The English-Learning and Languages Review Æ Homepage
A critical examination of the account of the definite article in
Quirk et al.’s A Comprehensive Grammar of the English Language (1985)
(The false trail of grammatical categorising 1)
This article is a slightly edited extract (pp.77-87) from my book Antilinguistics, published by Intellect in 1990.
Everybody, of course, should be granted a few mistakes without suffering excessively severe strictures, even if one would expect fewer from a team of four professors working for so long with numerous assistants. However, it is a tiny detail that reveals something even more important, and shows how wrong and impractical their whole approach is. On page 252 of Quirk et al.’s book there is a note [b]:
“It can be argued that some nouns, like weather, are neither count (*a weather) nor noncount (*a lot of weather), but these nouns share features belonging to both classes. Noncount noun features include the premodified structures a lot of good weather, some bad weather, what lovely weather. On the other hand, count noun features include the plural go out in all weathers, in the worst of weathers.”
This is typical of the mistake of applying abstract grammatical analyses instead of thinking naturally about meaning. They apply an inflexible test – “a lot of” – and fall straight into the trap. There is no grammatical quirk in the word “weather”. It is simply the particular meaning of “weather” that makes “a lot of weather” somewhat uncommon. One does not often, in life, mean that. But it is by no means impossible:
A lot of weather is determined largely by temperature.
As for “weathers”, that is a perfectly normal phenomenon found with ‘noncount’ nouns: the plural form is used to express the idea of ‘sorts of’:
You can find
some wonderful cheeses in
They make good wines in this region.
Quirk et al.’s and most grammarians’ attitude is strange. On page 288 Quirk et al. acknowledge that grammatical categories have unclear boundaries. Yet they immediately go on:
"... we might wonder whether King's in King's College is to be classed as a genitive common noun which is part of a composite common name."
They talk as if there was some absolute truth of the matter. We may wonder, but God knows the classes - if only we could find out too. If they do not believe that God knows, they are effectively admitting that they are engaging in a ritual of human-made labelling for its own sake, analysis for the fun of it, without meaning in reality.
The more they classify, the more categories they can establish, the happier they seem to be. Consider Quirk et al.'s treatment of "the". Even before they have separated "the" from "a" they straight away divide the articles into "specific reference" and "generic reference". By that they mean the difference between, for instance,
25 There are two giraffes in the garden.
26 Giraffes are strange looking animals.
where in 25 one means two 'specific' individual giraffes, but in 26 giraffes in general.
They do make an attempt to explain the meaning of "the" itself (p.265). They write:
"The definite article the is used to mark the phrase it introduces as definite, ie as 'referring to something which can be identified uniquely in the contextual or general knowledge shared by speaker and hearer',"
Any students of English who manage to understand this language will be misled by “…knowledge shared…” I guess fewer than half the people who have trouble with their cars and are told
27 It's the constant velocity joint,
share any knowledge of the matter with the mechanic who tells them. But they will not question the mechanic's "the", because they know which constant velocity joint it is - it's the one on their car. And in fact a vast amount of the "the" language that adults use to children refers to things that are not familiar to the children.
The confusion Quirk et al. have got themselves into is quickly confirmed, by a note on the next page (266):
"In practice, since a speaker cannot always be sure of the hearer's state of knowledge, use of the involves a certain amount of guesswork. In fact, in some cases the assumption of shared knowledge is a palpable fiction. Notices such as Mind the step and Beware the dog, for example, generally do not assume that the reader was previously aware of the hazards in question."
"The" has nothing to do with the sort of "shared knowledge" Quirk et al. are worrying about, and so there is no problem whatever with "Mind the step" or "Beware the dog". Just as "the" in 27 tells you that "the constant velocity joint" is the particular one on your car, so "Mind the step" tells us of the particular step right there in front of us, and "Beware the dog" tells us of the particular dog lurking around those parts. Such notices have to say "the" to make sense. Some people might get very alarmed if they found notices saying "Mind a step" or "Beware a dog" and watch out furtively for steps and dogs for the rest of their lives.
But by this stage Quirk and his colleagues have started on something even worse, though it is an extension of the same fundamental fault. Under one heading alone, "Specific reference...Uses of the definite article" (which is only the beginning of their treatment of "the"), they establish no less than eight categories. (The examples are mine.)
(a) Immediate situation - e.g. Would you like to see the garden?
situation (general knowledge) - e.g. The world Cup was
(c) Anaphoric reference: direct - e.g. We've a dog and a cat. The cat's called Tonk.
(d) Anaphoric reference: indirect - e.g. This new cassette's no good. I'll take it back to the shop.
