Father Gassalasca Jape, S.J.) I will believe them both.
Jonathon: Nor me. As a non-restrictive that it would be semantically unsound, I think.
John: A sensible policy too seldom adopted. But do let me know if one theory ultimately gets the upper hand.
If the choice is between an unusual “that” and an unusual comma, I’d vote for the unusual comma. There aren’t really a lot of firm rules for commas, which is a good thing, as it lets us use them to indicate pauses, as you say.
It would be interesting to know whether people who hear words in their minds when they read use commas differently than people who don’t. I suspect that they do, but I have no idea how you’d find out.
One other thought: does the meaning of the sentence depend at all on whether the “that” is restrictive?
An interesting case indeed! I think what’s making it hard to categorize for me is that “bitter” and “tasted like the waters of death” are in the same semantic category. If it were instead
“something pointy, that tasted like the waters of death”,
I think I would be in the restrictive-with-an-odd-comma camp. But it’s hard for me to think of the relative clause as restrictive or nonrestrictive in the true quotation because its information is somewhat redundant.
I like Harry’s suggestion of an elliptical construction, side stepping the trouble. But if I had to say that this was a relative clause, I think I’d come down on the side of non-restrictive, just because of the semantic similarity between “bitter” and “waters of death”.
My favourite use of which is in Patrick O’ Brian’s Aubrey Maturin series.Captain Aubrey’s steward Preserved Killick (and other members of the “lower orders”)
Aubrey: “Killick! rouse ut my Number one coat and scraper!”
Killick: “Which it’s laid out in your cabin (extend pause), sir”
Wonderful stuff!
Alan: Yes, the meaning of the sentence changes depending on the type of clause. If it’s restrictive, the fact that the taste in her mouth was “like the waters of death” is essential. If it’s non-restrictive, the phrase is dispensable. But it’s not a significant difference except grammatically.
Gabe: Thanks for your analysis. The semantic correspondence between “bitter” and “tasted like the waters of death” does seem to support Harry’s interpretation. Short of asking the author her intention, though, it’s likely to remain puzzling.
Jams: Marvellous! I must find time for more O’Brian; he’s a joy to read.
Great post, but in the example, for the sake of poetry, flow and emphasis, I’d have repeated the ‘something’ and used ‘that’. Then again, I’m all about the poetry and flow :-)
My American English ear can only hear the phrase as restrictive. I want to take out the comma. But, then, Americans tend to use “that” for restrictive clauses and “which” for nonrestrictive. I have a hard time reading them any other way.
bloowillbooks: I too would prefer the repetition of “something” for better flow and clarity. However, I wonder how long it would have taken me to consider that possible “fix” (or writer’s intent) without Harry’s comment and Stan’s reply.
Great discussion! Happy to be here.
bloowillbooks: Thank you. I’m all for poetry and flow too, but repeating something in that sentence would seem heavy-handed to me.
Erin: That’s the feeling I hoped to undermine with my earlier post! I realise it’s a losing battle.
G. Marie: It’s impossible to know, but fun to ponder. Thanks for joining in.
I agree, Stan, that repeating “something” would seem heavy-handed. I understand that it could be an elision, but to me — yes, even with an American English ear — the “that” stands on its own, and, oddly enough, sounds better *with* the comma. It doesn’t make sense to my American-copyeditor brain, but it somehow does to my ear. I would label it a “semirestrictive” clause, if such a thing existed.
And to your general question of spotting the nonrestrictive “that” in the wild, I most definitely have — in unedited prose. I may be alone on this, but I’ve often seen nonrestrictive clauses offset by “that” in the prose of untrained writers; I’ve seen it enough, in fact, that I haven’t bothered to note the various sightings. Maybe I should start! I do find it curious that no other editors have encountered the nonrestrictive “that” in unedited content . . . maybe because I work with more amateur writers than most do?
Interesting. The elided “something”, although apparently strained to some, seemed a given to me and, to my ear, the sentence flows more naturally without it being explicitly stated. We have to remember that the author is hearing this sentence with her own ear and may not even be aware that others would hear it differently. Not a mark of a good writer or a good editor but still, encountered often enough to be unremarkable.
Jared: I like how it reads, too. Fitzgerald is a fine prose stylist and may be presumed to have known exactly what she was doing. I put the line out of my mind for a day or two, and upon revisiting it I felt again that it was a restrictive clause with a comma introduced to create a slight pause necessary for the desired effect. The “elided something” hypothesis is also still perfectly plausible. (And yes, I think you should start collecting examples of non-restrictive that!)
David: The author would have heard it with her own ear, yes, but I’ve no doubt she considered how it would appear and sound to others, too, inasmuch as any writer can. The line would also have been read by other parties (proofreader, editor(s), perhaps friends or family); I wonder how they interpreted it, and whether there was any discussion about it or if its grammar simply went unnoticed.
To my mind, the most important feature in this example is that [bitter] and [tasted like the waters of death] are not independent attributes, but coupled in that one reinforces the other. The waters of death are the epitome of bitterness. Another way to express the same meaning would be, “In her mouth was something so bitter, it tasted like the waters of death“, or more simply, “In her mouth was something bitter, very bitter indeed.”
So if we want to label it as some particular type of clause, it has to be a type that can serve this role of reinforcing and expanding upon a description just given. I don’t find the version without the comma (*”In her mouth was something bitter that tasted like the waters of death“) to be acceptable; it makes the redundancy look like an error rather than a stylistic effect, and begs to have the word “bitter” removed. As for the “which” version (“In her mouth was something bitter, which tasted like the waters of death“), if forced to pick one of the two solutions I would go with this, but that doesn’t mean I like it much. In the end, I think the restrictive vs supplementary distinction is a red herring in this case, and that one needs to look along other axes of grammatical possibility to analyse it properly. The solutions involving some form of elision are probably on the right track.
Thanks for your thought-provoking analysis, Adrian. It’s true that one element could serve to reinforce the other, but I don’t think this necessarily disqualifies the restrictive, comma-less interpretation — a version that to me holds up OK and doesn’t warrant an asterisk marking unacceptability.
It might help to imagine a more-or-less equivalent sentence with different terms, e.g., “In her mouth was something sweet, that tasted like the summers of childhood”, and to try out variations analogous to those discussed above.
Ultimately, though, you may be right that the integrated relative vs. supplementary distinction is a red herring.
I am discombobulated!
It is quite metagrobolizing, Claude!
[…] earlier post on non-restrictive that gives an idea of how rare it is, and provides an ambiguous example from Penelope Fitzgerald; I […]
[…] 2011 I reported examples of the elusive non-restrictive that in prose by Penelope Fitzgerald, Anne Enright, and Robert Arthur, later adding lines from Peter […]