Kai von Fintel points to a detailed obituary of Donald Davidson in the Guardian. This goes into quite a bit of detail about Davidson’s work, and is I think a good tribute. (Although I was surprised to see the Locke lectures, which Davidson gave in 1970, described as ‘then-prestigous’. I didn’t understand the ‘then’.)

The Guardian says that there’s lots of work by Davidson still to be published, so there’s lots of papers to look forward to.

I also thought it was worth noting a comment from Richard Rorty in the NY Times obituary:

I don’t think there’s anybody else among American philosophers, after the deaths of Quine and Rawls, who gave him any competition.

I’d have added Lewis to that list, but Rorty’s point about how good, and important, Davidson’s work is seems right.

Josh Parsons has added a long comment to my (excessively snarky) post on perdurantism and semantics that is worth reading.

I know raising can account for lots of odd locutions, but even so I thought this was not well put:

The president and his aides don’t speak untruths because they are necessarily people of bad character.

(It’s from Josh Marshall’s article about Bushian lies.)

I very much doubt they are necessarily people of bad character. I’m pretty sure, in fact, that they are contingently people of bad character. The ‘necessarily’ needs to move at least in front of ‘because’, and I doubt familiar movement rules will take it that far.

I had this on the list of rhetorical questions to ask about time travel into the future, to try and get the students to think about what we do and don’t want to count as time travel. The earlier questions were Is waiting for time to pass time travel? and Is sleep time travel?. (If yes to the latter, the follow-up might be Is a bottle of gin a time machine?, but I don’t think I’ll take that path in my first year class!)

At first I thought the answer to each question was obviously no, and I still think that for the first two. But I’m not so sure about hibernation any more. The worry is that time travel is, as Lewis says, where personal time and external time don’t match up. Now there’s a few ways to measure personal time. If we just do it psychologically, then dreamless sleep will count as time travel, because no psychological time is passing. That’s clearly wrong, so presumably we should use some physiological measures of personal time as well. (Lewis mentions hair growing as one criterion of the passage of personal time.) But hibernation, at least as I understand it, also involves the slowing down of physiological processes.

If a bear wore a wristwatch, that wouldn’t slow down during hibernation, but we don’t want to make watches that important do we?

At some level, maybe cellular maybe lower, processes don’t slow down during hibernation. Maybe that’s what is relevant to time travel. But now I worry we’ll rule out some kinds of things we do want to count as time travel. Is H.G.Wells’s machine not really a time machine (when moving forward) unless the electrons in the time traveller’s body slow down? Perhaps, but that seems a much stronger condition on forward time travel than I’d have thought necessary.

I was following my links back and forth and ended up at Kaye Trammell’s metablog. There’s lots of neat stuff there if you want more of an insight into how blogs can work than you’d get following around the academic/political links I normally post.

I really liked her little post on thinking in terms of blogs. It reminded me of some of the ways this little page has changed my way of philosophising.

One of the changes is potentially unfortunate. I’ve really cut back on how much technical work I do, largely because it’s so hard to use symbols in HTML. It’s not impossible, it’s just hard, and that’s clearly had a motivational effect. I can still do some technical work – I was fairly pleased with the way I was able to use the canonical model for KT to build a semantics for truer that brought out some small differences between my approach and the supervaluational approach – but I’m doing much less than I used to be. And my technical skills are pretty clearly degrading as a consequence. (In the first logic class yesterday I completely forgot how to prove that the set of sequences of integers is the same size as the set of reals, and had to rearrange the class plan on the fly so that proof wouldn’t come up until the end, when we ran out of time. Not good times. It seems I’m going to have to do more preparation for logic classes in the future and rely less on my ability to just do things on the spot.)

The other changes are better.

I’m using many fewer footnotes now, again because footnotes are a pain in HTML. In part I’ve made up for that by using more parenthetical asides, but I think it makes my writing much clearer. It’s one of the (many) nice features of Lewis’s writing that he is very sparing in using footnotes. I think getting addicted to footnotes is a very bad habit, one that I fell into probably as an undergrad, and only got off when I started blogging.

There’s also a lot more jokes in the papers now that the papers are often trialled here before they appear. Sometimes I edit out some of the (less respectable) jokes, but usually they mostly stay.

