## Most and Many

To continue the mostathon: when I said that most "licenses a default generalization," I meant to suggest that it has a kind of generic quality — you can't account for its use by assigning it a purely quantitative or numerical meaning (i.e., "more than half"). One way to make the point is to borrow Mark's method and look at examples where most would have been licensed but the writer chose instead to go with many. These are actually quite plentiful:

Among people who did seek out help for their depression, many (68.8 percent) saw or talked to a medical doctor or other health professional.

Many parents surveyed — 62 percent — say they've taken away their child's cell phone as punishment.

Once again, many employers (53 percent) are not yet sure which action they will take.

Why didn't these writers use most instead? It isn't always easy to say, but some of the examples are suggestive.

Let's say that many means "a significantly large number," whereas most denotes a group large enough to represent the default case — which will most often but not invariably be more than half (cf Mark's examples of most = < 50%). It often happens that the fact that such-and-such a property applies to the majority of instances isn't in itself significant — that is to say, it doesn't license any conclusion that wouldn't be licensed if the property applied only to a large plurality. In those cases many seems the more appropriate term:

Many independents — 55 percent — are extremely or somewhat likely to vote for a Republican Congressional candidate this year with the "specific intention of providing a check on Democratic control of Congress and the president.

Many employers, 42 percent, plan to extend dental plan coverage to adult children in order to match their medical plan requirements, and 32 percent plan to extend vision benefits to adult children. Once again, many employers (53 percent) are not yet sure which action they will take.

In the first example, I think it would be odd to use most (though "a majority" seems okay): the point is not that this is a characteristic or dominant property of the independents surveyed, but only that it holds of a significantly large number of them — the interest of the result would presumably be little different if the figure were 45 percent instead of 55 percent. In the second example, too, most would be odd (but "a majority" less so): the point is simply that a large number of employers are not sure what they're going to do, not that this uncertainty is a dominant characteristic of the group. In fact you sometimes find many used in place of most even when the proportion is very high, so long as what matters is the absolute number of people who have the relevant property rather than their dominance in the larger category:

Many Echo Boomers (88 percent) agree that they do not know as much as they should about retirement investing and the majority (71 percent) would like to receive more information and advice from their company on how to reach their retirement goals.

The following example gives a nice minimal pair:

In this sample of 397 homeless adults, 72.1 percent were male. The median age was 35. Most respondents (72 percent) had completed high school. Many (66.5 percent) were black.

What the writer wanted to suggest, I surmise, is that the large proportion of respondents who had finished high school allowed one to take that as characteristic of the group, whereas the fact that more than half were black didn't license a similar conclusion — the implication might have been more or less the same if only a third were black, provided that was a significantly large number. (This really is a minimal triplet: the writer could have used either most or many with the percentage of males in the survey, but that figure was presumably neither significant in its size nor characteristic of the sample of the whole.)

In this connection, by the way, I don't think Mark is right to say that the following example offers "striking evidence that most = >50% is alive and well in some quarters":

Many dancers (49 percent) were comfortable or very comfortable in revealing their occupation to others. Interestingly, most (52 percent) respondents viewed exotic dancing as a promiscuous activity.

All this shows, I think, is that the writer feels that the proportion who regard exotic dancing as promiscuous was sufficient to make that a notable feature of the entire group — if the writer had been impressed merely by the absolute number of dancers who held this view, s/he could have used many in both instances.

Qualifications: I'm not sure if "default generalization" is quite the right way to describe this. And it may very well be that the inferences associated with most can be generated by a judicious application of the Gricean maxim of quality to a purely quantitative meaning, though this account would also have to explain why most is sometimes infelicitous in contexts where "a majority" is fine.

1. ### Sandy Nicholson said,

August 1, 2010 @ 5:52 am

If many is sometimes used instead of most when ‘what matters is the absolute number of people who have the relevant property’, as in the Echo Boomers example, it strikes me as odd that the writer would then quantify their claim using a percentage in parentheses rather than giving an absolute number of people. But I suppose that could just be down to inappropriate use of statistics (which is nearly universal) rather than any kind of insight into what the writer was thinking.

