One of them, plus two others… were or was?
« previous post | next post »
A tricky agreement situation arose for "Bagehot" while writing his eponymous column in The Economist last week. The topic was the "furious festival of blame in Britain recently". Among other scandals, two crude radio shock comics recently called a much-loved aging actor's answering machine and told him that one of them had fucked his granddaughter, and the call was recorded, and editorially approved, and actually broadcast. Bagehot wrote:
Two comedians make cruel jokes on BBC radio: heads must roll! (They did—one of the comedians, plus two executives, were forced out.)
My interest is in the agreement form chosen for the verb I have underlined. People like Stephen Fry in prescriptivist mood would say that the subject is the noun phrase one of the comedians, which is singular, so it should be was forced out. However, I'm not dinging Bagehot on the plural agreement form. I believe it's not just a simple binary decision in this case. Things are much more subtle and interesting.
I set aside, by the way, the fact that Stephen Fry claims none of them should take singular agreement. There is good evidence from usage that he is simply wrong there, contra prescriptivist dogma; but that's another story. Here we are concerned with noun phrases (NPs) like one of them, where things are very different, and considerably clearer; and also with subject NPs followed by parenthetical extra phrases like plus two executives.
Notice, if the sentence had been One comedian and two executives were forced out, then were would be uncontroversially fine, because coordinate subjects with and as the coordinator generally act as syntactically plural noun phrases. But Bagehot's sentence doesn't say that. It says one of the comedians was forced out. The preposition phrase (PP) plus two executives is dropped in as a parenthetical interruption, adding that two executives also were. (Yes, I think it is quite clear that plus is a preposition.) Parenthetical PPs dropped in after subjects do not affect agreement: in A young boy, with his mother, was sitting sadly in the corner of the doctor's waiting room, we get was rather than were.
Nor do partitive PPs like of the comedians, for the most part, affect agreement. True, there is more variation on this; but as a spot check, searching for and one of them are on .edu sites yields only about five genuine examples, some clearly written by people with a shaky command of English; while searching for and one of them is yields about 2,760 hits. It is normal for the plural NP in phrases like of them or of the comedians not to affect agreement properties. So one of the two comedians were forced out would normally be judged ungrammatical.
But as I said above, this does not mean Bagehot is to be dinged. It seems to me that he faced a genuine crux. A syntactic quandary. The distracting plural NP two executives intervenes, and is closer to the verb than the singular one; and there is another plural noun (comedians — irrelevant though it normally should be) intervening as well, upping the pressure to think plural. And to cap it all, in terms of the meaning expressed by the whole clause, notice that the total number of people said to have been forced out stands at 3.
I think Bagehot may well have been right to think that were would be the safest and least disruptive choice of agreement form to use here. Reasonable people could differ on the point, but one could say that what Bagehot wrote was thoroughly justifiable. To write one of the comedians, plus two executives, was forced out might very well be judged grammatically worse.
Now, I take it that for most Language Log readers, or for anyone with a certain amount of understanding of basic grammar, what I have just said will seem commonsensical enough. I hope so. But to theoretical linguists, I fear it may seem radical, even scary. I am suggesting that the right answer here is not fixed by the facts of anyone's internalized grammar. I'm implicitly taking the view that agreement can be a judgment call. It can be somewhat (though not in detail) like the esthetic judgment of whether this tie will go with that shirt, or the ethical judgment of whether the cost of the bureaucratic accounting involved would make it silly to take the dollar bill I found and turn it in to the police so I should keep it.
More specifically, I think there are situations in which the constraints set by the grammar of English leave things balanced in a state where nothing is fully satisfactory — there is no unimpeachable solution — so you just have to chance it, and go with one decision rather than another on a basis of… well, taste and educated discernment.
I'm suggesting that perhaps the psychogrammar in your head is not an automatically functioning module that strictly defines what's grammatical for you and what's not. It's a rather disorganized collection of conditions and prohibitions and desiderata, and it doesn't always provide a unique answer to every grammatical question. Sorry, but I think that might be the situation we face as linguists. And if so, then we can't make things otherwise by just stipulating or shouting — or by avoiding fine-detail description of subject-verb agreement altogether.