I’m probably not the best one to judge this, but I’m pretty sure that my writing has improved significantly from when I started blogging. This is improvement from a pretty low base, but we take what we can get. Some of the improvement is just from practice. And some of it is from thinking that I have to write for an audience that doesn’t all read Linguistics and Philosophy from cover-to-cover as soon as it comes out. (Though if you don’t I really think you should.) Between those two things I’m losing some of my old annoying habits, and getting better at communicating. I’m not claiming I’ve suddenly turned into one of the great stylists of philosophy. At best what used to be a rather striking negative characteristic of my work is en route to not being a major flaw. As I said, progress from a low base, but progress.

One surprising non-consequence of letting blogging be first philosophy is that my papers aren’t getting noticably shorter. In part that’s because I’ve always been prepared to have very long blog posts. But in part it’s because I always wrote papers that were fairly modular, so sections of the paper could, without much tinkering, stand as posts on their own. And paper sections were never much longer than a TAR-sized blog post. I don’t know whether this is a good thing or a bad thing. On the one hand, churning out a 3000-5000 word paper every three to five weeks would keep the tenure-file ‘scoreboard’ ticking over nicely. On the other, serious 8000-12000 word papers are more likely to be noticed. On the third, my ‘long’ papers are actually a bit too long for good journal articles, so a bit of a weight loss program for papers wouldn’t be a disaster.

The Age is running a story today headed Asteroid Heads for Earth. Which sounds fairly scary I guess. The article then says that the best estimate is that the asteroid has a 1 in 909,000 chance of hitting the earth. I guess Asteroid might be heading for earth, like you might win the lottery this week was too long to fit above the story.

Oops, there’s a small mistake in one of the arguments against the contextualist account of colour below. But it’s an instructive mistake, so it’s not an utter disaster.

At one stage I used the following form of argument

Premise: There exists a phrase P such that I can say “Tunn and Larry said that P”
Conclusion: There exists a proposition p such that Tunn and Larry both expressed that proposition.

Well that’s not right in general, and in fact the exception is relevant here, so I retract that argument. Here’s an exception – or at least what looks like an exception to me:

Tunn: That looks orange.
Larry: That looks orange.
Me: Tunn and Larry said that looks orange.

I think that’s OK, and I think that it just says that each of Tunn and Larry said that the desk looks orange to them. In general it’s possible I think for a variable in the P phrase to be filled in by the speaker in each case, even when that makes for a different proposition in each case. Since that’s much of what the contextualist needs in this case, my argument doesn’t go very far.

This leads to an interesting possibility in that there are things one can say using elison that cannot be explicitly said (at least without seriously changing the grammar of the sentence). Consider

(1) Tunn said that looks orange, and so did Larry.

I think that can be used to mean Tunn said that it looks orange to her, and Larry said it looks orange to him. But you can’t say that by making the relativisation explicit, as in:

(2) Tunn said that looks orange to her, and so did Larry.

That has to mean that Larry said it looks orange to Tunn. If we have two speakers of the same gender, there’s still an ambiguit, but in cases like (2) I think it’s unambiguous and means something different to (1).

Anyway, the argument about “Tunn and Larry said that… ” below seems to me to fail. I still think the contextualist claim is a pretty implausible claim about the syntax/semantics of colour sentences, but the argument I gave for that doesn’t work.

I should follow up my hasty little comment about colour relationalism below, because I want to be a little bit clearer about what I am and am not endorsing, and what I am and am not dissing.

The following position seems to me to be well nigh untenable. When S says “That is C” (for some colour term C) she expresses (i.e. says in Grice’s preferred sense) the proposition “That looks C to observers O in conditions N”, where context fixes O and N, usually to be friends of S and conditions like she’s normally in. It’s untenable because it doesn’t account for the following facts.

(The baby ‘argument’ in the previous post was meant to show it was implausible, but some implausible things are true. John kissed Jack has an existential quantifier over events at the front of its LF. Who knew?!)

Indirect Reports
For all the examples, I’ll imagine Tunn is a visitor from the planet El, where everyone has such severe tunnel vision that Tunn could not possibly see (part of) my desk while having anything else in her visual field. So my desk, which looks brown to me, looks some other colour to Tunn, just like it would look to me without contrasts. (Let’s call the colour it would look ‘orange’, just for fun. I really don’t know the relevant psychophysics here.) In the following ‘that’ always denotes my desk, which by the way is brown.

