"Often more [difficulty] than in this chosen pair"

« previous post | next post »

We've often complained about the ignorant aftermath of E.B. White's ignorant 1959 incitement to which-hunting, which launched the idea that restrictive (or integrated, or defining) relative clauses in English should always and only be introduced by that, while non-restrictive (or supplementary, or non-defining) relative clauses should be introduced by which. (See "Reddit blewit" 12/24/2012 for details and additional links. Note that for simplicity, I'm considering only relative clauses with inanimate/nonhuman heads, though the fundamental point remains the same when we add who to the mix.)

My point today is that the whole distinction is a false one.

More exactly: The traditional restrictive/non-restrictive dichotomy merges distinct morphological, syntactic, semantic, prosodic, rhetorical, and psychological questions; the correlation among these different dimensions is loose at best; several of the relevant distinctions are gradient rather than categorical; and some of the distinctions are sometimes a matter of pragmatic vagueness rather than grammatical ambiguity.

If I'm right, then modern linguists have been committing White's sin in a less extreme form, trying to impose an over-simplified rationalist taxonomy on a more complex linguistic reality.

Let's start with something that H.W. Fowler wrote in 1926, following his famous suggestion that "if writers would agree to regard that as the defining relative pronoun, & which as the non-defining, there would be much gain both in lucidity & in ease". Fowler notes that "it would be idle to pretend that [this] is practice either of most or of the best writers" — but what interests me is his parenthetical admission that the distinction itself is often rather unclear.  Here's the relevant passage from the first edition of  "A Dictionary of Modern English Usage" [emphasis added]:

A defining relative clause is one that identifies the person or thing meant by limiting the denotation of the antecedent: Each made a list of books that had influenced him ; not books generally, but books as defined by the that-clause. Contrast with that: I always buy his books, which have influenced me greatly; the clause does not limit his books, which needs no limitation ; it gives a reason (= for they have), or adds a new fact (= & they have). There is no great difficulty, though often more than in this chosen pair, about deciding whether a relative clause is defining or not; […]

(A longer excerpt is given at the end of this post.)

I recently began looking into the prosody of relative clauses. And my first discovery was that in fact, I often experience a great deal of difficulty in making the restrictive/non-restrictive distinction. In particular, apparently restrictive (or integrated or defining) cases with that are often easily re-interpreted as non-restrictive (or supplementary or non-defining). And this matters to my investigation of the prosody, because such clauses are often set off by significant prosodic phrase breaks.

For example, when I looked at the home page of the New York Times a few days ago, the very first relative clause I saw was the sub-hed from a editorial about the Benghazi hearing:

The hearing was a counterproductive exercise that accomplished nothing.

A restrictive relative clause, right? Certainly no competent editor would let you punctuate it

*The hearing was a counterproductive exercise, that accomplished nothing.

But when I read the phrase out loud, I naturally (though not obligatorily) put a fairly major phrase break between "exercise" and "that accomplished". And surely the writers weren't taking denotational care to exclude those counterproductive exercises that have accomplished things? (There seems to be a broad consensus that the hearing actually accomplished a great deal for Hillary Clinton's presidential campaign, but that would be a matter of mere fact, not grammar…)

Over the years, we've discussed a few similar cases, forced on our attention by writers' choice of punctuation. Thus in  "An ivory-billed relative clause", 12/1/2005, Geoff Pullum discussed the phrase

"The key point, that all the popular reports missed, is that FOXP2 is a transcription factor…"

The punctuation suggested a supplementary relative clause with that  (a construction Geoff thought might be extinct), but to be sure, Geoff checked:

I contacted the author of the sentence, Alec MacAndrew, and asked him what sense he had in mind when he wrote the article in question. He has kindly confirmed that I correctly divined his intention: he did indeed intend the supplementary semantics, where the meaning is "The key point — and incidentally, all the popular reports missed it — is that FOXP2 is a transcription factor".

And in "Non-restrictive 'that'", 3/22/2012, I cited a similar case from a webcomic. Some commenters complained that this is a comic we're talking about, and a proper editor would have etc.

But when I look at speech rather than text, and prosody rather than punctuation, those "ivory-billed relative clauses" are everywhere. Picking some audio from my hard drive, literally at random, I chose a 9/29/2014 Fresh Air interview with Lena Dunham. Here's the second that-relative in the interview, where Ms. Dunham is reading from her book:

I dress in neon spandex that hugs in all the wrong places.

Another restrictive relative, right? But there's a substantial phrase break between spandex and that hugs. And does the relative clause that hugs in all the wrong places really "[identify] the person or thing meant by limiting the denotation of the antecedent"? Or does it merely "add a new fact"?  Both interpretations seem plausible. Does a speaker really need to make this choice before constructing the sentence?

