## Less with plural count nouns in formal usage

Looking over the comments on Geoff Pullum's recent post "Stupid less/fewer automatism at the WSJ", I see one point — implicit in many contributions, and explicit in a few — that deserves to be underlined with some empirical evidence. When a numerically-quantified plural noun phrase refers to an amount that may be fractionally divided (grams, seconds, meters, dollars, etc.), it's generally incorrect to follow it with or fewer rather than or less. And even when the counted units are not normally divisible, but are being considered as a mass-like quantity (points, votes, cents, items, soldiers), less may also be permitted or even preferred in formal writing.

Google Scholar indexes a wide range of scholarly, scientific, and technical sources, most of which are carefully edited. If we search Google Scholar for "grams or less", we get almost 5,000 hits — things like:

Intellectual and functional status at school entry of children who weighed 1000 grams or less at birth
It appears that for those passerine species weighing 22 grams or less, the lower critical temperature is commonly above 28°C
This cell was designed for samples of 20 grams or less …

In contrast, if we search for "grams or fewer", we get about 25 hits. Some of these strike me as flat-out ungrammatical, and were probably introduced by someone who had overdosed on bad usage manuals:

In a first aspect then, the invention is directed to a nut spread including nuts, and added vegetable oil, the nut spread having 3.5 grams or fewer of total adjusted carbohydrate per 2 tablespoon serving.

Others are merely awkward, or reflect a context in which people seem to be counting grams as indivisible units:

Each product contains six grams or fewer of net carbs per serving.
A main dish or meal must meet the 3 grams or fewer criterion per 100 grams and not derive more than 30 percent of calories from fat.
So if you're on a 2,000-calorie-a-day eating plan, then it should contain no more than 65 grams of fat, 20 grams or fewer of which are saturated fat.

It's worth noting that plural quantified noun phrases of this type often take singular verb agreement:

Ten seconds is a long enough period of time for TCP to ramp up new transfers to their full capacity.
BF3 is bubbled through a glass tube into the methanol until 125 grams is taken up.
The Cobb-Douglas specification suggests that 3.0 meters is also an important threshold in terms of public preferences.

As for things like votes and points, we can find plausible examples that go both ways, in approximately similar numbers, from well-written and well-edited sources like the New York Times:

… what would have been a disastrous night, considering how many House seats were decided by margins of 5000 votes or less.
in fully a quarter of the races, the difference between the two candidates was 1000 votes or less.

... 197 municipal elections this year had been decided by 100 or fewer votes, 74 by 25 votes or fewer and 15 by five votes or fewer.
Franck, Mozart, Rachrnaninoff, Haydn and Schumann, in that order, received 100 votes or fewer.
Several earlier unofficial recounts gave each candidate leads of 25 votes or fewer.

Six of the Chargers' games … have been decided by 5 points or less.
And the Broncos are 5-0 in games decided by 4 points or less.
It seems so because all six of Notre Dame's losses have come by 7 points or less.

The record was established in 1997, when 27.9 percent of regular-season games were decided by 3 points or fewer.
Seven of Cleveland's games have been decided by 7 points or fewer …
The 76ers have lost two games by 3 points or fewer …

And again, things likes points and votes often take singular verb agreement, when they are being considered as mass quantities:

“Sixty points is a lot” Llodra said …
… unless the danger is so extremely remote that 30 total points is too high a price to pay …
OPA wondered for a long time how to find out whether, for instance, 8 points is too much or too little for a can of peas.
The 40 points is the most ever scored against the Huskies at Gampel Pavilion.

At this stage of the game, 225 votes is a substantial margin.
60000 votes is a significant difference between the Democratic and Republican turnouts.

[Note that these points are emphasized in the treatment of less/fewer in Merriam-Webster's Dictionary of English Usage, as discussed here.]

1. ### old_maltese said,

December 5, 2010 @ 10:10 am

I hope that you have it backwards in the first paragraph: it's generally incorrect to say 'fewer grams' or 'fewer dollars'?

[(myl) Phrases like 'fewer grams" are obviously correct, though such things are rare except in constructions like "fewer grams of food per gram of body weight", "ten or fewer grams of cocaine", etc. But when explicit numerical quantification is added, I do mean that things like "less than ten kilograms" are strongly preferred to "fewer than ten kilograms" in standard formal written English. Specifically, we have the following counts in Google Scholar:

 "less than __ kilograms" fewer than __ kilograms" five 153 2 ten 74 0 50 129 1 100 212 2

or for grams:

 "less than __ grams" fewer than __ grams" five 151 0 ten 117 5 50 649 6 100 876 6

In both cases, the preference is roughly 100 to 1. The corresponding facts for "__ or less/fewer" are in the body of the post.]