Sky Onosson said,
November 26, 2008 @ 11:18 am
I think your last paragraph is 100% correct. When I've taken theoretical syntax courses in the past, I've always had this gnawing feeling that something wasn't right – and it is exactly what you describe.
Levi Stahl said,
November 26, 2008 @ 11:19 am
This strikes me as the sort of sentence that, were I to find it in my own writing, I would rearrange: "one of the comedians was forced out, along with two executives." That solution allows the writer to avoid the difficulty while not, I think, altering the meaning.
jamessal said,
November 26, 2008 @ 11:33 am
I think this is right too, I'm just surprised to hear Professor Pullum saying it. I mean, isn't this all an argument for judging language in terms of its "acceptability" rather than its "grammaticality"? Or am I missing something?
William F Dowling said,
November 26, 2008 @ 12:03 pm
It's been a long time since I studied syntax, but I'm a little surprised that the conclusion here would be scary to theoretical linguists. I could understand it would be scary if you took the view that a language is a set of sentences. But I thought that view was pretty much history post James McCawley's statement: Defining a language as a set of sentences is like defining a car as a set of trips to the supermarket.
Faldone said,
November 26, 2008 @ 12:23 pm
I don't think your example, "A young boy, with his mother, was sitting sadly in the corner of the doctor's waiting room" is a valid comparison. It's not clear from the wording that the mother was sitting sadly. In fact, I would say that it is clear that she isn't sitting sadly. In the "Bagehot" sentence, the two executives clearly were forced out. The off-the-shelf argument that if it ain't got an 'and' it takes the singular doesn't apply here, in my opinion.
Jonathan Lundell said,
November 26, 2008 @ 1:05 pm
Like LS, my first inclination was to say "rewrite", and in practice I suppose it's what I'd do; contra Pullum, "so you just have to change it".
That rather misses the point, though: our grammar is genuinely fuzzy. Sure, we can find a non-fuzzy alternative, and in many cases that'll be the thing to do. But the subject sentence is still a sentence in English that somebody wrote and published.
A quick Ghit experiment yields:
"one and one is two" 43,900 (first hit: Paul McCartney)
"one and one are two": 9290 (first hit: Bertrand Russell)
"one plus one is two": 5450
"one plus one are two": 598
Curiously, "one plus one is four": 419
but "one plus one are four": 0
Johanne D said,
November 26, 2008 @ 1:15 pm
My native language is Québec French and I work as an En-Fr translator. Personally, I’m not a prescriptivist (except for reasons of national survival, which I won’t go into). But professionally, I need to be aware all of the little rules that can make French so frustrating.
So I read this blog with as much envy as pleasure. And lo, here is a rare example where French gives more leeway! In this particular case, the official rule is persnickety, although few people follow it to the letter: if the interruption is between commas, it doesn’t affect agreement, and if not, it does. (I know a lot of phone numbers by heart too.)
But for collective nouns, we have exactly what you describe as « go(ing) with one decision rather than another on a basis of… well, taste and educated discernment ». I think the most frequent phrase in « le bon vieux Grevisse » grammar manual is something like « the agreement depends on the point of view of the speaker/writer ».
Bill Scott said,
November 26, 2008 @ 2:24 pm
Re." The preposition phrase (PP) plus two executives is dropped in as a parenthetical interruption, adding that two executives also were. (Yes, I think it is quite clear that plus is a preposition.)"
When did plus become a preposition? It acts like a conjunction, IMHO.
Boris Blagojević said,
November 26, 2008 @ 3:30 pm
Well, if this uncertainty about number agreement isn't convincing enough, a recent huge post about coordinate possessives by Arnold Zwicky might be.
I don't even think it's hard to find cases which are confusing for native speakers. For example, a grammatical difference which is slowly falling out could create similar uncertainty.
