Against the “For an F” Myth: Drafts of Papers on Gradable Adjectives

I have a draft of a paper on gradable adjectives — currently entitled “Gradable Adjectives: A Defense of Pluralism” — on line here (pdf). It’s largely an attack on what I call the “‘for an F’ myth” (or “Implicit Reference Class Theory”), advocating instead a “pluralist” account of the semantics of GAs, with some remarks about how this all may affect the debate regarding epistemic contextualism. Comments are welcome.

Kent Bach, who was kind enough to send me some comments on my paper, also recommended to me a new paper by Chris Kennedy: “Vagueness and Grammar: The Semantics of Relative and Absolute Gradable Adjectives.” Kennedy is still revising the paper, but a draft is available on-line here. I haven’t yet read Kennedy’s paper myself, but the title of his section 2.4. — “Eliminating comparison classes” — makes it sound as if he may be pushing here in at least roughly the same direction I am — though “eliminating” at least sounds more radical than I’m proposing. (But maybe that’s just eliminating c-classes from some exaggerated role they’re often assigned.)


Against the “For an F” Myth: Drafts of Papers on Gradable Adjectives — 22 Comments

  1. Pingback: Certain Doubts » The (Mostly Harmless) Inconsistency of Knowledge Attributions

  2. Hi Keith,

    Thank you for posting this unsurprisingly excellent paper.

    I have a question about how you’re understanding “group-relative” (as opposed to “group-indexed”) standards. I’ll pose my question by considering an example that you mention from Stanley’s book:

    (30) That butterfly is large, but that elephant isn’t large. (Stanley 2005, 134-5)

    According to you, an asserter of (30) might be using “large” to mean, roughly, large for a thing of its kind. (And so such an asserter can truthfully assert (30) without having to change standards for “large” mid-sentence, and without having to find a butterfly or an elephant that would make it into the Guiness Book of World Records.) (30) is true iff that particular butterfly is large for a thing of its kind, and that particular elephant is not large for a thing of its kind. Right?

    And now my question is: For each of these particular creatures (the particular butterfly that’s being indicated and the particular elephant that’s being indicated), which of the many kinds to which it belongs is ITS kind? Is it the genus? the species? the sub-species? the current generation of the sub-species? Does it depend upon how the thing is described in the assertion? For instance, would it make a difference if, instead of asserting (30), the asserter had said “that animal (pointing to the butterly) is large, but that one (pointing to the elephant) isn’t”?

  3. Right?

    When group-relative standards are in play, what context supplies — in addition to that the standard is a group-relative one — includes what the relevant kinds are (though this can often be left rather vague). As long as butterflies and elephants are the relevant kinds for these two animals, as seems natural to the example, (30) gets handled. Esp. in the mouth of a young child, but probably for many uses by adults, too, the relevant kinds in such cases are not as specific as species, but consist of groups that, for instance, all fall under the common word “butterfly”. But for other uses, types can be individuated by species, or in many other ways.

  4. I learned alot from the paper. In particular, the point about changing the threshhold percentage within a single c-class was something I had overlooked in thinking about this before. It seems right to me that this disposes of strong IRT pretty decisively.

    The director though: Why isn’t her c-class the class of props reasonable for the location in the scene? When she says that by “I want something tall” she means “I want something around 15 or 16 ft” she doesn’t mean that anything that height would satisfy her want, so in either case there seems to be a class of objects that is implicitly understood by everyone. If someone brought a 16 ft greek collumn into a small town 1950s scene, they would not have satisfied the want expresed in even the latter assertion.

    Suppose that within the class of props reasonable for the location in the scene the available 15 or 16 ft props are actually in the shorter half. Suppose she tells her assistant “I want something tall over there” and nothing else. The assistant returns with something 30 ft tall that is in the taller half. The director is mad and says “no, I wanted something 15 or 16 ft.” Wouldn’t it be reasonable for the assistant to later complain to her coworkers that the director doesn’t say what she means?

    In any case, when we ask her to clarify and she says “something 15 or 16 ft.” it’s more appealing to me to think that she’s said something new that more accurately represents her wants than to say that she’s given an account of the meaning of “tall” in what she had just said.

