1. Quirk et al.'s A Comprehensive Grammar of the English Language

Quirk et al. do, near the beginning of their account of some and any (Quirk et al., Reference Quirk, Greenbaum, Leech and Svartvik1985: 783–4), try to differentiate between them in meaning terms. But not only are their descriptions wrong; they quickly convert reasonably precise meaning into vague general meaning. Some is said to be ‘assertive’ and any ‘non-assertive’. (‘Non-assertive’ is a very questionable term for any – a good case could be made for calling it a very assertive word.) The authors then move even further away from meaning: the choice of some or any is determined by the grammatical category of the sentence. The any series, they say, ‘appears’ in negative, interrogative, conditional, and restrictive relative clauses.

Quirk et al. say that ‘nonassertive items appear … after words that are morphologically negative or that have negative import’. A single one of their examples under this heading should suffice to demonstrate how they seem to be ignoring the fact that some and any have totally different meanings:

24 ‘They can prevent any demonstration.’

One might want, however, to convey a very different meaning:

25 They can prevent some demonstrations, but not others.

Much earlier in their grammar (p. 84) Quirk et al. write of the impossibility of a positive statement containing nonassertive forms. They give as an example

26 ‘*I have any ideas.’

Like so many grammarians, they seem to lack the imagination or the will to think of examples that invalidate their claims. Here, change the particular meaning of the verb, and one gets a perfectly sensible sentence:

27 I welcome any ideas.

There are other flaws in Quirk et al.'s account of some and any. See an extended review of these in Gethin, Reference Gethin2010c.

2. Rodney Huddleston, Geoffrey K. Pullum et al.'s The Cambridge Grammar of the English Language [CGEL]

The basic approach of this massive work, with its huge number of technical terms and minute examination of sentences, is similar to that in Quirk et al., but the grammatical categorising is taken even further. The work devotes, partly or wholly, more than fifty pages to the subject of some and any and their compounds (mainly in chapters 5, by John Payne and Rodney Huddleston, and 9, by Geoffrey K. Pullum and Rodney Huddleston).

The CGEL describes any as a non-affirmative ‘expression of existential quantification’, indicating ‘a quantity or number greater than zero’(Huddleston & Pullum et al., Reference Huddleston and Pullum2002: 358–9). This tells one nothing useful. It conveys nothing of the unique essence of any. One might as well call a or the expressions of existential quantification, since they too indicate a quantity or number greater than zero. The authors go on to maintain (on page 360 by implication, on pages 381 and 383 explicitly) that any has the same sense as some. Yet on page 359 under [5] they give perfect examples, apparently without noticing it, of how crucially different in meaning some and any are (cf. 18 and 19 above):

28 ‘He hadn't eaten some of the meat.’

29 ‘He hadn't eaten any of the meat.’

This sort of confusion is almost bound to arise when meanings are converted into grammatical categories. On page 829 the CGEL insists that ‘positively-oriented’ some is ‘inadmissible’ in the negative

30 ‘*They didn't make some mistakes.’

but that, in contrast, in changed ‘scope’ conditions,

31 ‘I didn't understand some of the points she was trying to make.’

is an ‘admissible’ sentence, which they paraphrase as ‘some of the points…had the property that I didn't understand them’.

In fact 30 is in principle exactly the same as 31. The ‘scope’ relationships can be interpreted as identical in the two sentences (some mistakes had the property that they didn't make them). It is a good example of how unusual or unnatural situations in real life can deceive analysts into thinking that the language expressing such situations is prohibited by some grammatical law. We can make 30 a bit more realistic by expanding it a little – This time they didn't make some of their usual mistakes – or by stressing some. But the CGEL seems to have determined in advance that 30 is ‘inadmissible’ and so declares, back-to-front, that the ‘scope’ is such as to make it so. See 18 and 28, and my discussion under General principles above.

Recognising that any is clearly very often not ‘non-affirmative’, the CGEL creates (p. 361) a new category, ‘free choice’ any, as in:

32 ‘Any of these computers will do.’

Of this any the authors say (p. 382) ‘an arbitrary member (or subquantity) can be selected from the set (or quantity) denoted by the head and the predication property will apply to it’.

This distinction may be valid up to a point. For example, 33 is ambiguous.

33 I won't eat anypies.

The sentence could mean I will eat no pies at all: the any is the CGEL's ‘non-affirmative’ type. Or it could mean I will not eat just any pies, irrespective of quality: the CGEL's ‘free choice’ type of any. (In practice the intonation of the sentence will probably vary according to the meaning intended.)

