# “Tall” Tales

1. Four years ago, when she was 8, Sally was the tallest kid in her class. But she has grown only 1 inch since then, and now, as a 12-year-old, she is the shortest kid in her class. Can we describe the situation in the following ways?:

1P. Four years ago, Sally was tall.
1C. Sally is not tall now, but she was tall four years ago.

2. Sam goes to pre-school with kids his own age. Sam is the only child who is at both the morning and the afternoon sessions of the pre-school. As it happens, all the other children who go to the morning session are shorter than Sam, and all the other children who go the afternoon session are taller than Sam. It’s now afternoon, so Sam is now surrounded by kids who are taller than he is. Can we describe the situation in the following ways?:

2P. Sam was tall this morning.
2C. Sam is not tall now, but he was tall this morning.
2CF. If many very short children came to the afternoon session, Sam would now be tall.

Case 1 above is loosely based on an example Jason Stanley discusses in his Knowledge and Practical Interests. Like Jason, I’m ultimately interested in such things because of how they may effect arguments concerning “knows”. But for now, I just want to ask: Any reactions?

(See p. 64 of K&PI for Stanley’s own case. One difference from Stanley’s presentation is that I’m considering sentences in the third-person about my young characters, while Stanley considers a first-person description he considers as being said by the child in his example. I worry that when we consider sentences as said by young children, we cut them a lot of slack, and find acceptable things that otherwise wouldn’t seem quite right.)

#### “Tall” Tales — 24 Comments

1. For what it is worth, I think both the claims in 1 sound true (or at least appropriate) and all of the claims in 2 sound false (or at least inappropriate).

If I had to diagnose my own intuitions, I’d say that they are tracking something Delia Graff notes somewhere, namely that comparison classes have to be somewhat natural kinds. Being an 8 year old (or a 12 year old) is a natural kind, for present purposes, being in the afternoon preschool class is not.

2. Hi Keith,

If we consider 1P and 1C to be equivalent to the following, I think we may justifiedly believe 1P and 1C if we (at least) justifiadly believe that 1 and if we base our beliefs in 1P and 1C on 1. I also think that we may adequately assert that 1P and that 1C provided the equivalence holds and our audience shares the relevant information conveyed by 1.

1P* Four years ago, Sally was taller than everyone in her class.

1C* It is not the case that Sally is taller than someone in his class now, but she was taller than everyone in her class four years ago.

Similarly, it appears to me that the only way we can justifiedly believe in (or adequately assert that) 2P, 2C or 2CF on the base of 2 is by considering 2P, 2C and 2CF to be equivalent to

2P* Sam is taller than his morning-classmates.

2C* It is not the case that Sam is taller than his afternoon-classmates, but he is taller than his morning-classmates.

2CF* If many very short children came to the afternoon session, Sam would not be taller than his afternoon-classmates.

3. It’s hard (for me, in any case) to see the difference between (1) and (2). Sally was the tallest kid *in her class* four years ago, but that might be for the same (seemingly coincidental) reason that Sam was the tallest kid *in his morning class*. Both are relativized to classrooms of people.
On the other hand, it might not be coincidental that Sam is not tall in his afternoon class; rather, it might be that Sam is short for his age and (as might be expected) among the shorter people in his afternoon (or any peer) class. In that case 2CF would be read counterfactually, and would not (I think) affect the fact that Sam is now short.
I take it that I am supposed to read these without entertaining possibilities that are, in one way or another, irrelevant to the context. Maybe I’ve violated that restriction.

4. As far as intuitions go, to my ear, 1P and 1C sound true.
And 2P, 2C and 2CF are odd; none of them sound true.

A guess why:
I hear 2P and 2C as saying or suggesting that
Sam’s height has changed. But in these cases he has
not grown or shrunk.

5. Keith: To me, 2 seems more odd than 1 precisely because of the time difference in question. Sally can be said to be tall or not tall because of the time (4 years) over which her comparison class has changed (this seems a bit like a ‘Cambridge change’); we are used to childrens’ height changing over a period of years. (Note that the case of 1 could refer to (mostly) the same kids in her class who have grown more over the last 4 years; whereas in 2, the two comparison classes of morning and afternoon kids are actually referring to different kids who are not undergoing any changes themselves.)

But in 1, the time difference is only from morning to afternoon, and that sounds stranger. In other words, perhaps our intuitions regarding how to use ‘tall’ are linked to the duration of time which has passed.