That's all I have time for this morning. But the more I look into this question, the more questionable the traditional dichotomy looks to me. And the questioning has broader implications, because there's a list of other cases that have traditionally between treated as exhibiting an analogous or even identical distinction: appositives (e.g. "my sister Sarah" vs. "my sister, Sarah"), post-modifying prepositional phrases (whether reduced relatives or arguments or adjuncts of the head noun); pre-nominal adjectives and other modifiers, and so forth.

I don't expect this brief exposition to persuade you. But I thought I'd get the idea out there.


As promised, here's a longer portion of the relevant passage from the first edition of H.W. Fowler's "A Dictionary of Modern English Usage":

A defining relative clause is one that identifies the person or thing meant by limiting the denotation of the antecedent: Each made a list of books that had influenced him ; not books generally, but books as defined by the that-clause. Contrast with that: I always buy his books, which have influenced me greatly; the clause does not limit his books, which needs no limitation ; it gives a reason (= for they have), or adds a new fact (= & they have). There is no great difficulty, though often more than in this chosen pair, about deciding whether a relative clause is defining or not; & the practice of using that if it is, & which if it is not, would also be easy but for certain peculiarities of that. The most important of these is its insistence on being the first word of its clause; it cannot, like whom & which, endure that a preposition governing it should, by coming before it, part it from the antecedent or the main sentence; such a preposition has to go, instead, at the end of the clause; that is quite in harmony with the closer connexion between a defining, (or that-) clause & the antecedent than between a non-defining (or which-) clause & the antecedent; but it forces the writer to choose between ending his sentence or clause with a preposition, & giving up that for which. In the article PREPOSITION AT END it is explained that to shrink with horror from ending with a preposition is no more than foolish superstition; but there are often particular reasons for not choosing that alternative, & then the other must be taken, & the fact accepted that the preposition-governed case of that is borrowed from which, & its possessive from who; its cases are then: subj. that; obj. that; poss. whose; prep.-preceded (in, by, from , for, &c.) which. Another peculiarity of that is that in the defining clauses to which it is proper it may, if it is not the subject, be omitted & yet operative (The man you saw means the same as The man that you saw), while which in the non-defining clauses to which it is proper must be expressed (This fact, which you admit, condemns you cannot be changed without altering the sense to This fact, you admit, condemns you).

Note that his choice of phrasing — "in harmony with the closer connexion between a defining, (or that-) clause & the antecedent than between a non-defining (or which-) clause & the antecedent" — might be taken to hint at a gradient aspect of this family of distinctions.



21 Comments

  1. BenHemmens said,

    October 25, 2015 @ 8:04 am

    Well, since I've never felt I undestood that distinction myself, I'll appreciate anything that allows me to believe it wasn't just because I'm stupid ;–)

  2. Jerry Friedman said,

    October 25, 2015 @ 11:20 am

    "That" is used non-restrictively after a pause more often than some think. For instance, COCA has

    "A restricted zone was painted around this car, that was legally parked."

    "So, if O.J. Simpson doesn't get on the stand and say, 'Look, I could never have done this to my former wife, that was the mother of my children and this other young man.' "

    I feel that such uses are more common in British English than in American English, but they're very hard to search for, and I don't have any data. The journalist Louise B. Gray, who writes on the environment, is fond of them, as here at her blog:

    "The springer spaniel will bark if it is a grey squirrel, that needs to be dispatched immediately."

    (She's describing a project to kill eastern gray squirrels, imported from America, because they're wiping out Britain's native red squirrels.)

    At alt.usage.english, it's mostly people who aren't Americans who say the distinction between restrictive and non-restrictive doesn't matter to their word and punctuation choices and they don't notice it in what they read or hear.

    As for the ones that are difficult to distinguish, I think a lot of them come after "a", as your example did, or other indefinite words. (I'm not sure how to say what I mean—the heads of the relative clauses are indefinite?)

    Along with

    The hearing was a counterproductive exercise that accomplished nothing.

    one could write

    The hearing was a counterproductive exercise, which accomplished nothing.

    (I don't think I'd say the "that" version with a pause.)

    However, I think that after "the" the distinction is often quite clear.

    I'm looking forward to more about your research. Are you planning to look into the question of how many readers and listeners get any information from these distinctions?

  3. Jerry Friedman said,

    October 25, 2015 @ 11:29 am

    If anyone wants fourteen examples of Gray's non-restrictive "that"s in edited text, plus my peeve and that of the guy who found them, go to this page and search for "Zotero".