Although I would have counted myself squarely in the ‘most = 50 per cent’ camp, the examples here persuade me that something more subtle is going on in many if not most cases.

2. ### SeanH said,

August 1, 2010 @ 6:44 am

I am definitely in the "most = >50%" camp. Somebody saying "most people voted for me" when 50.1% of people voted for her is trying to present the facts in the best possible light, but not actually lying (at least the way I have always seen the word used).

The following example gives a nice minimal pair:

In this sample of 397 homeless adults, 72.1 percent were male. The median age was 35. Most respondents (72 percent) had completed high school. Many (66.5 percent) were black.

Strikes me as a stylistic decision not to repeat "most", although I don't know where that's from.

"Many" is definitely relative. I'd write "many people (as many as five percent) claim to have witnessed alien activity" because one would expect the proportion of people claiming to have witnessed alien activity to be significantly lower.

3. ### Itamar said,

August 1, 2010 @ 7:20 am

I don't believe this was mentioned earlier, but there's an interesting 2009 paper by Martin Hackl tackling "most" and "the most". The first half is theoretical, the second experimental. He basically advocates the view that "most" means "the largest subgroup bigger than any other subgroup", which is why we can say "most Brits voted Tory in the last elections".

Martin Hackl (2009). On the grammar and processing of proportional quantifiers: most versus more than half. Natural Language Semantics 17:63-98. DOI 10.1007/s11050-008-9039-x

[(myl) Thanks for the reference! Note that this is consistent with the etymology (i.e. superlative of more) and is what lies behind the relevant OED gloss. Hackl treats most as (compositionally) "MANY + EST", which is not quite the same, but close.

From his conclusion:

In the case of quantification, fulfilling this obligation [to furnish the pieces that processing theories require to draw systematic distinctions that occur during real time comprehension] hinges on what type of semantic primitives one assumes for quantification in natural language. A compelling example is provided by the pair most and more than half, which are standardly treated as truth-conditionally equivalent quantifiers. I presented experimental evidence from real time verification studies that differentiates these two expressions in ways that seem to correspond to specific differences in their form – the former being a superlative and the latter a comparative expression of proportions. […]

Specifically, I offer a compositional analysis of MOST as the superlative of MANY, arguing that the proportional reading is in fact a special case of the superlative reading. Extending the analysis to FEWEST offers an explanation for a currently unexplained systematic gap in the paradigm of proportional quantifiers, namely that FEWEST cannot be used as a proportional quantifier. Such an analysis presupposes that the set of semantic primitives of quantification includes e.g. degree expressions, measure phrases, and comparative and superlative operators but not relations between sets as GQT would have it.

]

4. ### SeanH said,

August 1, 2010 @ 7:35 am

@Itamar: the example you use is interesting. Is the sentence "most Brits did not vote Tory in the last elections" true? I think it clearly is (and sentences like that are often put forward by proponents of electoral reform in this country, although at the moment, a Conservative/Liberal Democrat coalition is actually based on a majority of votes, a very rare occurrence here).

5. ### Chris Buckey said,

August 1, 2010 @ 7:58 am

I personally try to reserve "most" for what might be called an overwhelming percentage, say something like 70-80% or more. Then again I'm an academic-in-training, so I tend to try to perhaps be overprecise when I write.

6. ### theophylact said,

August 1, 2010 @ 9:45 am

"Many people" can suggest a large number, but "most people" has to mean a large proportion; certainly more than half, perhaps a great deal more.

Many Americans believe in UFOs; most believe in angels.

7. ### language hat said,

August 1, 2010 @ 9:57 am

In the first example ["Many independents — 55 percent — are extremely or somewhat likely to vote for a Republican Congressional candidate…], I think it would be odd to use most

That's presumably because you yourself do not use it that way. I do, and I find the use of most in that context completely normal. Much of the "argument" here seems to be speculation ("Why didn't these writers use most instead? It isn't always easy to say…") and petitio principii.