Simon Cauchi said,
November 26, 2008 @ 3:55 pm
To answer Bill Scott's question. Here is what the Cambridge Grammar of the English Language has to say in Chapter 15, pp. 1318-1319 (but I'm not sure if Geoff Pullum, who wasn't one of the authors of this chapter, would entirely agree with it):
"_Plus_
This is another item that straddles the boundary between prepositions and coordination.
[Examples omitted; I'll interpolate some of them later]
_Plus_ is predominantly followed by an NP, making it more like a preposition than a coordinator. Examples like [v] ["_He spoke with a funny accent, plus he wore socks with his sandals_"], where it introduces a main clause, are restricted to informal style . . . It differs from prototypical prepositions in that it does not permit fronting (*_Plus other control refinements the cost-billing system has reduced the deficit_) and only very rarely occurs as head of a predicative complement (_The electrical charge is plus, of course, the initial pulse of current_). "_X plus Y_" tends to count as singular for agreement purposes, as in [ii] ["_The cost-billing system plus other control refinements has reduced the deficit_"], but it is sometimes taken as a plural, as in [iii] ["_His stamina plus his experience make him unbeatable_"] — in the singular case it is being treated as a preposition, in the plural case as a coordinator. . . ."
Emily Lilly said,
November 26, 2008 @ 4:14 pm
Great article!
It sounds to me like you have just given a very effective argument in favor of seeking out an optimality-theory paradigm for syntax!
Stephen Jones said,
November 26, 2008 @ 4:47 pm
I'd have thought this was glaringly obvious. Apart from anything else if it weren't true how would you explain language change.
Stephen Jones said,
November 26, 2008 @ 4:56 pm
But the difficulty you are referring to is purely imaginary. There are types of noun-verb agreement in English; 'grammatical' or 'formal agreement', notional agreement and proximal agreement. In an ideal world all three would concur, but there are numerous examples, such as this one, where this is not the case.
Here notional and proximal agreement concur and we have a plural verb. If grammatical agreement and proximal agreement were the same, as in,
The presenter, plus his sidekick, was forced out. then we would have a singular verb. Take away the commas, or the equivalent pause in spoken English, and then you could reasonably say that 'plus' is acting in the same way as 'and', and that notional and grammatical agreement would demand a plural verb.
The producer plus his sidekick were forced out.
Nathan Myers said,
November 26, 2008 @ 5:26 pm
I won't argue with "the right answer here is not fixed by the facts of anyone's internalized grammar". The final paragraph seems trivially correct. Theoretical linguists need all the discomfiture that can be mustered.
However, the example is not even a little bit ambiguous to any regular speaker: the number agreement isn't syntactic at all, it's semantic. Three people were sacked, so we must say "were" to encompass them. Try to find one non-linguist (and non-grammarian prescriptivist; don't worry, there are lots) to agree that "was" could even be in the running.
Is it controversial, among linguists, to assert that people don't strictly segregate semantics from syntax? In nature, levels are a matter of organizational convenience, not law. Quantum physical effects leak out to the macroscopic world continually; your computer's operation depends on this fact. Language is a natural behavior of human animals. Semantics affects syntax. Sorry.
Sridhar Ramesh said,
November 26, 2008 @ 5:38 pm
I'm a non-linguist; I think "was" could be in the running. Do I get a prize?
Semantics is playing some role here, as even Pullum acknowledged, but not exclusively or even primarily; if it were purely a matter of semantics over syntax, then would not the situation be the same with "In addition to two executives, one of the comedians was forced out"? But "were" instead of "was" is flatly ungrammatical here.
Sridhar Ramesh said,
November 26, 2008 @ 5:39 pm
Er, because I used "here" twice with different referents, just to clarify, in my last sentence, "here" refers to the example I gave, not what Bagehot actually wrote.
Nathan Myers said,
November 26, 2008 @ 5:59 pm
Sridhar: It would be even more idiotic to claim that syntax has no role than to claim that semantics has no role. Therefore, I do not make such a claim. The point is that levels mix, inconveniently for theoretical linguists, but that's their problem. It is not the responsibility of native speakers to make things easier for theoretical linguists.