Cataphoric reference - e.g. The history of
(f) Sporadic reference - e.g. We're not on the telephone.
(g) The 'logical' use of the - e.g. Rico is the eldest son.
(h) The use of the with reference to body parts - e.g. He was hit in the chest.
All this categorization is pernicious in two ways, and the second evil arises out of the first.
First is that the categorizing gives a wholly wrong impression of how language works. The truth is that "the" (when used in the common way I am discussing here, as the 'definite article', though academic linguists waste energy debating that term too) has just one simple basic meaning. All the other meanings of complete phrases or sentences arise merely as the result of that one simple basic meaning in combination with other meanings. Exactly the same applies to, say, "a", or plurals, or words for uncountable things without "the", and so on. (And I mean "things", not "nouns". This is a crucial point. You can perhaps see what I have in mind if you think of the example above of "weather" again.) There is no such thing as an 'anaphoric' "the" as opposed to a 'situational reference' "the" as opposed to a 'cataphoric' "the" etc. They are all exactly the same "the".
It is not possible to describe any word accurately with other words. The only perfectly true description of the meaning of "the" is "the". But one can give a pretty good and useful idea of what it means if one says it has the sense of "that particular limited (thing), you know which I mean". Actually, the second part of that description is almost certainly unnecessary - the "you know which I mean" idea must surely always follow inevitably from the first basic idea.
28 We've a dog and a cat. The cat's called Tonk.
there is no "the" of 'anaphoric reference' in the second sentence. There's just plain "the". Of course the second sentence is about the cat in the first sentence, but that does not excuse saying that the "the" is anaphoric. The second sentence is about the cat in the first sentence because "the" - old, familiar, constant "the" - has been used together with "cat". Quirk et al. want us to believe virtually the reverse: that a special sort of "the" (anaphoric "the") is being used because the second sentence is about the cat in the first sentence.
The absurdity of dividing "the" up into categories can be seen if one imagines a grammarian analysing "does" who says that when a speaker uses "he does", "does" has male reference; but it can also have female reference, as in "she does", or "the hen does", or "his wife does". But naturally the male/female distinction is in no way the foundation or starting point of the grammar of "does", and it is sheer befogging analysing to present language on this basis. Meaning is the basis, and the result.
The categorizing approach gives the wrong impression of language in general; and it leads to wrong practical conclusions on individual points of grammar. Here is a single example of the latter - under (c) Anaphoric reference: direct, Quirk et al. write:
"the first reference to an object will ordinarily be indefinite, but once the object has been introduced into the discourse in this way, it can be treated as 'contextually known', and can thenceforward be referred to by means of the definite article." (p.267)
Readers who speak a language without articles can draw quite the wrong conclusion from this, for it means that the following conversation would be normal.
29a Bill: What are those?
29b Ted: A melon and a pineapple, of course.
29c Bill: It's the very small melon.
29d Bill: Yes, I can see it's the melon, but what's it doing here?
Now, I dare say the classifying grammarians will find a way of wriggling out of this. That will be quite beside the point, because the only important issue is: What is the practical use of their grammars? Mistakes like those in 29c and 29d are inevitable with their methods. But give students a simple explanation based on the meaning of the word itself, and no more, and everything falls naturally into place. It will become clear, for instance, that one cannot use "the" for "melon" in 29c or 29d because there one does not mean "that particular limited thing"; one means instead "one of the several (or many) that exist."
Certainly one will sometimes have to point out to a language student what a particular sentence or phrase means, and so one will be involved in saying things like "Look, clearly we are talking about all tigers in general here, not particular tigers". But the meaning of the words are all one needs, not some artificial system of categories which will only be another barrier to understanding, insight and grasp.
The analysing and categorizing system not only gives students false information; it imposes a completely unnecessary burden on them. Naturally, describing words' meanings is not in itself enough. One has to give examples. But the larger part of Quirk et al.'s explanations in their section on the articles, for instance, is concerned purely with an elaboration of their categories or trying to get out of the tangles that their method must lead to. The whole of their section on Cataphoric reference (pp.268-69) is typical, but I will quote only two parts of it in illustration:
"Sometimes the definite noun phrase can be contrasted with an equivalent indefinite phrase. In such instances, the definite article is not in fact genuinely cataphoric but entails some degree of anaphoric reference."
"Note It is not necessary to postulate that the expression  presupposes some unspoken preamble such as [5a] but, rather, such as [5b]:
the mud on your coat 
There's some mud on your coat. [5a]
You know there's mud on your coat. [5b] "
If students seriously get down to grasping and learning all this - and remember that it goes on and on for the greater part of 1779 pages - their heads will soon start to spin.
