November 15, 2006

If it was good enough for King Alfred the Great...

Do you own a copy of Merriam-Webster's Concise Dictionary of English Usage? If not, go immediately to your favorite bookseller and buy one. Believe me, it'll be the best $13.22 (or even $16.95, if you pay list price) that you've spent in a while. Geoff Pullum recommended it last year ("Don't put up with usage abuse", 1/15/2005), in response to a reader's question about what references or authorities to trust with respect to style and usage. Geoff used blurb-worthy phrases like "the best usage book I know of" and "this book ... is utterly wonderful", and I agree with him.

Why am I plugging this book today? Because it provides a perfect answer to a note from a reader about the use of less and fewer.

Matt Cockerill send in a link to an article in the Guardian (John Mullan, "M&S: the pedant's store", 10/6/2006). Apparently a customer complained about apostrophe placement ("I do not care to dress my child in a top containing a glaring grammatical giraffe gaffe"), and after an appeal to their "childrenswear technologist", M&S withdrew the offending item from their stores, apologized, and sent a refund. Matt focused on the article's passing mention of an earlier M&S capitulation to customers' grammatical prejudices:

M&S, of course, likes to project a classy image and this confession of grievous fault rather neatly confirms it as the favoured shop of those with high standards, in grammar as in everything else. A few years ago it changed its "6 items or less" checkout signs for replacement signs declaring, more correctly, "6 items or fewer", reportedly after customers had grumbled.

Matt observed that this "crops up as a standard example of the shoddy grammar of our modern age in newspaper articles here all the time", and registered a counter-grumble:

... the weird thing is, I've never once seen anyone point out that there's nothing grammatically wrong with '5 items or less', and in fact it's much more natural and less stilted sounding than '5 items or fewer'.

The key, as far as I'm concerned, is to realize that it's quite valid to think of '5 items or less' to imply an ellipsis:
"5 items or less... [than that amount of shopping]"

and in that it's no different from any number of standard grammatical usages which make use of ellipsis.

I'm also always tempted to ask whether they would replace the sign outside a kids playground to indicate that it may be used only by children who are "5 years old or fewer"...

Matt's grammatical instincts are exactly right. He's also correct in observing that with ages -- and in certain other cases of countables as well, which MWCDEU summarizes as "distances, sums of money, units of time, and statistical enumerations" -- less is generally preferred to fewer. And Matt's observation about a possible construal of "5 items or less" also seem valid to me, although I think it's a secondary point. The primary point is that the now-standard pedantry about less/fewer is in fact one of the many false "rules" that have recently precipitated out of the over-saturated solution of linguistic ignorance where most usage advice is brewed.

But not the usage advice at MWCDEU. This is the start of its entry on less/fewer:

Here is the rule as it is usually encountered: fewer refers to number among things that are counted, and less refers to quantity or amount among things that are measured. This rule is simple enough and easy enough to follow. It has only one fault -- it is not accurate for all usage. If we were to write the rule from the observation of actual usage, it would be the same for fewer: fewer does refer to number among things that are counted. However, it would be different for less: less refers to quantity or amount among things that are measured and to number among things that are counted. Our amended rule describes the actual usage of the past thousand years or so.

As far as we have been able to discover, the received rule originated in 1770 as a comment on less:

This Word is most commonly used in speaking of a Number; where I shoudl think Fewer would do better. No Fewer than a Hundred appears to me not only more elegant than No less than a Hundred, but strictly proper. --Baker 1770

Baker's remarks about fewer express clearly and modestly -- "I should think," "appears to me" -- his own taste and preference. [...]

How Baker's opinion came to be an inviolable rule, we do not know. But we do know that many people believe it is such. Simon 1980, for instance, calls the "less than 50,000 words" he found in a book about Joseph Conrad a "whopping" error.