6. Oops; obviously that last paragraph of mine is supposed to begin “But in *2*,…”

7. I agree with everyone that 1 is good and 2 is bad. And I think Matt Benton is on to something: the acceptability of 1 may piggyback on the fact that it’s not unreasonable to expect absolute height to change over that period.

This might be compatible with a view of ‘tall’ on which when we use ‘tall’ there’s a sort of pretense that ‘tall’ is a matter of absolute height only, even though it is in fact a matter of height relative to a certain class. Then we would expect conjunctions of the form “a was tall at time t but not tall at time t1” to be abominable when t and t1 are very close together, but not when they’re very far apart; which perhaps leaves room for cases like (1).

Perhaps this raises the question, on such an account why would we ever be able to say “X was tall but isn’t anymore”? But the answer might be that “tall” applies to things besides people, which may readily shrink in height (for instance, the grass on my lawn).

8. Thanks to all. 2 certainly does sound worse to me, as well. (Though even 1 doesn’t strike me as clearly fine. Upon hearing 1 first, a couple of my non-philosophical “respondents” — who don’t read this blog — didn’t like those descriptions, saying one has to put “for her age” in at the appropriate places.) An explanation for the difference may be along the lines that Brian suggests in comment 1 — though of course spelling out what’s meant by a “natural kind” in this context isn’t easy. But, supposing that’s worked out, I wonder about “comparison classes have to be somewhat natural kinds.” Have to be for what? It seems — at least to me — that the language allows us to use “tall for an F” for just about any F, no matter how unnatural. Even “tall for a thing on my desk” (Isn’t that at least close to Graff’s example?), however silly, seems allowed. One could instead say that although one can put in just about any F in such phrases where the c-class is explicit, simple descriptions like “A is tall,” governed by an *implicit* c-classes, can only have natural c-classes. But I wonder about contexts where an “unnatural” c-class has been made super-salient — like one in which F has been featured in many explict constructions. Can’t “tall” be correctly used for “tall for an [unnatural] F” then? Well, I’m not sure. Or thirdly, one can say that while “tall” can take unnatural c-classes, both explicitly and implicitly, you can only get the kind of “shift” needed to make the likes of 1-2 come out OK where the c-class is natural. (“Shift”: the value contributed by context is the one that would be supplied at the time [or, in the case of 2CF, by the counterfactual situation] being talked about, rather than that of the utterance.) Well, just thinking out loud here. If you still have your ears on, Brian, I wonder what of this sounds right to you?

9. Suppose that Ray Brown breaks into the NFL in 1986 as one of the largest linemen playing. By the time he retires in 2006, he has put on some weight, but linemen as a whole have got so huge that he is under average size for an offensive lineman.* Would it be OK to say this:

In 1986 Ray Brown was a big man, but he isn’t any more.

I don’t know. I think it would be OK to say:

In 1986 Ray Brown was huge, but now he’s undersized

but that may be partly because “undersized” may specifically mean “below average size” or “too small for current purposes, ceteris paribus.” And I don’t know whether “NFL linemen” counts as a natural kind, either.

*I don’t think this is actually true, he seems to be 6’6″ and 325 lbs. which I think is still plenty big.

10. Matt: Do you want to change your example from “was a big man” to something else, like “was big”? Explicitly putting in the c-class of men seems to rule out what your example needs to make a point: that the c-class is offensive lineman.

11. I don’t think it does rule it out; for instance, you can say “Sam is a big kid” when Sam is distinctly below-average size for the class of kids, because Sam is four and big for a four-year-old. And actually I think the example sounds worse without ‘man’, though my intuitions here are shaky. But I can hear it as OK with ‘huge’ in for ‘big man’. Imagine Ray Brown struggling to block a much larger defensive lineman. Dialogue between sportscasters:

“Ray Brown has lost a step or two since he was a rookie phenom.”
“It’s not quickness. In 1986 Ray Brown was huge, but he isn’t any more.”

Does that sound OK?

12. This may be just be me, Matt, but to my ears, the example works better w/o the “man.” With it in, supposing that Brown hasn’t gotten smaller, your original seems (at least to my ears) to suggest that men in general have gotten larger — which seems to be true, judging from the health reports one sometimes hears on the news. (I know that *I’ve* gotten bigger since 1986!)

Your “big kid” example works pretty well for me. So, at least to my ears, there’s a difference between
a. Sam is a big kid [4-y-o] and
b. Brown is a big man [NFL lineman]
in that it’s easy (at least for me) to hear (a), without the part in brackets, as being made relative to the class in brackets, but very difficult to hear (b), without the part in brackets, as being made relative to the class in brackets.