2. ### Lauren said,

December 5, 2010 @ 11:15 am

For the singular verb agreement, isn't that more a matter of covert meaning? That is, 3 meters is the threshold — it's not that there are 3 meters, all of which are a threshold, it's that there is a single point (3m) which is the threshold.

Similarly with X grams or less/fewer… since grams can be further subdivided, less is meaning that this is not a discrete measurement. If we were dealing with weights that couldn't be subdivided or were impractical to divide (say, atomic weight), then 'fewer' seems to be fine, although I haven't done a search for X atoms or fewer/less. (Intuitively, 'less' feels weird there.)

All in all, it seems like these situations aren't anomalous in their prescriptive forms, at all, unless the user happens to be unfamiliar with the signified.

[(myl) Unfortunately, many people are not as reasonable as you are.]

3. ### Twitter Trackbacks for Language Log » Less with plural count nouns in formal usage [upenn.edu] on Topsy.com said,

December 5, 2010 @ 11:15 am

[…] Language Log » Less with plural count nouns in formal usage languagelog.ldc.upenn.edu/nll/?p=2826 – view page – cached December 5, 2010 @ 10:03 am · Filed by Mark Liberman under Prescriptivist Tweets about this link […]

4. ### GeorgeW said,

December 5, 2010 @ 11:16 am

Are there other languages that make a similar distinction? My sense is this is highly marked.

5. ### BA said,

December 5, 2010 @ 11:17 am

Doesn't this have something to do with common math? With the inequality 1 < 2, we say 1 is less than 2 because, while those integers are discrete, 1 is less X than 2, right? A lot of those "fewer than"s or "or fewer"s sound wrong because while integers are discrete, and you can indeed have fewer than 60 grams of saturated fat, unless you're counting in whole numbers of grams, you're probably having less than 60 grams. But we also tend to round in day-to-day counts. Surely very few meanings are obscured by substituting a whole number for its closest imperfect fraction. If a height-weight rule advises a group of people weighing "fewer than 60 kilograms," is it bad English since it nominally excludes people weighing 58.8 kg? Don't we usually say those people weigh 59 kg, and they are therefore included in the target range?