    I guess what I’m pressing for here is that one could defend IRT by arguing that when people use a word like “tall” to attempt to indicate that the object satisfies a tape-measure standard that they are communicating poorly. If you mean to use a tape-measure standard then you should say what you mean and give a rough estimate in familiar units of measurement.

    Also, this example from a footnote: “All I was claiming was that I know it… by any reasonable standard.” If such a claim is possible, then there should be some common features shared by all reasonable standards for knowledge against which we could judge the claim. Maybe that’s what invariantist epistemologists are looking for: a general account of “reasonable” standards for knowledge ascriptions.

  5. Thank you for the work. I’ve learned quite a bit.

    The argument in this paper seems reasonable enough on its surface. After all, to take a more general tack: what is “precision”, in general, except the accurate interpretation by a hearer of the speaker’s intentions? And what could invite pluralism more readily than the ad hoc intentions of the speaker?

    It has been my understanding that a thing is vague only if it fails to satisfy some conversational goals, which (if we are being felicitous) break by default in favor of the speaker’s intentions. So sure, sometimes we use comparison classes, or further specifications (as with the poor epistemic uses of ‘know’), or what have you, and sometimes not at all. But this is not an issue of comparison classes alone, or gradable adjectives alone; it is an issue of conversation goals in general, which in principle may include implicit conversation classes, and may not. (By default, I maintain, the interpreter wants to attribute a comparison class to an adjective which is typically used in a relational way, and will likely want to do it on the basis of norms. But this is not an absolute rule.) Thus, (contrary to, say, the views of Chris Gauker), the intentions of the speaker are the only necessary implicit factor. The competent listener’s sense of “closure”, mentioned in Matt W.’s recent paper, may be the stamp of precision.

    However, this insight may be revealing when we put it to work in the Director case. When no comparison classes are presented to the listener in the case of a (typically relational) adjective, they have reason to believe that they have not understood the speaker well enough — that the speaker has not been precise. We should not be surprised if the stagehand may respond with puzzlement to the director’s command; and if they did respond by fetching a tall plastic tree (or whatever), they would likely keep their eye out for the director to show pleasure of displeasure at a particular choice they made. What’s going on in this situation is that the stagehand would get more and more of an idea of what the director’s intentions were wrt a fuzzy, and emerging, comparison class.

    The point is that the listener requires certain things for interpretation and closure. This may very well be a fuzzy class which is shaped by successive approximations generated by ostension. (“Is this tree one tall enough? How about this crane? How about Godzilla…?”). And that’s plausible. This may show that your argument against the weak thesis is too hasty.

    But admittedly, this analysis may give counterintuitive results in the Grandpa case. It would mean that the grandfather, who has forgotten the age and height of his grandson, would be being perfectly precise if he were to say, “My grandson is in the 98th percentile for children his age, whatever his age is”, so long as he feels unambitious in the truth-specificity of his communicative goals.

  6. Very interesting paper! I think DeRose is right that the movie director, who says, “I need something tall over there on the left, to balance the shot” might not have a particular comparison class in mind. “Tall for an object on the set”, for example, won’t work because there might be much taller objects located in the center. DeRose suggests that if the movie director is asked what she meant, she might mention a “tape-measure standard” (“something about 14-16 feet tall”). However, isn’t it still true that an object is tall only if it is tall relative to something else that is contextually salient? Suppose the average height for a human were 20 feet (houses, cars, chairs, tables etc. would be of equal size). Then it clearly wouldn’t make sense for the movie director to say “I need something tall” (if what she wants is something about 14-16 feet tall).

  7. Brit,

    It may be that the director is just being vague until they either specify a comparison class, or engage in ostension (which is a suitable substitute for the provision of a CC).

  8. Yentz, I guess the point I was trying to make was that when the movie director says “I need something tall over there on the left, to balance the shot” she might mean that she needs something that is tall for an ordinary medium-sized object, or that she needs something that is tall compared to a human being. In other words, I am not fully convinced (by that particular example) that there is a need for a “tape-measure” account of gradable adjectives. This is not to say that I endorse a standard contextual semantics for ‘tall’, ‘rich’ or ‘huge’. As Egan, Weatherson, and Hawthorne point out (“Epistemic Modals in Context”), it seems that Ant Annie (who is 1 cm) could truly say about the shortest NBA player ever, Muggsy Bogues at 160 cms, that he is huge. In that case what makes it true that Muggsy is huge might be that he is huge compared to Annie (no implicit reference to a comparison class).