  4. Brett said,

    October 25, 2015 @ 11:42 am

    I've particularly noticed this lack of a firm distinction in appositives. I cannot divine any systematic difference between "my sister Sarah" and "my sister, Sarah" in terms of meaning or real-world pronunciation.

    [(myl) It's certainly possible to separate an appositive prosodically to different degrees. But even a prosodically very parenthetical appositive sometimes "identifies the person or thing meant by limiting the denotation of the [head noun phrase]" — e.g.

    my friend — you know, the one that lives in Oakland — told me the other day that …

    Of course, what seems to be the same degree of prosodic separation might (maybe more often?) be associated with an appositive that "gives a reason or adds a new fact" or other commentary on the side.]

  5. Coby Lubliner said,

    October 25, 2015 @ 12:16 pm

    I first learned of the distinction in freshman English in 1953-54, that is, several years before Strunk & White. At the time English was in the process of becoming my primary language (replacing German) and the distinction became so natural to me that I have had to make a conscious effort to overcome it and sometimes force myself to use a defining which. Other languages I know don't make the distinction formally. Spanish makes it prosodically, that is, with pauses reflected in writing by commas or dashes. French may add something like d'ailleurs to a non-defining clause. And in written German all relative clauses require a comma before them.

  6. Coby Lubliner said,

    October 25, 2015 @ 12:20 pm

    And, incidentally, there is the oft-stated fact that relative clauses introduced by who, where, of which etc. function perfectly well as either defining or non-defining.

  7. Jerry Friedman said,

    October 25, 2015 @ 12:53 pm

    Coby Lubliner: Why did you sometimes force yourself to use a defining which?

  8. Matt Keefe said,

    October 25, 2015 @ 1:32 pm

    I'm an editor, and I've definitely noticed that this distinction really is not clear cut – not only when attempting to identify the intended interpretation and to punctuate accordingly, but perhaps more tellingly I've noticed it when trying to explain the difference to writers to help them improve and more thoroughly check their own work. Indeed, there have been times when consulting with a writer on what their intended interpretation might be has not only talked me around from my own initial interpretation, but also made me question what is permissible/grammatical. On the basis of a writing explaining their intended meaning and reading their sentence aloud to me, I've permitted cases of restrictive that which my own initial reading wanted to exclude; likewise, I've encountered instances that really made me question how clear cut the distinction between appositives really is. Maybe I should start keeping a record.

  9. Chris Cooper said,

    October 25, 2015 @ 3:21 pm

    @Jerry Friedman @11:20

    > I feel that such uses are more common in British English than in American English

    That's very interesting. As a Brit, I've always felt they were characteristically American, and I've hated them.

  10. Doug said,

    October 25, 2015 @ 4:03 pm

    I agree with Brett. In my experience, when people say "my sister Sarah," they do not generally adjust their pauses to indicate whether they have more than one sister. I had this thought way back in English class when I was about twelve (but I kept it to myself rather than argue with a teacher.)

    On a related, note, I've read that some languages allow only defining/restrictive relative clauses. I have to wonder how thoroughly the difficult cases were tested.

  11. Jerry Friedman said,

    October 25, 2015 @ 4:15 pm

    Chris Cooper:

    These are hard to search for. Picking a verb arbitrarily, I found one example of nonrestrictive "that brought" out of the 86 at COCA (not counting translations of the beginning of the Iliad), and one out of the twelve at the BNC. so clearly… :-)

    American: "In point of fact, we're going back to the very pressure tactic, that does not involve troops, that brought about agreement in the first place."

    British: "The celebration of the Eucharist itself was a very enjoyable and moving occasion, that brought people of all ages together."

    The search was for ", that.[cs*] [v*]". you might be able to do better, but I don't understand the results I'm getting, and it's time to do something else.

  12. Coby Lubliner said,

    October 25, 2015 @ 9:37 pm

    Jerry Friedman: because sometimes it just sounds better, e.g. "a day which will live in infamy."

  13. JS said,

    October 26, 2015 @ 12:37 am

    The terms "restrictive" and "non-restrictive" should at least still be fully meaningful when there's a minimal contrast — like the one Pullum was trying to sort out, or

    I found some reluctance from white people who don't even want to hear about black history. (link)
    I found some reluctance from white people, who don't even want to hear about black history.

    But the suggestion of a correlation with that vs. which, etc., was a bad idea…

    I don't know to what extent you could get at a distinction between "real" NPs of the form [N + relative clause/N/PP…] (e.g., "white people who don't want to…") and stuff you could just call parenthetical information ("white people—who don't want to…") by pointing to structural restrictions on the former, or to differing capacities to fill certain syntactic roles.