6. ### KCinDC said,

December 5, 2010 @ 11:17 am

old_maltese, do you really say things like "He is fewer than 6 feet tall" or "She makes fewer than $25,000 a year" or "The town is fewer than 5 miles away"? 7. ### richard said, December 5, 2010 @ 11:28 am I do love reading these posts — but my constant companion is the question, "What does this post mean for class discussion and for marking in my first-year composition courses?" I haven't had much success introducing effective nuance on the points where LL recommends it, because my efforts have tended to generate assignments where the clear message received has been "anything goes." Of course, anything does go, and I understand what they mean, but if the goal is to produce writers capable of writing in conventional forms, and of recognizing conventionality as no more than conventionality, the "anything goes" outcome isn't exactly preferred…. 8. ### Brandon said, December 5, 2010 @ 11:49 am For the first example I didn't actually read it as described, that is: "…children who weighed [1000 grams or less] at birth." Where the less is qualifying the number of grams that the children weighed. I read it more as: "…children who [weighed 1000 grams] or [(weighed) less] at birth." Where the less is actually qualifying the weight of the child, which is a mass noun, and not the number of grams. So to me this example doesn't do much to further your point, but I could be wrong. I apologize if my formatting is confusing. 9. ### Army1987 said, December 5, 2010 @ 12:00 pm @BA: In "1 1" is "2 is greater than 1". 10. ### KCinDC said, December 5, 2010 @ 12:05 pm Brandon, how would your analysis make any difference? The less there would still need to be a comparison to something, so it'd be "[(weighed) less (than 1000 grams)] at birth". 11. ### Army1987 said, December 5, 2010 @ 12:41 pm OK, since my post was screwed up: in "1 < 2" pronounced as "1 is less than 2", I think of "less" as the comparative of "little". By comparison, "1 > 2" is pronounced "1 is greater than 2". 12. ### Helma said, December 5, 2010 @ 1:51 pm GeorgeW asked, Are there other languages that make a similar distinction? My sense is this is highly marked. Not the same issue, but something of a number conundrum in Greek and Latin: you can find neut. sing. plus / πλέον as well as plures/πλείους followed by a numeral plus count noun (plus quam tres/plures quam tres N 'more than three N'). Syntactically, you'd have to go with treatment as a comparative adverb for the singular, vs. treatment as a modifier for the ones that do show agreement. The Greek corpus offers a lot of distance measurements, but also troop/casualty counts. I'm tempted to say that the singular is more at home with the former, but really haven't done more than glance at the search results. 13. ### MJ said, December 5, 2010 @ 2:18 pm The idea that "counted units" that "are not normally divisible" can be "considered as a mass-like quantity" and its corollary–that amounts that are normally divisible can be considered as countable units–seems to me to be crucial; whether the extralinguistic referent is in fact divisible or not is not determinative. If you're treating hours as countable units, you might say "I worked five fewer hours this week than last" ; if you're thinking in terms of quantity you'll probably say "I worked five hours less this week than last." 14. ### Brandon said, December 5, 2010 @ 2:55 pm @KCinDC From my understanding the issue is that "less" is only supposed to be used with mass nouns, and "fewer" is supposed to be used with counts. So the question is whether the "less" in the first example is parsed as qualifying the count phrase of "1000 grams" or the mass noun of "weight." To me there is a distinction between the two, but this might just be because I didn't really grow up with a noticeable distinction between less and fewer, so I can't really recognize when it's supposed to be one or the other. 15. ### Paul said, December 5, 2010 @ 3:00 pm @MJ: Would your average English-speaker actually say, "I worked five fewer hours this week than last"? (Other than someone being carefully mindful of the rule, that is?) To my ear, it doesn't sound like natural speech. 16. ### MJ said, December 5, 2010 @ 4:20 pm @Paul It looks like a certain number of speakers do–a quick search through Google Books for 1900-1999 of "two fewer hours"/"2 fewer hours" through "ten fewer hours"/10 fewer hours" yields about 300 results. I don't expect that kind of construction to be as common as "I worked five hours less this week than last," but I don't see it as wrong. 17. ### Iulus said, December 5, 2010 @ 7:35 pm "When a numerically-quantified plural noun phrase refers to an amount that may be fractionally divided (grams, seconds, meters, dollars, etc.), it's generally incorrect to follow it with or fewer rather than or less. And even when the counted units are not normally divisible, but are being considered as a mass-like quantity (points, votes, cents, items, soldiers), less may also be permitted or even preferred in formal writing." Maybe I'm being a bit foolish here, but isn't it pretty impossible for a language to have naturally developed this distinction? Unless you are, in an apparently confusing way, trying to say something about a less and fewer each being exclusive to either singular or plural (semantically, not grammatically) nouns? It is also worth noting that in your response to the first comment, "In both cases, the preference is roughly 100 to 1. The corresponding facts for '__ or less/fewer' are in the body of the post," that the preference for less vs. fewer is always 100 to 1, according to quick Google searches of "fewer" and "less." 18. ### GeorgeW said, December 5, 2010 @ 8:32 pm myl: Thanks for the information about the sporting use of 'parquet.' I didn't think any self-respecting Mamma Grizzly would use words of French origin? Maybe there is a sporting exception. 19. ### GeorgeW said, December 5, 2010 @ 8:39 pm Whoops! I posted my comment (8:32pm) on the wrong thread. Sorry, this absolutely violates the relevance commandment. I confess my sin and beg forgiveness. 