[(GN) Well, there are two claims involved here. All we know is that the writer chose to use many here rather than most; we assume he had a reason and speculate as to what it might be. In that sense speculation is essential to this sort of analysis. The second, counterfactual claim — that most would have been odd and not simply dispreferred in this instance — doesn't have the same kind of empirical grounding, which is why many people are generally troubled by the of question marks and asterisks in linguistic argumentation. I am too, but nonetheless find it terribly useful. When I read, e.g., "Most independents are closet partisans," I assume a substantial majority and not a narrow one. But in other contexts "most independents" seems okay with a narrow majority (e.g., "most independents don't vote.") For me, this hinges on the question of whether a proportion much below 50 percent would substantially change the significance or interest of the claim. My guess is that that's the root of the disagreement over judgments in these cases. The crucial point is that the choice of most or many isn't simply a function of the size of the group — it depends on the particular claim being made.

8. ### Ran Ari-Gur said,

August 1, 2010 @ 11:09 am

I agree with Language Hat. Search results are great, but they're not really telling us anything we don't already know from all the comments: some speakers take "most Xes" to mean "a majority of Xes", some speakers take it to mean something stronger, and some take it to mean something even weaker, roughly "a plurality of Xes". (You can count me in the second group, by the way. "Most people are men, but most Americans are women" sounds ridiculous to me.)

9. ### bloix said,

August 1, 2010 @ 11:46 am

The ambiguity is frequently exploited by hack political commentators and the like, who will write a literally true statement that "most" whatevers [people, dollars, events] are something-or-other, meaning 50% plus one, and go on from there as if they've demonstrated that virtually all whatevers are in that category.

10. ### Ron Stack said,

August 1, 2010 @ 12:53 pm

I hadn't thought about it before but I think I'm in the "dominant property" camp. In the example of the voting intentions of independents the point that "the interest of the result would presumably be little different if the figure were 45 percent instead of 55 percent" is telling.

I also think the writer's (or speaker's) attitude toward the property in question is relevant. If the "most" and "many" in the example about homelessness were switched, I think the passage would not only take on a different meaning but also would invite inferences about the author's point of view.

11. ### Rodger C said,

August 1, 2010 @ 12:57 pm

I think I see from all this what happened to Dubya in 2004: he got 51% of the vote and then got it into his head, "Whoop-de-doo! Most Americans want me!"

12. ### groki said,

August 1, 2010 @ 1:17 pm

SeanH: Strikes me as a stylistic decision not to repeat "most"
yeah, in the "many/most/majority" examples, the word choice seems partly driven by a dont-be-boring constraint, not just by the literal* semantics–which complicates the analysis. otoh, maybe that is helpful: having to think more carefully about word choice might throw the semantics more fully into relief.

for me, "most" = ">50%" is not lying, but does glance in that direction: a less extreme version of calling a within-statistical-error majority a landslide.

actually, I have two meanings at least competing: "quite a bit more than half" (the "real" meaning) and "anything over half" (technically correct, but a little shifty). could these be homonyms with very close overlap?

(*as opposed to all those other semantics, I guess. :)

13. ### groki said,

August 1, 2010 @ 1:22 pm

and Rodger C beats me to a punchline.
harrumph, I need to refresh more often!

14. ### James Kabala said,

August 1, 2010 @ 2:36 pm

Rodger C.: Well, if we leave aside the issue of non-voters, most people did want him. And in 2008, most people wanted Obama. In 1992, 1996, and 2000, when no candidate received a popular vote majority, no one could claim the "most" banner.

I said below that "more than half" or "a majority" was probably a better stylistic choice in close cases, but Rodger C.'s joke and groki's claims about "glancing in the direction of lying" and "shifty" really startle me. I had no idea there people who believed so strongly that the correct meaning is misleading.

15. ### Ellen K. said,

August 1, 2010 @ 3:15 pm

It seems more than a bit arrogant to assert that one's own understanding is the correct one, and the others wrong.

16. ### Jonathan said,

August 1, 2010 @ 3:37 pm

If I have 20 students in my class, and 11 pass and 9 fail, it seems perverse for me to say that "most students pass my class." I am in the camp where most has to mean all but a small percentage, not 50% + 1. My statement would be fine if I failed 2 and 18 passed, for example. A good test is to see whether the statement holds up if you give an actual percentage, and the word "most" is either revealed as a distortion or holds up as pragmatically valid.