13. I’ll take a stab. . .

In judging an assertion that X is r, we invoke a c-class (sometimes explicitly, others not). Maybe we do something like:

“X is r” is true iff P(XrY| X in c-class)>n

That says something like (setting n=7/10) “Sam is tall if and only if he is taller than %70 of his peers,” where “his peers” is some c-class that speakers understand to be invoked or is explicitly invoked.

Some shifts in c-class seem expected, while some seem unexpected. The unexpected shifts (eg, those that come in quick succession) can make us uneasy, but usually that doesn’t happen. There are some collections important of c-classes, which are frequently invoked, like the collection of classes of 1st graders. Sometimes the criteria for membership in some c-class is arbitrary (eg, being on one’s desk) and sometimes they aren’t (eg, having a spine). Something about that arbitrariness has something to do with how natural the c-class is. I don’t think that I’m using “arbitrary” as synonym for “unusual” since there are probably some non-arbitrary unusual c-classes, but I’m not sure where to lay down criteria for “arbitrariness.”

To illustrate: Frequently, the c-class invoked by statements like “Sally is tall” is the members of Sally’s class. That’s who kids and parents use for comparison. So the collection of 1st grade classes is an important collection and the union over that collection (ie, the class of all 1st graders) seems pretty natural.

That class is pretty natural, but you might not be able to tell just by looking at Sally whether she’s in kindegarten, 1st, or 2nd grade. You can’t tell what year she was born in either, unless “looking at her” includes looking at her over her whole life. I can look at her now and have some idea of her stage of growth and development, so maybe that’s a more natural c-class than “born the same year.” After all, it is somewhat arbitrary when that we start a new year in January rather than June. I’m digressing, but that’s kind of an interesting tradeoff between naturalness and fuzziness in this instance, since the class of children in a certain stage of development is fuzzier though more natural–I’m suggesting–than the class of children born in the same year according to our calendar.

The union over the collection of classes of 1st graders is the same as the union over its own power set, but its the former collection that is important, not the latter. We can get arbitrary classes out of the power set. For example, we could take a kid that is tall [for a first grader] and stick him in a class with all kids that are taller than him. He’d be short relative to that c-class, but it would be unimportant. So, it doesn’t follow that all subsets of natural classes are important.

One thing that makes the shift in (2) unexpected is that at least one of the c-classes that are invoked in the respective assertions is going to have to have an abnormal distribution of heights compared to the distribution of heights in the union over all pre-school classes. If the morning class has a normal distribution of heights compared to the class of all pre-schoolers then the afternoon class is abnormally tall, and vice versa, mutatis mutandis. So, maybe it is false to say (eg) “Sam is tall” when he only turns out to be tall in a c-class C that has an abnormal distribution compared to some appropriately selected and sufficiently natural c-class N which contains C. In that case we might be inclined toward corrections like “Sam is not tall. He’s tall for his class, but is pre-school class is short.”

14. Keith,

No fair waiting until I go to Kenya to post criticisms! However, I have fortunately tracked down an internet connection in Nairobi. Suppose that Bill was bragging all morning about how tall he is (he is six foot four). Then I take Bill out on the basketball court, which is full of professional players, all of whom are taller than six foot four. It seems to me that I can easily say “You’re not so tall anymore Bill!”.

Now, what is going on with (2)? I agree that it is odd. But the case I’ve just given shows that one certainly can change the comparison class for “tall” within a day, as it were. I do take negative evidence very seriously, so I think Keith is right to worry about (2). But I thing the other cases of felicitous altering of the value of the comparison class variable suggest we should look for some more particular pragmatic explanation of the oddity of (2). I don’t yet have it, and anyway I’m going on safari now.

15. Didn’t know you were gone when I posted, Jason. (I’ve since read your post on Leiter’s blog saying you’d be gone.) Anyway, I was just trying to figure out what to say about such cases. Wasn’t ready yet to discuss your treatment — so no criticism intended. Hope you’re enjoying your trip!

Since it may be relevant to sorting this all out, it’s worth observing that characters in my story #2 can say what your character says, with the same propriety. Suppose Sam has a friend in the morning class, Sue, whom Sam teases about her height. Sue has forgotten her lunch box, and so returns to the class in the afternoon to retrieve it. Seeing Sam surrounded by taller kids, she says: “You’re not so tall anymore, Sam!”