Again, under (d) Anaphoric reference: indirect, the authors are exercised by the problem of how much indirect anaphora is combined with cataphoric reference, and so whether the use of "the" is explained by ellipsis. And so on. It starts all over again when Quirk et al. get to "The articles in generic reference" (p.281). On page 283 there is:
“The Romans is thus a generalization (like the mathematicians, the teenagers, the birds, etc in similar use), whereas the Roman implies a generic statement. Nevertheless, it will be convenient to apply the term 'generic' to both the singular and plural uses."
Convenient for whom and why? Surely convenient only for analysers trying to fit everything into their classification scheme. The authors of this tome are only the tip of the iceberg. Year after year after year grammarians publish articles in the linguistic journals earnestly debating how generic 'generic' "a" and "the" really are. Naturally the debate is only a minute part of the whole, of the constant analytical linguistic debate among academics.
Yet, in the more than 30 pages of their section on the articles, with all their classification Quirk and his colleagues have no room for two of the most important practical problems connected with "the" that face students of English as a foreign language: nouns with modifying adjectives, and nouns modified by "of' phrases.
the electrical industry
the French impressionists
And without "the":
history of great interest
literature in the nineteenth century
the history of music
the literature of the nineteenth century
(See Cook, Gethin, Mitchell, A new way to proficiency in English, 1967/1980, pp.79-81.)
All that Quirk et al. have to say on the two problems is on page 286 under the heading "The articles in generic reference":
"Normally the zero article also occurs when the...noun is premodified.
she's studying European history."
[By “the zero article...occurs" they mean that the article does not occur.]
when the same noun is postmodified, especially by an of-
phrase, the definite article normally precedes it: She is studying the history
There is no explanation of why and when some nouns modified by adjectives take the article and others don't; and no explanation of why and when some "of' (and other prepositional) phrases make "the" necessary and others don't. Yet these, particularly the first, are 'mysteries' that cause difficulties for millions of students of English. It is hard to see what is the purpose of a grammar that makes no attempt to clear them up. It is true that in an earlier version of their book, A university grammar of English (1973/1979, pp.71-72), Quirk and Greenbaum had very slightly more to say on the subject:
"...English tends to make a liberal interpretation of the concept 'generic' in such cases, so that the zero article is used also where the reference of the noun head is restricted by premodification. ...Canadian paper..."
What is the good of making these beds of Procrustes for themselves? The uselessness of their sort of classification and analysis is shown up clearly by the vagueness of the weak attempts to get round difficulties. In this case, for example, they have got themselves into a tangle only by insisting on the 'generic' classification in the first place. Quirk et al. talk of "the generic the" as if the word "the" itself had different functions, and itself gave generic (or other) meaning to nouns - instead of it being the meaning of each specific phrase or sentence in its context that gives the meaning - i.e. the meaning is the meaning! It isn't "the" that is generic - it is the phrase that has generic meaning, or not, as the case may be. In any case, as we have seen, there are different sorts of 'generic', and disagreement as to what has true generic meaning or not. This surely confirms that the linguists have got the wrong end of the stick.
What students need to have shown and explained to them are the most basic meanings. Here are three of them:
They can then be illustrated in varying contexts. But Quirk et al. and many other grammarians insist on lumping them together under the heading 'generic' and then find they have to unravel them again. Quirk and his fellow authors fail to do this properly and so only achieve confusion. All I can find in A Comprehensive Grammar...(pp.281-82) is:
A bull terrier makes an excellent watchdog [1b]
Bull terriers make excellent watchdogs [1c]
....It should not, however, be assumed that the three options [1a], [1b], and [1c] are in free variation. One difference between them is that, whereas the [1a] keeps its generic functions in nonsubject positions in the sentence, a/an [1b], and to a lesser extent zero [1c], tend to lose their generic function in these positions:
the medieval mystery play [3a]
medieval mystery plays [3c]
Of these, only [3a] refers to mystery plays as a genre; [3b] refers
to only one play; and [3c] is most likely to refer only to a subset of them.
The generic use of a / an picks out ANY REPRESENTATIVE MEMBER OF THE CLASS. Thus any can be substituted for a / an in examples like:
The best way to learn a language is to live among its speakers.
Generic a / an is therefore restricted in that it cannot be used in attributing properties which belong to the class or species as a whole. Thus:
The tiger is
becoming almost extinct.
BUT NOT: *A tiger is becoming almost extinct.
The generic use of zero article with both plural nouns and noncount nouns identifies the class considered as an UNDIFFERENTIATED WHOLE (cf 5.39 f):
Cigarettes are bad for your health.