17. ### James Kabala said,

August 1, 2010 @ 4:23 pm

Well, all the dictionaries (except Encarta) are on my side. I am sorry for my wording, but since I never heard of this alternative meaning until yesterday, I hope you can understand my surprise.

18. ### James Kabala said,

August 1, 2010 @ 4:24 pm

I also think that words like "lie" and "shifty" are far harsher than "incorrect."

19. ### sarang said,

August 1, 2010 @ 4:55 pm

I think I completely agree with the "licensing a generalization" idea. I hear "most people X" as "it is safe to assume that a random person Xes." Of course there's a lot of disagreement as to what licenses a generalization. It seems to me quite misleading to generalize from a mere plurality, regardless of how small the other groups are, but I don't know to what extent this is a _linguistic_ issue rather than a broader cultural one. Does one's interpretation of "most" correlate with (say) one's reactions to pop psychology?

20. ### a said,

August 1, 2010 @ 5:09 pm

@GN and @languagehat

I also agree with languagehat. The problem I see with GN's analysis is this. He says:

Well, there are two claims involved here. All we know is that the writer chose to use many here rather than most; we assume he had a reason and speculate as to what it might be. In that sense speculation is essential to this sort of analysis.

I think this is a false choice. Why should we assume that when a writer uses a word, s/he is always choosing between some minimal pair? I don't think it's warranted to say if a writer uses the word 'many,' s/he has (actively) decided not to use 'most,' or vice versa. I think there are a number of more likely possibilities: the writer just happened to think of the word 'many,' and wrote it, without ever thinking that 'most' might also be suitable; (as in one of your examples) the writer just used the word 'most,' and to avoid repetition, uses 'many,' (this is a stylistic phenomenon, not, IMO, a linguistics one). There are probably others I haven't thought of. My point is that I don't think it's reasonable that whenever someone uses a word, we can conclude that (a) that was the exact word they meant to use; (b) they necessarily have a good reason for using that particular word and not a close 'synonym.'

GN: Well, I don't know how that "actively" crept in before choose. And it's very often hard to know why someone chose one course rather than another. But philosophical considerations aside, here's an experiment: Do a search over your own writings and find all the places you've used many where you could logically have logically substituted most, and ask yourself in each case what difference the substitution would have made, if any. Then vice-versa for the mosts that could have been manys. And then focus on the cases where the replacement would leave you dissatisfied.

August 1, 2010 @ 6:55 pm

Itamar: "which is why we can say most Brits voted Tory in the last elections"

But we can't say that! Who does say that? It's not true. It would be true to say the Tories got most votes. But not (?) that most votes went to the Tories. Now my brain is starting to hurt…

Chris Buckey: "I personally try to reserve "most" for what might be called an overwhelming percentage, say something like 70-80% or more. Then again I'm an academic-in-training, so I tend to try to perhaps be overprecise when I write."

But if you use "most" to mean "an overwhelming percentage, say something like 70-80% or more", when it doesn't actually mean that, you're being underprecise when you write.

22. ### Rodger C said,

August 1, 2010 @ 9:07 pm

It had also never occurred to me until this series of posts that "most" might mean anything than "over half." My joke (or conjecture) was based on the idea that GWB might have one meaning of "most" and, say, Cheney and/or Rove the other, so that if one of the latter kept repeating the "most" notion … Whatever.

23. ### James Kabala said,

August 1, 2010 @ 9:17 pm

Rodger C.: Then we're on the same page. Sorry!

A more substantial point: It seems to me that "most" might be used more often for a narrow majority when the outcome is surprising or unexpected.

Thus, while I find it completely accurate and not at all "perverse" to say "most students passed" when 11 of 20 did, I could see how someone (like the hypothetical Cheney described by Rodger C.) could use the word "most" to create a misleadingly positive assessment, and I might use a more precise phrase such as "a bare majority passed." If 11 out of 20 flunked, however, I would have no hesitation at all about saying "most students flunked."

This doesn't really reply to the original Irving interview, though, since his interviewer described a surprising situation and Irving rejected the word "most" anyway.