But the distinction is very fuzzy, and ultimately questionable. Just one instance from the several one could cite from the CGEL is the case of conditional sentences. These, such as (p. 745)

34 ‘If anyone has a solution to this problem, please let me know.’

are supposedly part of the justification for categorising any as ‘non-affirmative’. But they can often be paraphrased to provide examples of what the CGEL would call ‘free choice’ any:

35 Anyone with a solution to this problem should kindly let me know.

See Gethin, Reference Gethin, Huddleston and Pullum2010d, for further detailed discussion of this and all the other difficulties in the CGEL's account of some and any.

The CGEL fails to convey anything of the essential meaning of any. It is much more accurate, practical, and psychologically real – in the sense that it is the way native English-speakers experience any, unless they are grammatical analysts – to regard all anys as one and the same. They all carry the ‘whatever, whichever’ idea.

The CGEL, instead, tries to fit any into its huge, elaborate system of grammatical categories. It claims that ‘non-affirmative’ contexts ‘admit’ ‘negatively-oriented polarity-sensitive items (NPIs)’, of which the ‘non-affirmative’ any series of words, but not ‘free choice’ anys (pp. 823, 826), are an example; while ‘affirmative’ contexts ‘exclude NPIs’ (pp. 822–38). All this rigid categorising, however, leads to difficulties, and sometimes to what seem to be rather desperate attempts to reconcile apparent contradictions. For instance (p. 836) in

36 ‘It would be foolish to take any unnecessary risks.’

there is an ‘NPI’ in what appears to be an affirmative context. Not so, says the CGEL (p. 835). Foolish is one of a number of ‘covertly negative lexical items with clausal or clause-like complements’, one of a ‘large array of lexical items expressing unfavourable evaluations’.

I don't think languages work like that: grammatical categories admitting or excluding other grammatical categories. We use any in 36, not because foolish is covertly negative, but simply because any is the word we have to use in this context if that is the meaning we intend. What else could we say if that is what we want to say? We could use some, or many – but they would give the sentence quite different meanings. Similarly, we do not say things like I have any books. That is not because the any is a ‘non-affirmative quantifier’, but because it is nonsense, while I read any books makes sense. Fundamentally, creating language is not a negative process of ‘admitting’, or ‘excluding’ or ‘sanctioning’ categories, but a positive one of combining different meanings to make the sense one wants. Actually, at this point in the CGEL the authors are not very far from saying precisely this; but their whole approach prevents them admitting any such thing.

The other key distinction the CGEL makes is between the ‘proportional’ and ‘non-proportional’ use of both some and any. The authors say that with ‘proportional’ use ‘we are concerned with quantity relative to some larger set’ (p. 381); but, they say, in a sentence like

37 ‘There are some letters for you.’

some is non-proportional.

On pages 364–5, though, they offer the sentence

38 ‘Some people don't know how to say ‘No’.'

as an example of the ‘proportional’ use of some, because, they say, one is ‘talking about people in general, but people in general constitute a set’. But (in its most common use) some is practically always ‘proportional’, always means a ‘proportion’, a portion, part of a larger set. If I say I have some letters, or some cheese, I am saying that I have a portion of a larger set.

The CGEL compounds the confusion by linking the analysis of some and any to the use of stress. ‘Proportional’ some, the authors say (p. 381), ‘is stressed, and not reducible to /səm/’, as in

39 ‘Some people left early.’

But in talk about human affairs, the idea of people is normally implicit. So if, say, one is talking about a party or a meeting, one normally has no reason to stress people. One wants to stress the some idea. If, though, one changes the topic, for example to ladies or officials, one switches the stress:

40 Some /səm/ ladies/officialsleft early.

See also sentence 11 above.

Their confusion over stress continues. They not only say (p. 383) that ‘non-affirmative’ any ‘is usually unstressed’, while ‘free choice’ any ‘is always stressed’; the examples of any that they give on page 382 as unstressed are in fact stressed, and the ones they give as stressed are in fact unstressed!

3. Ronald Carter and Michael McCarthy. Cambridge Grammar of English [CGE]

There are good things here. One is the attention paid to spoken English. Many of the examples are taken from the huge Cambridge International Corpus of real texts taken from everyday written and spoken English. The authors show wisdom, though, in emphasising that ‘the pedagogic process should be informed by the corpus, not driven or controlled by it’ (Carter & McCarthy, Reference Carter and McCarthy2006: 12). Another excellent feature is the opening section of 141 pages which gives detailed explanations of 81 words known to cause difficulties for learners of English.