  9. [This is directly prompted by the comment immediately above, but is also relevant to several of the other above comments.] On the movie director, I guess I just don’t understand where this unrelenting drive to squeeze such cases into the c-class straightjacket is coming from. (I’m a little extra-suspicious of using c-classes like ordinary medium-sized object — things of the “tall/short/big/small/medium/etc. F” variety. (“He’s tall for a short guy”?)) It’s easy to imagine the case so that director means something stronger than “tall compared to a human being.” In fact, as I was already imagining the case, she meant something stronger than that. A 12′ tall item just wouldn’t do, though that would certainly meet the “tall compared to a human being” standard. And she may well mean something stronger than “tall for an ordinary medium-sided object.” I’m not sure whether being 10′ tall counts as being “tall for an ordinary medium-sized object,” but I at least suspect it does. Suppose the director is with me on that issue. Then in looking for something that has to be still significantly taller than that to count as “tall,” she doesn’t just mean “tall for an ordinary medium-sized object.” Yes, one can strain to find a c-class. Maybe she means to be using “tall for an ordinary medium-sized object,” where to count as “medium-sized,” an ordinary object has to be something like between 4 inches and 16 feet tall, and she’s looking for something in, say, the upper 5% of objects in that class. But that really is straining. If you ask her, she’ll just say that she meant something like 14′-16′ tall, and, as I started this comment, I guess I’m just not seeing why we would strain so hard to find some way of forcing that into a c-class standard.

  10. Keith,

    I’m largely motivated by a desire to defend empirical “idea theories” of lexical meaning in at least its weakest forms. That seems to motivate a defence of comparison classes as playing some role. But as you indicate, “tape-measure” standards are also plausible.

    Anyway, the point is that, according to the kind of hard-nosed pragmatics standard I endorsed, the speaker is being too vague for us to legislate the matter properly. Asking us what so-and-so really meant by their vague use of language (as in the director case) is effectively like asking us to describe the face of the invisible man.

  11. But the director’s use of “tall” doesn’t seem problematically vague. In fact, her use is quite precise compared with most uses. When asked what she means, her answer is that she means “about 14 to 16 feet tall.” In a typical use of the word — like when I claim, “Yao is tall” — the speaker typically isn’t in a position to, if asked, specify his meaning in such a way as to make clear what would and wouldn’t count as “tall” acccording to his current use with nearly as much precision as the director.

  12. Keith, I agree that we shouldn’t force it into a c-class standard. But I wonder whether “I need something about 14′-16′ tall” could really be the semantic content of “I need something tall”. How could I possibly grasp the proposition expressed (without asking the movie director)? There are no contextual clues that I can think of. “I need something about 14′-16′ tall” may well be what she meant, but I am not sure I see how it could serve as the semantic content. Suppose I say “I feel like something sweet”. When asked to be more specific, I might reply: “something chocolaty. Maybe chocolate chip cookies”, but the latter is not the content of “I feel like something sweet”. Your case is obviously different but couldn’t the movie director’s reply be seen as a clarification of her earlier remark?

  13. There seems to be no special problem here. The director can say that’s what she means. Or it can be tolerably clear from context — esp. to someone used to working with the director on such things. Same ol’ stories as how we ever know what someone means on a particular occasion by use of context-sensitive language. Yes, the director’s reply could be seen as a clarification of her earlier remark, specifying what she intends and intended to mean by it.

  14. Keith,

    That’s where Jeremy and I seem to part ways with your analysis. In the utterance “put something tall there”, prior to her specification of a tape-measure standard or c-class, the director really is being vague. However, it’s problematic (for me) because it fails to meet the pragmatic criterion for lucidity: we can expect the stagehand (the listener-interpreter) to be confused, and to make tentative choices until further instruction is provided. That means that communicative goals are lacking in some important way, and so, there is vagueness. Note, though, that this is what *we* would expect, because thought-experiments tend to begin with all other things equal. On the other hand, if, as you now indicate, the stagehand has extra knowledge of how these sorts of requests play out with this particular director, then that changes the matter entirely.