(I’ve followed you in my phrasing. I wonder how the example sounds without the “so” in it — which is probably the more relevant test. At least to me, “You’re not tall anymore, Bill” doesn’t sound as good.)

And if we tell your story in the “mode” of my post, the result is something that sounds pretty bad, at least to me:
Bill is 6’4″, and has spent all morning bragging about how tall he is. But now, in the afternoon, he’s playing basketball against professional players, all of whom are taller than Bill. Can we describe the situation in the following way:
3C. Bill was tall this morning, but isn’t tall now

This isn’t to explain what’s going on, but this does suggest that the difference lies in such matters as what exactly is being said, rather than in such things as how much difference in time there is. (On this particular point, I take it you’d agree, Jason.)

16. (I was just kidding Keith!) I agree that adding “so” helps. But maybe 3c is odd because there is a more natural way to say it. The oddity of 3c (if it is odd) might be due to the fact that it is out of any natural context.

17. I disagree with Brian, Patrick, and Nick, etc, and agree more with Rodrigo and Mike. Taken independently of a conversation, atatements 1P and 1C are both false because of the hidden pragmatic truth-conditions which are involved in both propositions.

The word, “tall”, is always transitive. The default constant that exists in the relation is that of normalcy. So, taken on its own, 1P means “Four years ago, Sally was tall relative to most people”, and similarly for 1C. I believe this view was taken by Boguslowski.

What the speaker would want to say about Sally is that “Four years ago, Sally was tall relative to her classmates”. But that’s not what’s been made manifest by default in statements 1C and 1P.

This default of normalcy, however, can be bumped off if the conversation demands it, and the shared background assumptions are made mutually manifest.

18. Keith,

You write:

“…this suggests that the matter lies with such matters as what exactly is being said, rather than with such things as how much difference in time there is.”

This seems mostly right; that the time difference is irrelevant was the point of my first comment. But it doesn’t follow that the matter lies with what exactly is being said. If “You are not so tall anymore” and “You are not tall anymore” express roughly the same content, it might have to do with how what is said is being said (something manner-related). Maybe the anaphoric reference involved in the use of “so” is required to allow the audience to hook onto the comparison class?

Btw: Though I get the oddity in your original example, I don’t get the oddity of the versions without ‘so’ in the basketball example, and neither did my single non-philosopher informant.

19. Jason, you write,

“Then I take Bill out on the basketball court, which is full of professional players, all of whom are taller than six foot four. It seems to me that I can easily say “You’re not so tall anymore Bill!”.”

My ear might be bad, but I can’t imagine saying to Bill sincerely “you’re not so tall anymore”. I can imagine saying “not as tall as you thought you were, eh?”.
Something like that. Seems to me like his bragging was the result (in part) of a naively favorable comparison class.

20. I can also imagine something similar being said in a situation where we would be likely to deny the literal truth of what is being felicitously uttered. For instance, we are at a costume party and I am wearing a sack over my head. You are trying to guess my identity, but you think that I am out of town (that is, you would assent to “Matt Weiner is out of town”). So you say, “I know it isn’t Matt Weiner.” I rip off the sack and say, “You don’t know anymore.”

This strikes me as felicitous, but if we take “anymore” to mean “You once knew and now you don’t” most of us would take it to be literally false. (Not all! I’m reading a paper that denies that “know” is factive.)

21. Matt,

That example seems crucially different, because the conversants in your example are actively breaking semantic conventions in order to express certain pragmatic meanings, somewhat like when someone says “I don’t know nothin'” to mean “I don’t know anything”. Whereas in the Sally case, people are just filling in implicit slots and not breaking any semantic norms for pragmatic gains.

22. Three relevant facts. First, when I go to NBA games, I frequently say things like “Wow, everyone out there is really tall,” and I am referring to everyone on the court, including the guards who are 6’2. Second, when I go to a WNBA game, I also say “wow, everyone out there is really tall,” and here again, I am referring to everyone on the court, including the guards who are about 5’9. Finally, I think that if I was told that all of the kids in the aforementioned morning and afternoon classes were well below average in height, for all kids their age, then I would say that none of those kids are tall. So I think that there are only three comparison classes for all applications of “tall” to human beings: adult males, adult females, and children. Finally, then, I think that the relevant features of the context of a use of “tall” (for human beings) simply determine one of these three comparison classes. I don’t see any reason to think that the comparison class varies from context to context any more than this.

23. One qualification: there will have to be separate comparison classes for children of different ages.