In the case of "In addition to…", we must already know, for the sentence to make sense, that the two executives had been sacked. Semantics again.
BTW, I think I meant "non-grammarian-prescriptivist" above, but that's not so great either. Maybe it is better expressed as "Mrs. Grundy".
dr pepper said,
November 26, 2008 @ 6:01 pm
Very well, my grammar is inconsistent. My linguistic faculty is immense, it contains a multitude of rules.
Aleksei Nazarov said,
November 26, 2008 @ 6:11 pm
It is certainly true that there are "gaps" in agreement systems, where there is in fact no "right answer" in certain cases. The most interesting part is that in cases like the one presented here, it is still acceptable to use the sentences in question, whereas many grammatical impossibilities are impossible to use in a serious context.
But does all this really suggest that (psycho)grammar is a "disorganized collection of conditions and prohibitions and desiderata"?
This is an open question, I am simply curious. Of course the following is biased rhetorics, but Chomsky once said that anything looks like a disorganized mess if you don't understand it. I wonder if it could turn out that in the end, gaps in agreement and repair strategies for them are the result of a unified grammar combined with an idiosyncratic lexicon that sometimes provides the strangest things to work with, or something along those lines.
Chris said,
November 26, 2008 @ 6:14 pm
If "plus two executives" were the same kind of prepositional phrase as "with his mother," shouldn't we be able to move it to the end of the clause?
The sentence, "A young boy, with his mother, was sitting sadly in the corner of the doctor's waiting room" sounds better to my ear as "A young boy was sitting sadly with his mother in the corner of the waiting room."
"One of the comedians was forced out plus two executives" sounds like nonsense, however. I think "plus two executives" has to be next to "one of the comedians" because it's a part of the noun phrase, albeit a parenthetical one.
On balance, I don't think it's even a prepositional phrase at all. Compare to "one of the comedians was forced out along with two executives" or, awkwardly, "one of the comedians along with two executives was(?) forced out."
Ivan said,
November 26, 2008 @ 6:43 pm
Geoffrey K. Pullum:
For me, linguistics is just a hobby interest, so it might be a bit presumptuous that I feel entitled to comment on this, but still, I'd say that the above conclusion is plainly obvious for speakers of more synthetic languages with very irregular inflectional morphology. As a native speaker of Croatian, I've always been struck by how rarely the situation you describe happens in English relative to my native language. Maybe it's because these things happen most often with inflection and agreement rules, of which English has very little in the first place?
To take my native language as an example, like other Slavic languages, Croatian is full of irregular inflections and maddeningly complicated and illogical rules on number and gender agreement. Furthermore, there's a lot of dialectal variation in grammar, and nearly everyone's grasp of the standard language is significantly influenced by the local peculiarities, far more so than in English, so you get to hear conflicting grammar rules while growing up. In addition, some areas of grammar are passing through rapid changes, with old rules on their way out, but not quite forgotten, and no new ones firmly established to replace them. As a result, there are many situations where native speakers will get totally confused over what's grammatical, and not in the sense of being confused over the correct prescriptivist rule, but having an honest mental blockade.
One good example are the verb conjugations in archaic past tenses (aorist and especially imperfect). People generally still recognize them and occasionally use them for some verbs, but they'll often disagree or be confused when asked to conjugate them. Another similar case are the declensions of numbers followed by nouns; sometimes all options you can think of sound disturbingly "halfway correct". And I emphasize again that I'm not talking about being confused over what could be prescriptively correct, but about really having your natural intuition stalled.
Simon Musgrave said,
November 26, 2008 @ 6:48 pm
@Chris:
"One of the comedians was forced out plus two executives" sounds pretty good to me, even without any intonation break before 'plus'. But I don't see it as strong evidence that 'plus two executives' is a PP in that sentence – I think that it is also possible to see that structure as two conjoined clauses with ellipsis in the second one (an analysis which is of course not possible for the original example).
blahedo said,
November 26, 2008 @ 9:34 pm
"the right answer here is not fixed by the facts of anyone's internalized grammar"
An alternative conclusion would be that the right answer is that neither option, *no* option, is correct. Some have hinted at this, suggesting "just rephrase it" sorts of strategies, but I still detect an air of "one of them must be right, we just don't know which" in a lot of their posts. Or maybe "we just *can't* know which", or some variation. But what would be problematic with an analysis that said that neither the singular form nor the plural form agreed?