    If you compare this to ordinary uses, like “Yao is tall”, there’s little difference. The vagueness in both cases depends on whether or not the listener’s interpretive expectations have been met. If Malcolm says “Yao is tall” to Jean-Luc, then in a crucial way, if Jean-Luc fails to understand the implicit c-class or tape-measure standard, and he has reason to believe that such a standard is required, then Malcolm’s usage has been vague.

  15. No, she can’t say here that the 17′ tall object “isn’t tall.” If, as she says, she’ll accept something in the 14′-16′ range as “tall” she has to accept 17′ as “tall” as well. (Of course, she can reject the 17′ object as what she was looking for for various reasons, incl. that it’s “too tall.” She just can’t deny that it’s “tall.”)

  16. Hi Keith, a quick question about Brit’s example. Can the director correctly complain, when presented with the 17′ tall object, “That’s not what I asked for!”

  17. Jon & Brit: I was imagining that the director wanted something around 14′-16′ tall, not significantly more or less than that. It’s fine by me if you want to instead imagine it the other way and suppose that what she wants is anything that tall or taller — nothing seems to be riding on that difference in the case. Imagining it my way (she doesn’t want anything taller than that), if she’s given something taller than that (and yes, of course, this applies to 18′, 19′, etc., etc.), then, as I wrote in comment 17, she can’t complain “No, that isn’t tall.” She should instead say something like, “No that’s not what I’m looking for. That’s too tall.” The content of her use of “tall” is at least 14′-16′ tall, even if what she wants is (as I was imagining it) something 14′-16′ tall and no more. Can she complain “That’s not what I was asking for”? Answer that either way you want, so far as I’m concerned. She said she wanted something “tall,” and she was given something that is “tall” by the standards that governed her use. So she was given what she was asking for. But, though she was given what she literally asked for, she also wanted something that wasn’t taller than 16′. Whether she can complain in the way proposed depends on the legitimacy of complaining in that way if you’re given something that meets the specifications you gave, but isn’t everything you want. I have no horse in that race, so far as I can see.

  18. Keith,

    What benefits does your account have over rival ones? Do you disagree with the kind of pragmatic formulation outlined? If so, is the feeling of satisfaction that the stagehand has (that they have understood a meaning lucidly) the product of something especially linguistic (i.e., to do with the word “tall”), or of some other extra shared knowledge?

  19. The whole paper’s about the advantages over IRT; can’t really summarize that all here. Don’t know enough about the pragmatic formulation outlined to have an opinion.

  20. Right. But my concern was with the notion which seems most intuitively appealing, outlined in post 5. It seems capable of producing opposite conclusions to yours. I probably rambled, so I’ll repeat the conclusions here.

    An utterance is lucid only if it satisfies the conversational goals of the speaker and the goals of the interpreter. If the listener is being felicitous, their goals are to pursue what the interpreter believes are the speaker’s intentions. An utterance is vague if it is not lucid (caused either by failure to form appropriate goals, or failure to satisfy them). [A correlary assumption would be that there are no vague words, only vague uses.]

    It seems to me that all other things equal, we can expect that interpreter in the Director case would not be able to understand the intentions of the speaker; thus, the utterance would be vague. I just have a hard time understanding how this could not be.

    In addition, part of the account I provided in that post was the idea of an emerging comparison class, which is defined by acts of ostension.

    That should be enough here to supply at least some intuitions which you may respond to.

  21. Keith, I have a different sort of question as well. To what extent is your pluralistic account still a “standard” contextual account? I guess what I am wondering about is whether you would grant that a tallness or richness ascription (or the like) might, relative to a context, express a minimal proposition (e.g. the proposition that John is rich)? The meta-linguistic truth-conditions for such minimal propositions might still depend on the speaker’s intentions (as suggested by e.g. MacFarlane).

Leave a Reply

Your email address will not be published. Required fields are marked *