20. ### old_maltese said, December 6, 2010 @ 1:43 am Prof. Liberman's argument is correct. And MJ's point is good: although hours are countable units if we treat them that way, sometimes we treat them as 'a bunch of hours', or a mass quantity of time. Perhaps that's why "I worked five hours less this week than [the bunch of hours I worked] last [week]" works fine. 21. ### bloix said, December 6, 2010 @ 2:15 am If the word "fewer" disappeared tomorrow, the language would be none the poorer. The less/fewer distinction has no purpose other than to allow members of a certain in-group to recognize one another. 22. ### jc said, December 6, 2010 @ 11:38 am We might also point out to such prescriptivists that mathematicians rarely if ever use "fewer" to compare numbers. It's always "less than" or "less than or equal". They seem to see no need for "fewer" at all, except maybe as a way to add a bit of variety to informal writing. Mathematicians also normally use "greater" for the opposite comparison order, not "more". I've wondered about that. They could have saved themselves a lot of syllables and keystrokes over the years if they'd chosen "more" instead of "greater". Not that it's one of the great issues of the day, though. 23. ### Nathan said, December 6, 2010 @ 1:15 pm @jc: In defense of the mathematicians, fewer and more seem to imply some sort of concretish quantity, whereas greater and less work comfortably for any total order relation, no matter how abstract. 24. ### Ray Dillinger said, December 6, 2010 @ 6:15 pm I have recently been mildly amused at a billboard regarding some PDA or other for which, apparently, many applications are available. In Billboardese, this is rendered, "The amount of apps is staggering." Making "apps" apparently a mass noun rather than a count noun. On one hand, It is hard for me to imagine a fraction of an application as significant or useful, but on the other hand it's hard for me to imagine someone who natively speaks and writes billboardese understanding or caring about a distinction between mass nouns and count nouns. I don't think there's really a mass/count grammatical distinction in billboardese. In terms of grammar and derivation, this may indicate that billboardese could be a subdialect of headlinese. 25. ### Ellen K. said, December 6, 2010 @ 8:20 pm Seems to me that "amount" suggests there's so many you can't count. So many it's not a large number of apps, but an huge amount that one can't number. 26. ### Chargone said, December 7, 2010 @ 5:37 am if i understand that right, in 'amount of apps', that 'of' is important… it combines with 'apps' to form an ownership thing… the noun is 'amount'. the 'amount' belongs to 'apps'. genitive, adjectival phrase. amount is mass, thus what you get. also, so far as i can remember off the top of my head, i only ever here 'fewer' when there's no actual number involved. (and even then, rather than saying 'there are less/fewer people than before', lots of people will say 'the number of people has gone down/reduced' or, depending on context 'a lot of people left' of course, my experience is limited to various dialects found in the Christchurch and the surrounding area in New Zealand, so sample size is limited (but i have often noted that Kiwis dodge issues of 'correctness' by rephrasing to avoid the problem word completely, rather than adopting or rejecting 'rules' presented to them by prescriptivists or mispronouncing words they have trouble getting right.) 27. ### Chargone said, December 7, 2010 @ 5:38 am gack! hear, not here… *hunts in vane(vain?) for an edit button* 28. ### Ben Hemmens said, December 8, 2010 @ 5:27 am if you want to refer to units of something as a countable plural, then it should be "150 or fewer liters", not "150 liters or fewer", shouldn't it? One might want to say the former if something comes in liter bottles. "or fewer" is part of the number expression, "or less" isn't. Put the other way around, "150 or less liters" and "150 liters or fewer" both feel not quite right, don't they? "100 dollars, or less" can live happily with a comma. "100 dollars, or fewer" seems worse than without the comma. I think we overestimate the number of things that are inherently uncountable. Some circumstance will evetually pop up that makes them countable, and then we make them so by the appropriate structure. "So that's five coffees and two mineral waters?" 29. ### MJP said, December 8, 2010 @ 4:45 pm @bloix: "we may be fewer, but we are no less"? 30. ### Jangari said, December 9, 2010 @ 11:13 pm Clearly this has been said no fewer than 15 times, but I might just sum up the distribution: 'Fewer' is retricted to discrete countable quantities. This general principle allows for 'less' to be used with discrete countable quantities (as in "15 items or less", which in my view is totally fine) as well as continuous countable quantities (the child weighed less that 1000 grams) and uncountable (mass) quantities (in future, put less butter on the bread). But 'fewer' must only be used for the first kind (15 items or fewer) but not for the second (*the child weighed fewer than 1000 grams) or the third (*In future, put fewer butter on the bread). It appears to me that 'fewer' forms a constituent with the numeral as some kind of quantificational specifier before the whole thing forms a noun phrase: a) [less than [20 dollars]] b) [[fewer than 20] dollars] a) denotes a wider range of possibilities, including, for example$19.99, while the set described by b) has only 20 members: $0,$1, $2, …$19.

And since the grossly underweight baby has received so much attention already, try replacing 1000 grams with 1kg:
…children who weighed 1 kilogram or fewer at birth
To me this can only mean children who weighed either nothing, or 1000 grams.

31. ### Ellen K. said,

December 14, 2010 @ 4:00 pm

This is a belated reply, I know.

I find Jangari's comment, just above, that "fewer" forms a constituent with the number, makes sense. Perhaps that's why "or fewer" after the noun sounds odd to me. "3 or fewer items", fine. "3 items or fewer", wrong.