Rigid Complementarity
« previous post | next post »
Over on the American Dialect Society mailing list, we've returned to a topic last aired there in 2007: the alternation between and and zero in numerical expressions like "two hundred (and) six" (in speech and in writing), in particular when they are used as determiners, as in "two hundred (and) six elephants". These discussions quickly range over a variety of types of numerical expressions, uses of them, and contexts for these uses. Plus a lot of back-and-forth about the acceptability of the variants.
Several sorts of numerical expressions recur in these discussions, among them those expressing a whole number plus a fraction, and those in the related case of a dollar amount plus a cent amount. Until this morning, I'd attributed the appearance of these cases (which seemed to me to be irrelevant to case above) to simple thread drift, one phenomenon reminding people of phenomena that are similar to it in some respect. But then Russ McClay posted a collection of net discussions that suggested to me that something much more interesting — something familiar to me from discussions of other alternations — is going on. A bit of what I wrote to ADS-L on 5/2/07:
I recall being taught at some point in school that things like "one hundred and two", "two hundred and thirty", etc. were vulgar errors (in both speech and writing), that "and" should never be used in such expressions. (This might have been an instance of Omit Needless Words). The lesson seems not to have stuck with me, since I sometimes use one version, sometimes the other.
I spent some time searching through usage manuals for pronouncements on the matter, but with no success, though I did come across evidence that the Chicago Manual of Style 15 considers both variants acceptable. What I did get, them and now, was lots of reports about what particular people had been taught in school and what their current practices were. A number of people (all, I think, Americans) said that they had been taught that this and was just wrong, and some of them seem to have taken this advice to heart, condemning this and in strong terms. A few reported that they had been taught that both variants were acceptable. A great many recalled no instruction at all on the matter. I concluded that in the United States, at least, the proscription of and was maintained entirely as a bit of English-teacher lore, with some effect on some students.
More recently, commenters have suggested that the and variant is used by British and Australian speakers, the zero variant by Americans; this is surely overstated, but there might well be some statistical differences in use (possibly resulting from school instruction in the U.S.!). In any case, both variants are used, by speakers and writers of all sorts. There are all sorts of further details, but it's sufficient for my current discussion that both variants are standard, so that the question is why the and variant is stigmatized by some. (Omit Needless Words seems insufficient to explain the phenomenon, and indeed ONW is almost always appealed to "after the fact", as a justification for some proscription that arises from some other source.)
(The general acceptability of "numerical" and is entertainingly illustrated by some titles with cardinal number expressions in them that are virtually always spoken with the and: 1001 Nights, 101 Dalmatians.)
Illumination began to come to me when Michael Covarrubias wrote to ADS-L (I have edited this a bit):
In the 3rd grade, Mrs Wolford taught us to use "and" only when there was a decimal — 'one hundred fifteen and seventeen hundredths' — but I never understood why that should be the only use. She probably made some argument about a garden-path-like confusion, even tho there really wouldn't be any confusion there.
'One hundred and fifteen and seventeen hundredths' works just as well.
It's the "only" that's crucial here. Then came Russ McClay's collected comments, which included some referring to one or both of the two cases I mentioned above — whole number plus a fraction, dollar amount plus a cent amount — saying that that's the only time you can use numerical and.
Now, in the whole-number+fraction case, numerical and is not only possible, but obligatory: "one hundred fifteen seventeen hundredths" is just not acceptable (in speech or writing). The dollar+cent case is a bit more complicated, because in certain technical, legal, and administrative contexts, the zero variant is well attested:
… an amount of not more than two dollars fifty cents per acre-foot per year. (link)
Taking and approving each bond and returning it to the proper court when necessary, seven dollars fifty cents. (link)
But in ordinary speech, and in writing outside of these contexts, numerical and is vastly preferred over zero — "That cup of coffee cost me a dollar fifty cents" is odd (though the colloquial "a dollar fifty" is fine, and British usage with monetary amounts is different from American) — and I imagine that many teachers treat the and as obligatory here.
(What follows is a slightly revised version of a recent posting of mine to ADS-L.)
The upshot is that there seems to be a principle of "grammatical reasoning" here that almost no one (even usage advisers, much less non-specialists) makes explicit; probably they're not aware that they're using it. The version here is:
(1) The only places variant X is acceptable are where it's obligatory.
(with the understood continuation: otherwise, the alternative, Y, is used.)
This is equivalent to the formulation:
(2) Don't use variant X unless the alternative, Y, is unacceptable — in which case, use X.
The version in (2) is given at the end of my posting (on my own blog) on for vs. because, where it follows a more elaborate unpacking of the reasoning (in three steps), as applied to for/because, linking but/however, and restrictive which/that. (See the Appendix below for a version of this formulation.) The principle has the function of enforcing rigid complementary distribution (rather than partial free variation), via a "division of labor" between the variants — all this in service of One Right Way.
The posting on my blog has links to earlier postings on various case studies. The historical precedent is "Fowler's famous suggestion that the labor of signaling relative clauses might be divided between that and which" (as I put it in this posting).
Fowler's proposal was merely a tentative suggestion, but it since been elevated to a rigid rule (which we've blogged about many times here on Language Log). I was originally going to call the general principle of grammatical reasoning Fowler's Syllogism, but that seemed unfair to Fowler, so for the moment I'm settling for the less punchy name Rigid Complementarity Principle (RCP).
In any case, and/zero is yet another case of the RCP in action (I also have a piece in preparation on apostrophe/zero alternations, where some people have appealed, implicitly, to the RCP).
I am not, of course, recommending the RCP, which I think is a silly idea.
Appendix: the Rigid Complementary Principle (elaborated formulation)
[An exemplification: variant I is the relativizer that, and variant II is the relativizer which; context A is restrictive relative clauses, and context B is non-restrictive relative clauses. Another exemplification: variant I is zero, and variant II is and; context A is cardinals used as determiners, and context B is whole-number+fraction expressions.]
1. Variant I occurs in context A but not in context B.
2. In actual practice, variant II can occur in both A and B.
3. To enforce complementary distribution, II should then be confined to the list of contexts in which it's obligatory (so it's barred from A), and I occurs elsewhere.