24. ### Neal Whitman said,

August 1, 2010 @ 11:32 pm

There was an interesting talk on precisely this subject at LSA 2010, by Stephanie Solt. The slides are available here. And here's the abstract, taken from the conference program:

Stephanie Solt (Zentrum für Allgemeine Sprachwissenschaft) Session 6
On the expression of proportion: Most and more than half
While most and more than half are often treated as equivalent, the true picture is actually more complex. This paper draws on corpus data to demonstrate fundamental (and previously unrecognized) distributional and interpretive differences between the two. These are shown to derive from a basic distinction in how proportion is expressed: more than half expresses a comparison between numbers, and is explicitly based on counting or other form of measurement, while most expresses a comparison between sets, which may – but need not – involve precise counting of set members. Connections are made to recent findings on the
psychology of number cognition.

GN:Thanks for this. The slides are very clear and thorough. From her summary of the corpus data:

Most and more than half are used to convey distinct ranges of proportions
Most yields a generic interpretation in contexts where more than half is infelicitous or has a ‘survey results’ interpretation
More than half (but not most) requires an enumerable domain and a precisely defined predicate
Use of more than half (but not most) is typically supported by numerical data (count/survey/ analysis)

25. ### Neal Whitman said,

August 1, 2010 @ 11:37 pm

All this discussion of most and many prompted me to blog about an ambiguity of more I'd been wondering about: If the server brings you one wedge of lemon with your tea, and then you ask for "more lemon," there are two interpretations: Bring another wedge of lemon, or bring at least two wedges of lemon. Whence the ambiguity?

26. ### grackle said,

August 2, 2010 @ 2:38 am

I don't see anyone mentioning the pairing I assume of most and few. If I am doing a repetitive task, like grading 80 exams, for example, I would never think that I had completed most of them when I had reached 41; it would be only when I had just a few more to grade that I would be confident that I had completed most of them. By the same token, if I had completed 35 or 40 of them, that would be more than a few but still far from most.

27. ### Nick Lamb said,

August 2, 2010 @ 6:26 am

"All we know is that the writer chose to use many here rather than most"

This sentence is silly, we don't know that at all. When you look at the quotes in this present context you (Geoff) see each writer as having chosen many rather than most. But the writer did not share your context and may never have made such a choice. I am quite certain I have written both "many" and "most" in sentences without considering the other as an alternative.

A philosopher is walking through the picturesque centre of an English industrial town. He sees a teenage girl in a pink dress. "Why did she choose to wear pink rather than blue?" he muses briefly. But the premise of his question is mistaken, the girl attends a high school with a uniform policy. The uniform requires all girls to wear green, and so (inevitably) out of school the girls never wear green. The dress was chosen because it isn't green, not because it is pink.

[ We see this mistake in attempts to understand voting patterns. The analysts tend to shortcut from actual votes cast to support of a political party. But this assumes a fact not in evidence, that the votes are made in support of a political party. In practice the electorate may (and often do) vote for a candidate they've heard of, the first one on the list, one with the least "foreign sounding" name, the one with a yellow logo etc. ]

28. ### Moritz said,

August 2, 2010 @ 8:07 am

The writers might not have made a conscious choice to go with "most" instead of "many" (or vice versa), but they did go with one or the other. Unless you are assuming that the alternative choice simply wasn't available to them, which seems very unlikely. Of course there could be any number of reasons for choosing "most" or "many" besides quantitative ones, e.g. stylistic choice in order to avoid repetition. The analogy you make is cute, but doesn't seem to fit very well.

29. ### language hat said,

August 2, 2010 @ 9:48 am

The writers might not have made a conscious choice to go with "most" instead of "many" (or vice versa), but they did go with one or the other. Unless you are assuming that the alternative choice simply wasn't available to them, which seems very unlikely.