Tim Silverman said,
November 27, 2008 @ 7:48 am
@Blahedo: the problem is that people actually produce and understand these forms, often with no awareness of any problem. In these case, both singular and plural agreement sound correct to me; they just have slightly different nuances (whether the comedian is foregrounded relative to the executives—singular agreement—or they are grouped together on the same level—plural agreement.
Compare the agreement of institutional nouns like bank or army in British English, which can take either singular or plural agreement depending on whether one is treating the institution as acting as a unified entity, or its members as acting as individuals. Likewise the "Fluid-S" system for intransitive verbs in Dixon's treatment of accusative/ergative phenomena.
On a separate note—while I appreciate Arnold Zwicky is bothered by topic wander in comments threads, surely he can open comments to receive readers' congratulations on Language Log's recent award from the LSA! Well, I guess not … so, congratulations awkwardly inserted here …
Aaron Davies said,
November 27, 2008 @ 10:36 am
"one and one is two" 43,900 (first hit: Paul McCartney)
whereas Lennon had "one and one and one is three" (~2000:1 vs "are" on google, fwiw)
Aleksei Nazarov said,
November 27, 2008 @ 1:09 pm
@Ivan
The situation that you describe is very interesting.
I think that having different paradigms and agreement models doesn't mean that there are different forces in play, it's more about extensive idiosyncracy, which is a related but different topic (but I might be wrong, of course).
But the situation that people are regularly confused about what is grammatical is really interesting. This probably must have had something to do with the historical developments in recent years (rapid language change because of social instability), but that doesn't explain anything about the situation nowadays.
Christian Campbell said,
November 27, 2008 @ 3:58 pm
@Jonathan Lundell, I think it should be pointed out that the people writing "one and one is two" are using 'and' prepositionally, exactly as they would 'plus.'
OTOH, I'm stumped by the "one plus one are two" people using 'plus' apparently as a coordinating conjuction.
kyle gorman said,
November 27, 2008 @ 6:01 pm
one doesn't need to accept the chaotic nature of syntactic knowledge to get this kind of thing (it looks a little bit like paradigmatic gaps). what i'd suggest is that simply there is not enough reliable evidence to allow a learner to converge on one or the other solution during acquisition. the idea that learning for all parts of say a parameter-space of syntax may not always converge IRL has been around for a while. i think the most compelling evidence comes from:
C.-h. Han, J. Lidz, and J. Musolino. Verb-raising and grammar competition in Korean: Evidence from negation and quantifier scope. Linguistic Inquiry, 38(1):1–47, 2007.
Han et al. find that Korean speakers just don't know whether they have quantifier raising or not, because there's no way for them to tell. So they just choose one option.
Skullturf Q. Beavispants said,
November 27, 2008 @ 6:16 pm
Nathan Myers at 5:26 pm November 26th makes an interesting point, about word choice sometimes being dictated by semantics rather than syntax.
Whether or not the sentence in the original post is the best example of this (intuitions seem to be divided there), perhaps there are others that can illustrate this tendency.
Example: "My family is getting together in Buffalo for Thanksgiving; I am cooking dinner for it."
As far as I know, nobody would say the above. (Though I could be wrong — there are enough varieties of English that nothing surprises me anymore.)
There are regional divides and individual divides as to the acceptability of "my family is" vs. "my family are", but as far as I'm aware, no native speaker would ever refer to their family as an "it" in the latter part of that sentence.