The OED shows that less has been used of countables since the time of King Alfred the Great -- he used it that way in one of his own translations from Latin -- more than a thousand years ago (in about 888). So essentially less has been used of countables in English for just about as long as there has been a written English language. After about 900 years Robert Baker opined that fewer might be more elegant and proper. Almost every usage writer since Baker has followed Baker's lead, and generations of English teachers have swelled the chorus. The result seems to be a fairly large number of people who now believe less used of countables to be wrong, though its standardness is easily demonstrated.

MWCDEU then gives a couple of pages of illustrative example in both directions, dealing especially with the "common constructions" with countables where less continues to be used more often than fewer "in present-day written usage". The concluding advice:

If you are a native speaker, your use of less and fewer can reliably be guided by your ear. If you are not a native speaker, you will find that the simple rule with which we started is a safe guide, except for the constructions for which we have shown less to be preferred.

I've scanned the whole less/fewer entry, and made it available here. Now validate my stretching the boundaries of fair use, and go buy the book!

I'm going to add a couple of observations based on web searches. First, Google News validates MWCDEU's observation about the difference between countables in general -- where journalists and their editors prefer fewer about 2-to-1 -- and things like units of time and amounts of money, where less is preferred by a whopping margin. Here are counts for a few different countables, in the "less/fewer than N items" construction:

votes people players pages hours minutes seconds dollars
less than N __
fewer than N __

And the same tendencies can be seen on the web in general, except that the ratios are generally shifted in the direction of less. No doubt this is due to the effects of copy-editing on the Google News sample.

votes people players pages hours minutes seconds dollars
less than N __ 272,000 2,570,000 183,000 696,000 9,110,000 13,400,000 3,840,000 2,220,000
fewer than N __ 182,000 837,000 61,500 112,000 552,000 92,200 22,300 921
less/fewer 1.50 3.07 2.98 6.2 16.5 145.3 172 2,410

But interestingly, in that "N items or less/fewer" construction, the less/fewer ratios generally shift away from fewer and towards less. At least, that's clearly true for the cases of countables where fewer is reasonably common to start with. Here are the counts from Google News:

votes people players pages hours minutes seconds dollars
N __ or less
N __ or fewer

And here are counts from the web at large:

votes people players pages hours minutes seconds dollars
N __ or less 14,200 134,000 13,200 215,000 1,890,000 2,100,000 944,000 248,000
N __ or fewer 495 30,600 1,570 19,500 24,200 757 6,760 125
less/fewer 28.7 4.38 8.41 11.0 78.1 2,774 140 1,984

I'm not sure whether Matt's ellipsis theory is the reason for this shift, though it's a reasonable possibility: when someone writes or says "1,000 votes or less", they may well mean "1,000 votes or less of a margin than that", rather than "1,000 votes or less votes than that". But King Alfred says that they'd be OK either way, and so do most other English writers in the millennium between his time and ours.

[A small pedantic confession: King Alfred had the genitive case at his disposal, and so his use of less with a count noun is actually a partitive construction of a type that we can't copy idiomatically in modern English -- "less of words":

c888 K. ÆLFRED Boeth. xxxv. §5 [6] Swa mid læs worda swa mid ma, swæðer we hit ȝereccan maȝon.  ("whether we may prove it with less words or with more")

But still. And I'll make up for doubting Alfred's modern relevance by filling in one data point from the intervening centuries -- a footnote by Alexander Pope, to book XIV, verse 291 of his translation of the Iliad:

But whoever considers his Circumstances will judge after another manner. Priam, after having been the most wealthy, most powerful and formidable Monarch of Asia, becomes all at once the most miserable of Men; He loses in less than eight Days the best of his Army, and a great Number of virtuous Sons; he loses the bravest of 'em all, his Glory and his Defence, the gallant Hector.

The use of less with a count of time-units has always been preferred, as MWCDEU observes. But I was surprised by that "bravest of 'em all", in a footnote no less. ]

[Update: Full access to Merriam-Webster's Dictionary of English Usage is now available from Google Books, with the less/fewer entry starting on page 592.]

Posted by Mark Liberman at November 15, 2006 08:41 AM