  14. Alan R said,

    October 26, 2015 @ 5:16 am

    This is something that's bothered me for a long time. I notice non-defining 'that' quite frequently in speech. If I pause later and try to recall the structure, I can normally only recreate something that sounds wrong. So it definitely seems to be the case to me that some uses of 'that' in non-defining clauses are unacceptable. So the question is whether there is some kind of sub category here, or is just that sometimes using 'which' seems a bit 'heavy'? That's my feeling but I can't explain it in a more scientific way.

    I've always found that the defining / non-defining distinction is easiest to explain/understand when you have the definite article (e.g. That's the man that I spoke to). When there's no article, or the indefinite article, things are harder. After all, isn't there something contradictory about having a 'defining' clause about something indefinite? For example, what's the difference between:

    They're looking for an employee who will manage the accounts.
    They're looking for an employee, who will will manage the accounts.

    It seems to me to be far less clear cut than the difference between these two:

    The employee who I spoke to yesterday is ill today.
    The employee, who I spoke to yesterday, is ill today.

  15. Jerry Friedman said,

    October 26, 2015 @ 10:35 am

    Coby Lubliner: Thanks. Roosevelt's famous line seems to be a matter of taste—you can easily find it misquoted with "that".

  16. Xmun said,

    October 26, 2015 @ 2:42 pm

    The employee who I spoke to yesterday is ill today.
    The employee, who I spoke to yesterday, is ill today.

    Alas, poor whom. I knew him well . . .

  17. Geoff said,

    October 26, 2015 @ 4:09 pm

    My day job is transcribing proceedings of parliament, so I pay close attention to the spoken word. I hear unambiguous non-restrictive 'that' quite often. Where the speaker's nonrestrictive meaning is clear from the prosody, but might be misunderstood in reading, I tend to edit to 'which'.

  18. Kaleberg said,

    October 26, 2015 @ 9:31 pm

    Back in the 1980s a friend of mine, an American, was hired by a software firm in England. His new employer offered him an employment contract to be executed in England and advised him to consult a solicitor. Luckily his solicitor had worked with Americans before and explained that in legal documents the meaning of "that" and "which" are inverted in the US and England. A clause that would be restrictive in one country would be non-restrictive in the other and vice versa. I do not remember which way the usage worked where, though I do remember my friend was quite happy with the contract once he understood the semantics within the relevant lexical scoping. A lawyer would have remarked on the language of jurisprudence.

  19. Jonathon Owen said,

    October 26, 2015 @ 11:39 pm

    I've been wondering about these things for a while too—both relative clauses and appositives. When editors decide whether it should be which or that or whether it needs to be set off by commas or not, they usually just consider whether it's limiting the denotation, as Fowler says. And that has felt wrong to me for a long time, though I've never really articulated why.

    But when I say "My wife Ruth", I'm not saying "My wife (whose name, by the way, is Ruth)". The latter seems to imply that her name is old information or is irrelevant to what I'm saying, when the usually the opposite is true—I'm introducing her as my wife and telling you her name.

    And I can't tell you how many hours I've seen wasted when an editor has tried to fact-check whether a person had more than one wife or more than one son so they can determine whether to put in those stupid commas. I wonder how many readers would even make the proper inference that "my son, Josh," was a person's only son while "my son Josh" must have had brothers.

    These rules waste a lot of people's time. I really don't think they help to clarify things for the reader, and occasionally they create problems when some well-meaning editor makes an ignorant assumption about whether a relative clause or appositive is restrictive or not.

  20. Suzanne Kemmer said,

    October 27, 2015 @ 11:00 pm

    I teach a Structure of English course, using A Student's Introduction to English Grammar by Huddleston and Pullum. It is a fine book, but when explorations with very large corpora are used with it it becomes obvious that many of the phenomena show gradience, and/or unclear boundary between grammatical vs. pragmatic distinctions (or rather, any boundary you choose to draw is an arbitrary one depending on your purposes.) A lot of 20th century linguistics has been a grand attempt "to impose an over-simplified rationalist taxonomy on a more complex linguistic reality." I don't think this invalidates linguistics, it just means we need theories that incorporate gradience and a better understanding of the relation of 'grammar' and 'pragmatics'. Which is what theories like Cognitive Grammar and other usage-based theories are aiming for.

  21. Victoria Simmons said,

    October 29, 2015 @ 3:54 am

    I correct students who write "Shakespeare's play, *Romeo and Juliet*, is a tragedy" to "Shakespeare's play *Romeo and Juliet* is a tragedy" because the commas make it seem Shakespeare wrote only one play.

RSS feed for comments on this post