Sadly, though, the authors have apparently succumbed, at least partly, to the disease that seems to infect most who enter the halls of academic linguistics. This is the Chomskyan illusion of the primacy of syntax. Large parts of both the book's organisation and its explanations are on the basis of grammatical categorisation. For instance (p. 481), under ‘Compound nouns’,

This back-to-front approach gives the impression that the CGE's task is to discover grammatical categories, rather than show students how to use pieces of language – in this case, where to put the stress. The CGE's approach imitates that of Chomsky and Halle (Reference Chomsky and Halle1968) in The Sound Pattern of English. I have demonstrated some of the flaws in that book in Gethin, Reference Gethin1990: 98–104. One cannot put any piece of language into a grammatical category until one understands its meaning (consider running shoes and running nose). Once one has established meaning, though, grammatical categorisation serves no purpose; it is simply a pointless extra burden on the memory. Unless, that is, it is considered enlightening to discover, by using the CGE's criteria, that OxfordStreet is a compound noun, and, mysteriously, Edgware Road is a noun modifier + head structure. Moreover, the CGE's account is confused and self-contradictory. See a detailed account of its mistakes, and of the principles that really determine stress, at Gethin, Reference Gethin2010e. The obsession with categories continues in chapters like Verb complementation. The authors fall into what I call the Hornby error. There are simpler, better, and more psychologically ‘real’ ways of showing how verbs are used. (See Gethin & Gunnemark, Reference Gethin and Gunnemark1996: 304.)

There are other things in the CGE that suggest that there are some basic principles of language that the authors are unaware of. For instance, they say (p. 7) that computer-assisted research has shown the patterned relationship between vocabulary and grammar, such as ‘the pattern of about twenty verbs in English … followed by the preposition by and an –ing clause’. But hundreds, indeed an indefinite number, of English verbs can be followed by by + -ing. The correct insight into the relationship between vocabulary and grammar is gained, not by registering with computers what a sample of people have said, but by recognising what people can say. See Gethin, Reference Gethin2010f, and Gethin & Gunnemark, Reference Gethin and Gunnemark1996: 298–300.

Where some and any are concerned the CGE is as confused as the CGEL. If students of English are ever told to study all three of the works I review here, they will surely be bewildered to find that each one makes distinctions between some and any that are different from the other two; and indeed that in the end none of them really distinguish properly between the meanings of the two words at all.

The CGE in effect tells us that it is an inexplicable, arbitrary system of stress which determines the use of both words. I have explained the error of this relationship of stress with some and any above. See my argument around sentences 11, 39 and 40. (The CGE in fact adds a further error of its own. See Gethin, Reference Gethin2010g, for a detailed account of all the CGE's mistakes concerning some and any.)

The CGE ends up making the same old basic mistakes (pp. 366–7):

But as we have seen:

1. (a) some and compounds are often used in negative declarative clauses, as in examples 9, 15, 18, 23, 28, 30, 31, 38;

2. (b) any and compounds are often used in affirmative declarative clauses, as in examples 12, 13, 20, 22, 32, 35, 36; while the statement that any does not usually occur with affirmatives would be useless even if it was true. (The CGE's specifying of weak forms is irrelevant. A variety of stress patterns is possible in the sentences listed.)

Unfortunately, even if the information in Quirk et al. and the CGEL was unfailingly correct, and did not so frequently illustrate false principles, studying these grammars would certainly not be the way to learn English. One learns a foreign language by observing it in use. One does not learn the true essence of words through definition and exhaustive analysis. It is hard to see what purpose these huge works serve. They add nothing to understanding or real knowledge.

The CGE could be helpful for students of English. Its greatest usefulness would be as a reference work. However, this is much reduced by the poor index – try looking up, for instance, the/definite article, -ing/gerund, or stress. Unfortunately the search function on the CD-ROM accompanying the book is in some ways even weaker as a reference tool than the printed index.

There is not just the problem that grammarians frequently get the facts wrong. There is the much more serious and fundamental problem that they often give the wrong impression of how language works. The Chomskyan illusion persists. Grammarians sometimes give the impression that meaning can be (or is always?) irrelevant, that the way words are used is determined by grammatical categories. But as Michael Bulley says (Reference Bulley2007: 56), grammar is not discovered; it is made up.