I have always felt that Gorgonzola has the emphasis askew in the above mention of dialectal differences in acceptability. It is the British-type dialects that allow both singular and plural agreement depending on how the group is perceived, and the American-type dialects that prefer to restrict agreement to the singular. This variability has been true in Britain for at least a century.

Moreover, it is a matter of formality. British written news sources overwhelmingly use the singular, but in speech the plural is more common. For me, the plural is the usual form.

Dialectal variability in acceptance

I recently acquired Corbett (2000) and can now back this up with studies quoted in that. In 1979 Stig Johansson tested about a hundred each native English speakers in Britain and the US, telling them that foreign learners had produced some sentences, and asking them to check if they needed any correction. In 1988 Laurie Bauer repeated the study in New Zealand.

The sentence testing this point was The audience were enjoying every minute of the show. The proportion who passed the sentence untouched were: Britain 77%, New Zealand 73%, United States 5%. Those who offered a correction to audience was were: Britain 15%, New Zealand 21%, United States 90%.

I can confirm from my own knowledge that Australian aligns as British-type, as expected, and from discussions on E2 that Canadian aligns with American usage.

In 1972 Graham Nixon studied British newspapers. In 100 000 words he found plural verb agreement in only 12% of cases and pronoun agreement in 27%.

In my own informal searches I have found it almost impossible to find plural agreement on the BBC news website or in my newspaper, The Independent. Not only does The Independent keep saying the government, council, or whatever is doing something, but they keep on referring to it as singular it.

However, listening to the BBC spoken bulletin this morning, I heard the Northern Ireland police were doing something, Scotland Yard have done something, though the government is under renewed pressure to do something. So the BBC write what they do not always say. And in my newspaper the only plural I could find for a long time was a speaker being quoted: 'Norwich Union are the first insurers to take such a measure...'. But in the sports pages, where colloquial forms are more common, all the teams become plural: Oxford still have, Argyle are only, Farnborough Town visit Darlington, Morecambe go to Ipswich, Chelsea have recalled him, Paris St-Germain are still waiting. I found two singulars there.


The term 'collective noun' is overworked. As well as covering the animal terms like pride, brace, murder, it can cover human groups like government, team, army, council and qualifiers like lot, number, majority, group and aggregations like fauna, verdure, drapery, verbiage. These need separate treatment. For the human groups, the point of most interest or at least most controversy here, the term noun of multitude has been used (Fowler & Fowler 1906), and one used by linguists is corporate noun.

The two possibilities of agreement are semantic agreement and formal agreement. The latter is also though less helpfully called syntactic agreement. Semantic agreement means you're thinking of the meaning: the audience are people, and it is they who are enjoying it, not some abstract set constituted out of them. Formal agreement means it is the form of the word 'audience' that decides: it lacks the plural '-s', so it's singular, so verbs agreeing with it and pronouns referring back to it are also made singular.

Historical evidence for variability

In the influential grammarians Fowler & Fowler we find variability was normal in standard British at the beginning of the twentieth century. There is no suggestion that it was then an innovation. In Fowler & Fowler (1906) they state the following, and Fowler (1926) makes a similar statement.
Though nouns of multitude may be freely used with either a singular or a plural verb, or be referred to by pronouns of singular or plural meaning, they should not have both (except for special reasons or upon deliberation) in the same sentence; and words that will rank in one context as nouns of multitude may be very awkward if so used in another.
In both books they then give example quotations of fluctuating number.
The public is naturally much impressed by this evidence, and in considering it do not make the necessary allowances.

The battleship Kniaz Potemkin, of which the crew is said to have mutinied and murdered their officers

Almost all their examples are like the above two in that they start with formal agreement next to the noun (public is; crew is) and move to semantic agreement later (public... do not; their officers). These sound natural to me, whereas the few that shift from plural back to singular sound very odd; you wonder how anyone can have written them. Recall also the figure for verbs and pronouns in the newspapers: 12% versus 27%. This reflects a general tendency in language that formal agreement drops off with distance.

Agreement and distance

This write-up will be far too long if I bring in examples from Cairene Arabic and Macedonian. But I'll mention a more familiar foreign example. It is a well-known quirk of German that 'the girl' is not feminine but neuter, das Mädchen, and that all her its articles and adjectives are neuter to agree with her it: ein gutes Mädchen. However, she doesn't trail this throughout her sentence. In Heine's poem Ein Jüngling liebt ein Mädchen, 'a boy loves a girl', the very next line begins die hat einen andern erwählt 'who has chosen another'; she takes a feminine relative pronoun die.