No matter how much the word "family" has the form of a plain ol' singular noun not ending in S, the physical reality that your family is a bunch of people, and not an amorphous blob, wins out in the end. Because of physical reality and not because of any underlying pattern in the language, your family becomes a "them".
blahedo said,
November 28, 2008 @ 3:58 am
@Tim Silverman "the problem is that people actually produce and understand these forms, often with no awareness of any problem":
The fact that people produce them is not inherently evidence (since people produce errors), nor is the fact that people understand them (since there are plenty of sensical but ungrammatical utterances). Which leaves us with awareness of the problem; you say you find either to be okay, but both sound a bit wrong to me (as well as, apparently, many other commenters). Anyway, I'm not trying to make the stronger claim that both forms are wrong here, just that *if* neither form sounds quite right, then "both forms are wrong" is one potentially valid conclusion that shouldn't be ruled out.
Jonathan Lundell said,
November 28, 2008 @ 9:28 pm
@Christian Campbell, it seems to me that there's an alternate reading, along the lines of: [the expression] "one and one" is [equal/identical/equivalent to the expression] "two".
Jonathan said,
November 29, 2008 @ 5:21 am
@Skullturf:
And when the family *is* considered as a unit, we can complete the semantic/syntactic dissociation:
One of the great problems of family therapy is the cost, in every sense, of bringing the whole family together and working with it. (ahpweb.org)
Jan Schreuder said,
November 29, 2008 @ 11:19 am
"But to theoretical linguists, I fear it may seem radical, even scary."
But maybe not to theoretical linguists of the Cognitive persuasion, especially those of the Constructive Grammar denomination. What surprises me most is that it took so long for the rest to realize that a grammar resembles a partially organized junk yard more than a smoothly purring electromagnetic motor. In the late sixties I studied linguistics at the University of Amsterdam (I later switched to medicine), and the department was frequently visited by a strange fellow with the beautiful name of Verloren van Themaat who would ask us to judge the grammaticality of sentences he would produce. Actually, it was a computer program that would produce these sentences. He was a talented programmer and an amateur linguist and Chomsky's notion of a grammar as a machine that produces the sentences and only the sentences of a particular language held him in its vise. The sentences he asked our native speaker judgments about, were of course of the marginal variety and would lead to many discussions and usually a split decision about their grammaticality. And often I would change my own idea about whether a certain sentence was "Dutch' or not several times. Slowly it dawned on me that my internal grammar was a less than perfect machine. A couple of months later I enrolled in medical school.
Phoevos Panagiotidis said,
November 29, 2008 @ 1:19 pm
Some similar phenomena led Annabel Cormack and Neil Smith to similar ideas on the status and the (in)determinacy of mental grammars. These are digested in their 2002 paper 'Parametric Poverty'. Glot International 6 (9/10): 285–287.
Daniel Ehrenberg said,
November 30, 2008 @ 10:18 am
I'm not sure if I understand the main argument. It seems like Pullum is arguing that the existence of ambiguity requires that we abandon the idea that grammaticality is on some level unambiguous. Why wouldn't it be consistent if, as Simon Cauchi quotes the Cambridge Grammar of the English Language, plus can function as either a preposition or a coordinator? You could argue that both interpretations are always present, so either way sounds imperfect. It seems to me that this isn't more of a problem for understanding and determining grammaticality than any standard example from a Chomskyan syntax textbook.
Chris said,
December 1, 2008 @ 10:38 am
I agree with Levi: the best solution is to rewrite. (An alternative rewriting would be to make the subject straightforwardly coordinate: "One of the comedians and two executives were forced out.")
It seems to me that this sort of halfheartedly compound subject often causes problems and therefore should be avoided (your young boy example isn't much better IMO; it's not absolutely wrong, but unnecessarily awkward). Either put the second subject in clear coordination with the first one ("A young boy and his mother were sitting…"), or banish it to the far end of the sentence so that its exclusion is clear ("A young boy was sitting… along with his mother."). One of those rewritings implies that the mother was sad and the other doesn't; the original example can be interpreted either way.
codeboot said,
April 11, 2009 @ 3:38 pm
I think this is right too, I'm just surprised to hear Professor Pullum saying it.