A large percentage is or are [duplicate]

I found this sentence in a book:

A large percentage of the population smokes too much.

The author treated the phrase 'a large percentage of' as singular. I am so confused with the use of verbs after 'a large percentage of' as some folks on the Internet consider 'a large percentage of' plural.

Do we have any rules for the phrase? Thanks!

Solution 1:

The construction a large percentage of… is not plural, but is rather an example of a number-transparent construction, where the number agreement is determined by the oblique, i.e. the noun phrase that is the complement of of.

In your case there is another complication, namely that the oblique in question, the population, has a head noun (population) that is a collective noun. Such nouns are formally singular, but they allow for optional plural overrides (as in the police are here instead of the police is here). These overrides are more common in British English than in American English, but they do happen in American English as well.

In short, the number agreement of the whole phrase a large percentage of the population will depend on whether population is taken on its face value as singular, or whether we choose to employ a plural override for it. This is why, in this case, you see that some people use a singular verb, and some plural.

Discussion

Most of the time in English,

the verb agrees with a subject with the form of an NP whose person-number classification derives from its head noun.

That type of agreement is called simple agreement (CGEL, p. 499). In your sentence,

A large percentage of the population smokes too much.

we seem to have a simple agreement. Namely, the subject of the sentence is the noun phrase (NP) a large percentage of the population. The head of that NP is the noun percentage, which is a singular noun. Thus the verb smokes, which is also in the singular, would seem to agree with that singular noun.

But appearances can be deceiving, as evident from the following examples, all from the Corpus of Contemporary American English (COCA):

For example, he notes that a large percentage of the -ump rimes have a three-dimensional sense to them.
A large percentage of the new immigrants come as educated professionals…
…a large percentage of the entering students are not true freshmen…

In these examples, the simple agreement fails: all the verbs are plural, despite the fact that percentage, which is the head noun of the subject NP, is singular. It seems that the fact that percentage is a singular noun does not matter after all.

Singular and plural overrides

What is going on here is the following. In English, there can be departures from simple agreement. When the departure is systematic enough so that one can describe it in terms of certain rules, we say that the expression in question exhibits a singular or plural override (CGEL, p. 501). There are different types of overrides, which are sometimes obligatory and sometimes optional.

One of the nouns with which overrides can happen is percent. In your sentence you have percentage, which behaves similarly to percent as far as overrides.

In order to explain what happens, it will be helpful to consider several other cases of override first.

Background: two relevant examples of override

A very clear case of obligatory override happens with number-transparent non-count quantificational nouns. A typical example is illustrated in

[5]  i  [A lot of work] was done.
      ii  [A lot of errors] were made.

In [5], the number agreement is determined by the oblique, the noun phrase (NP) that is the complement of of. Here are some other nouns that can function similarly to lot in this respect: plenty, lots, bags, heaps, loads, oodles, stacks, remainder, rest, number, and couple (CGEL, pp. 349-350).

Another override that will be relevant is the optional plural override that can happen with collective nouns:

[8] The committee has/have not yet come to a decision.

The noun committee is singular, and so simple number agreement would dictate a singular verb. However, there can be an optional plural override. CGEL explains it as follows (p. 502):

The optionality of the override with collectives reflects the fact that there is potentially a difference of meaning between the versions with singular and plural verbs. From one perspective a committee is a single entity, but since a committee (normally) consists of a plurality of members it can be conceptualised as denoting this plural set. The construction with a plural verb focuses on the members of the committee rather than on the committee as a unit. The plural override is therefore not permitted with predicates that are applicable to the whole but not to the individual members; it is, moreover, of questionable acceptability if the collective has one (or a/another) as determiner:

[10]  i  The committee consists/^*consist of two academic staff and three students.
         ii  This committee, at least, is/^*are not chaired by one of the premier's cronies.
        iii  One committee, appointed last year, has/^?have not yet met.

Overrides with percent

NPs whose head is the noun percent are subject to overrides. Here is CGEL (p. 504):

Proportional constructions with per cent (BrE) and percent (AmE)

[16]  i  One percent of students ^*takes/take drugs.
        ii  One percent of the electorate takes/take drugs.
       iii  One percent of the cheese was/^*were contaminated.

Percent is best analysed here as a noun taking an of complement (and for that reason we have used here the single-word AmE spelling). It belongs with the number-transparent nouns, with plural students in [i] requiring a plural verb and singular cheese in [iii] a singular verb: compare a lot of students and a lot of the cheese. In [ii] the head of the oblique is electorate, and since this is a collective noun, singular agreement can be overridden, just as in The electorate aren't going to like this.

A large percentage of…

As we have seen above in the examples from COCA, the phrase a large percentage of… seems to be number-transparent: for the purposes of subject-verb number agreement, what matters is whether the oblique (the NP that is the complement of of) is singular or plural.

But when the oblique is a singular collective noun, it is itself subject to optional plural override.

In COCA, that option does not seem to be taken very often, at least not in the academic portion of its collection:

A large percentage of the island's population depends on the French welfare system.
A large percentage of the world's population lacks access to safe drinking water.
A large percentage of the population was dislocated from their ancestral territory.
When such a large percentage of the population is involved in vehicular usage, …

The optional plural override with collective nouns is more common in British English. We can get a rough estimate of how much more by comparing the number of hits for "population is" and "population was" to the number of hits for "population are" and "population were".

Of course, a good portion of the hits will be false positives, e.g. if the full phase is many in the population are/were…; here population is not the head, while the actual head, many, normally has a plural agreement and is not number-transparent. But the same type of false positives will happen in both the American and the British corpus.

In COCA, the (apparent) singular agreement is more than seven times more numerous than the plural agreement. But in the British National Corpus (BNC), it is only 3.3 times more numerous.

And so it is not surprising that in the BNC we find an example where a large percentage of the population is used with a plural override for population:

A large percentage of the population appear to be earning their living as 'consultants' of one description or another.