The noun phrase is the most tightly bound unit, and formal agreement is obligatory there. The same in English: we can only say this committee; that audience, never *these committee; *those audience. This may exert influence over the predicate: I am doubtful over whether I would say ?That committee are taking too long to decide, though The committee are is natural. (* before a phrase means it's unacceptable, ? means doubtful.)

As the sentence and its successors move away from the original word, we tend to forget the exact form of the word but still remember who we're talking about, so anaphoric pronouns tend to be they and their, and verbs further away are more likely to take semantic agreement (The public... do not make...). Consider these:

The audience was enjoying itself.
The audience were enjoying themselves.

The audience roared out its delight.
The audience roared out their delight.

The audience was enjoying every minute of the show. After the curtain call they were treated to an encore.
In the first pair, note that the formal choice at the beginning forces agreement on the later pronoun. You can't mix was... themselves or were... itself. In the second pair, the verb doesn't show agreement so the first choice is at a greater distance: if in your dialect audience were is unacceptable, you might nevertheless accept their at that remove. In the third pair, the switch to plural is in a different sentence; and this is very likely to be acceptable to everyone.

The treatment of relative clauses in English confirms that the choice is between a group of people and a singular thing. If you make the audience plural you have to use the animate relative who for them, and if you make it singular you use the inanimate which:

The audience, which was roaring with delight, ...
The audience, who were roaring with delight, ...
Fowler (1926) quotes a couple of sentences where this is violated, and again they sound bizarre: The population, who was driven away on the morrow... and The excuse of the Admiralty, which were responsible...

Where no choice exists

There are a couple of words which though formally singular are functionally plural: people and police even take plural agreement within their noun phrase: these people; those police. The word people is effectively the suppletive plural of person (though both persons and peoples also exist, with minor uses); and the word police doesn't seem to have a simple singular. More oddities: you can say Police are combing the area or The police are, so it's different from the choices in The army is and The army are. And The Northern Ireland police is to consider... is much more acceptable than *The police is to....

On the other hand there are those uses of collectives where only the singular is possible. This is where the collective (or corporate) noun specifically means the set constituted from its members, and not its members themselves. The party was founded in 1901; my family contains four people; the army is on a voluntary basis; the Cabinet has an office in Downing Street.

Agreement sometimes shows an animacy hierarchy. Of the different kinds of collective noun I mentioned above, the inanimate ones verdure, drapery, verbiage can never take plural agreement: nor can garden, forest, gravel, which could also be considered other kinds of collective. You might treat animals differently from humans: ?The pride are gathering at the waterhole.

The qualifiers that link with of, such as a majority of people, present their own grammatical problems and I don't want to discuss them here.

Perception and markedness

All readers of this will have their own intuitions about how acceptable or natural the examples are. In those cases where you can say both, your choice will probably depend on how you perceive the group, and one or the other will be more marked, that is less natural or available, or applicable in narrower circumstances.

Of the remaining type, the human multitudes that can take either number, one main criterion is, as Gorgonzola said above, whether the people are perceived as acting together or separately. Some collocations seem to work on this basis, at least for me: unity in The government is going to lose the next election; the audience was evacuated after the fire alarm; the jury has been empanelled and is considering its verdict; the regiment is stationed overseas: versus individuality in The government are unable to decide; the jury are allowed to smoke in the jury room; the audience were laughing.

I would also say The government is agreed; the government are divided, perceiving them as results. But Fowler (1926) makes the entirely logical suggestion that we should perceive them as processes and say The government are agreed, because you need plurality to agree, and The government is divided, because division needs a prior whole.

In many cases the unity or disunity of the group is not obviously assignable. If the committee are/is preparing a report, it will be issued by a unity but is being worked on by the members. This is where markedness comes in. For me, the plural is the unmarked form. I use plural where there is no particular reason to do otherwise. Being unmarked also means being more widely applicable. I'm quite happy with The audience were evacuated. I don't have to mark it for 'acting as a unity' by switching to the marked form is.

Most North American speakers will treat the singular as the unmarked form and need special reason to prefer plural, or will feel it as very colloquial or perhaps ungrammatical.

Two illustrations of markedness. Normally I'd say Italy are playing Poland, and I visualize people, the teams on the ground. If I said Italy is playing Poland I'd visualize a tournament fixture in a newspaper: the teams have been bleached of people and are just names. In The pride were moving towards their next kill I see them walking normally, perhaps to a herd of zebra; in The pride was moving towards its next kill I see special cohesion, once they're homing on an individual zebra.

Fowler, H.W. and G.F., 1906 (3rd ed. 1930), The King's English, OUP.
Fowler, H.W., 1926, Modern English Usage, OUP.
Corbett, Greville, 2000, Number, CUP.