And you are assuming that there are only two "alternatives," which is simply wrong. There are "a lot," "more than half," "a great many," "a majority"… as many as you like. As several people have pointed out, it is artificial and misleading to state that the author "chose" between two "alternatives." If that was how we wrote, we'd never write anything. (Should I say "If" or "Assuming"? "was" or "were"? "how" or "the way"? "we" or "one"? "wrote" or "composed"?…)

30. ### mollymooly said,

August 2, 2010 @ 2:49 pm

OK, then. There is "majority": the boring, apparently older, meaning which is in the dictionaries; and there is "great majority": the other meaning, which is interesting because it is vaguer, pragmatically conditioned, and poorly documented.

From a practical viewpoint, the important thing is to be aware that "most" is now officially ambiguous; so we should use it and interpret it with caution, whichever our own favoured interpretation might be.

From a linguisticky viewpoint, an important first step before studying the properties of the exciting "great majority" meaning is to remove all data from sources who have the "majority" meaning. How you do this I dunno. Maybe throw a few grad students at it.

31. ### Alexander said,

August 2, 2010 @ 6:53 pm

Jeffrey Lidz, Tim Hunter & Justin Halberda (2009). The Meaning of 'Most': Semantics, Numerosity and Psychology. Mind and Language 24 (5):554-585.

Lidz, J., P. Pietroski, T. Hunter & J. Halberda (2009) Interface Transparency and the Psychosemantics of 'most'. Article in press. Natural Language Semantics.

Among other things, Lidz and colleagues have psychophysical data showing that people, when shown an array of blue and yellow dots for a very brief time, and asked to respond immediately, Yes or No, whether most of the dots are blue (or yellow, etc.), consistently respond as if the question is whether >50% are blue (or yellow, etc.). They do not demand a big fat majority. This sort of data is interesting, in being different than either usage data from corpora, or data from responses to pragmatically contextualized questions.

32. ### Mfahie said,

August 2, 2010 @ 7:20 pm

This is what I love about language log – learning things I didn't know about practical language use in the real world. I don't know if it's because I'm Canadian, but "most" to me means a great majority. If you asked me to be specific, I'd say at the bare minimum 75%. But usually even more. That sentence earlier "most people are male" is completely ludicrous to me. And to claim most as a 55% majority seems disingenuous at best. But now I will try to have a more open mind about it!

33. ### Jonathan said,

August 2, 2010 @ 9:49 pm

I'm pretty sure I use most with both meanings, depending on the contexts. That is, in some contexts I would avoid it for 55%, but in others it would be fine.

34. ### Zatch said,

August 2, 2010 @ 9:58 pm

I fall into the "great majority" group; grackle's example was particularly striking to me. But mollymooly and Mfahie have an excellent point – as these comments have illustrated, it's important for some of us to realize that the use of 'most' to mean simply 'more than half' is not dishonest if that is how the speaker conceptualizes it, while others should take into account that they may be unintenionally inflating figures for some interlocutors if they use 'most' in cases of a plurality or a narrow majority.

As for GN's homeless example, I think 'minimal pair' may have been too strong a term, since it wasn't a binary choice. That said, any time we speak, or write, we are making a choice of what words to use, based on our understanding of language. It may not be a conscious choice – most of the time, we use the first word that comes to mind. But the first word that comes to mind to convey a particular concept is highly relevant to a discussion like this, and overall I think the example was a good one. Even if the author was deliberately avoiding repetition, there was a choice involved of how to do so.

35. ### groki said,

August 3, 2010 @ 8:02 pm

mollymooly: OK, then. There is "majority": the boring, apparently older, meaning which is in the dictionaries; and there is "great majority": the other meaning, which is interesting because it is vaguer, pragmatically conditioned, and poorly documented.

well said!

36. ### Teuxe said,

May 23, 2012 @ 8:25 am

Interesting discussion. As a native French, I would make the analogy between the English terms "most" / "many" and the French translations "la plupart" / "beaucoup".

– "la plupart" comes from "la plus grande part" i.e. if you take a set of people/items/(other) and categorize them into subsets, the biggest subset will be "la plupart" : "most".
– "beaucoup" is not taking other categories into account ; it only considers that an important part of people/items/(other) is classed in this category : "many". In terms of etymology, "beau coup !" is composed of "beau"=beautiful => great, and "coup" as a part of a whole (to be thought as "couper"=cut, "coupe"=cup, thus "boire un coup"=have a drink, etc.).