The is rather limited in its generic function. With singular heads, it is often formal or literary in tone, indicating THE CLASS AS REPRESENTED BY ITS TYPICAL SPECI:MEN:
A great deal of illness originates in the mind.
No one knows precisely when the wheel was invented.
My colleague has written a book on the definite article in Spanish.
Marianne plays the harp very well."
These explanations do not, I think, explain differences - and 'impossibilities' - like the following:
30 Bill trains a horse.
31 Bill trains the horse.
34 The horse gets its training at places like Bill's.
35 Horses get their training at places like Bill's.
30 certainly fits what Quirk et al. say about their example [3b] (above) though Bill appears to have a rather limited life. But already here one notes that they merely say "tend to lose their generic function" (my italics). That is not satisfactory for any student hoping for clear guidance. 31 is impossible unless either the speaker is talking about one particular horse, or Bill is God. Yet there is nothing in the exposition above to tell one that. Is it not the fullest generic sense that is needed and is not the "the horse" form the very type of the generic? Students must be forgiven for believing both these things if they study Quirk et al., and they will get decisive confirmation of their conclusion when they note [3a] –
Nora has been studying the medieval mystery play.
- and what is said about it. For 'syntactically' [3a] is effectively exactly the same as 31. The only difference that might be significant is that 31 uses the 'simple' form rather than the "-ing" form of the verb. But any students who have read Quirk and Greenbaum's earlier A university grammar of English (1973/79, p.68) will have their conviction strengthened further by
“There is considerable (though by no means complete) interdependence between the dynamic/stative [i.e. "-ing"/simple] dichotomy in the verb phrase and the specific/generic dichotomy in the noun phrase...”
So what is to stop students thinking that 31 is just what they need? The answer to that question is once more: don't think about irrelevant classifications, but concentrate on the particular 'whole' meanings produced by the particular meanings of the parts - "studying" is different from "training", and "the medieval mystery play" is different from "the horse", and will affect the outcome accordingly.
Notice, too, not only the familiar vagueness, but also the old mistake of turning what is simply more common in life into a function of language. It is quite irrelevant that people do not so often want to express 'generic' with 'dynamic' as 'generic' with 'stative', and it is utterly misleading to draw attention to non-existent linguistic associations. If the idea one wants to express happens to be "Peanuts (in general) are doing something", that is what one has to say, irrespective of how rare that situation may be in the world.
Quirk- et al. might of course be ingenious enough to show that their description of the "the horse" form as "indicating the class as represented by its typical specimen" is precisely the right one for preventing the use of 31. But they can surely not seriously pretend that anybody who did not know how to use the articles in English already could work out from such an abstract formula that "Bill trains the horse" is not what they mean.
32 (above) is the 'right' one. How are students to know this from Quirk et al.? Is "horses" specific or generic here? And which do we want anyway? If generic, how generic? Presumably not anything generic in the sense of "Horses are beautiful". Perhaps their reference to a subset (but how generic is a subset?) in non-subject positions is just what we should bear in mind here. But then, why can't we say "Bill trains some horses", for isn't "some horses" a subset? And do we identify something as a subset first and then choose the appropriate words accordingly? And if that is how we work, what happens to ideas like "I like horses", where "horses" is certainly not a subset. All this is a perfect example of how totally misguided the classificatory approach is, not only to the theoretical problems of grammarians, but also, more seriously, to the practical problems of students.
From Quirk et al.'s examples [1a], [1b], [1c] (above)one would think that 33, 34 and 35 (above)are all perfectly acceptable. The various "horse" phrases are all subjects. But again it is the particular meanings that prevent this, not any of the distinctions noted in A comprehensive grammar... 34 is impossible in the same way as 31. If 33 and 35 are to be taken in a general or 'world' context, they are patently untrue; in the limited context of horse racing both are a possibility, although "The horses" seems to me more natural than merely "Horses".
Meanwhile bewildered students can only become more frustrated than ever by the failure to tell them the things they really need to know.
Quirk et al.’s treatment of the articles is not at all an isolated case. For instance, when they come to the distinction between "some" and "any" (and their compounds) they immediately start in their usual way. "Some" is declared to be 'assertive', "any" 'non-assertive'. This is exactly the same cart-before-the-horse approach as they applied to the articles. Once more they fail to see that it is the basic, original meanings of "some" and "any" that produce, sometimes, sentences that can be analysed or classified in this way………………….(See further www.lingua.org.uk/saquirk.htm)
Revised 8 April 2010
The English-Learning and Languages Review Æ Homepage