Tunn: That is orange.
Me: Tunn said that is orange.

Presumably for me, there’s a tacit ‘for humans in these conditions’ in the proposition I expressed, but that isn’t what Tunn said. She said (on the hypothesis) that it looks orange to her (people) in these conditions. So my report is a misreport. But it sounds like a perfectly good report to me.

Well, maybe the tacit conditions fix on Tun since it’s her speech I’m reporting. That won’t do, if we extend the story. Lying Larry is a human liar. He rarely says true things, unless he can’t help himself.

Tunn: That is orange.
Larry: That is orange.
Me: Tunn and Larry said that is orange.

(alternatively) Me: Tunn said that is orange, and so did Larry.

But there’s no proposition Tunn and Larry both expressed, on the view I’m attacking, so it’s hard to see how any of these reports could be coherent. Yet they are perfectly correct.

Valuations
Same scenario as above

Tunn: That is orange

The following are all bad follow-ups on my part

That’s true.
That’s right.
Tunn said that’s orange, and she’s right.

But if Tunn said that my desk looked orange to her (people) in these conditions, I should say she was correct.

One might try and defend the contextualist position here by saying that we’re just systematically and massively mistaken about these reports, judgements etc. But I don’t know how that position could be defended. It seems to me that these judgements are just as much part of the data about the semantics of colour terms as the simple statements that theorists spend a long time trying to capture.

As I mentioned earlier, I think none of this undermines the arguments in favour of the metaphysical thesis of relationalism. But just in case you’re disposed to draw metaphysical conclusions from linguistic data, here’s some linguistic evidence that cuts the other way.

Let’s assume I’m no longer one of the folk, in that I know that there are such things as contrast conditions, and I know my desk normally looks orange to Els. Then I would hesitate to say either of the following things.

Tunn spoke falsely.
Tunn made a mistake.

Maybe Tunn never said a false thing in her life. It’s certainly compatible with the story that Tunn never had a false belief in her life. If ‘orange’ meant ‘looks orange to me’, we couldn’t explain that. Tunn would have spoken falsely.

The simplest way to capture all this data (I don’t know if this generalises, as the above makes clear I’m no colour expert) seems to be some variant of a relative truth account. Here’s one way of spelling that out – I make no stand on whether it’s the best way, just that it’s a way. (This way obviously owes a lot to John MacFarlane’s relative truth theory, though I’m not sure he’d agree with the last sentence or two.)

Tunn simply expressed the proposition that the desk is brown. (I can say Tunn said the desk is brown, because that’s all she said.) That proposition is true in a context iff most observers in that context see the desk as brown in most conditions. (Or something with more bells + whistles, but you get the idea.) I won’t say ‘Tunn was right’ because in my context she isn’t right. But, when talking about her speech, rather than about what she said, I focus on the context she was in, and so I won’t say she spoke falsely.

Why do I care about this? Well, colour is interesting. But I think some of our reactions to colour speeches by people like Tunn resembles our reactions to speeches involving epistemic modals by people with different knowledge states to ours. And epistemic modals are really interesting.

(All and only the true propositions in here are due to conversations with John Hawthorne and Andy Egan.)

This could go on all day, but I have serious work to do, so let’s finish with one of my favourite topics. It’s also from Laura’s 2D paper.

In that case, it may make best sense to say that your word is ambiguous and to assign two distinct semantic values. In Hartry Field’s terminology, your word partially refers to H2O and it partially refers to XYZ. This is quite different from saying that it refers to the disjunctive kind H2O-or-XYZ. Clearly, deciding which of these competing interpretations is correct will be neither easy nor automatic.

(Bonus nitpick: I rather doubt that Field’s partial reference is anything at all much like ambiguity. When a word (or, better, a sequence of letters) is ambiguous the inactive meanings are irrelevant to the truth conditions of sentences it contains. When a word has multiple partial referents, they are all relevant to the truth conditions of sentences it appears in.)

Actually this is usually fairly easy. I don’t know whether it’s automatic. These days nothing is automatic, not even taxes. (You still have to make an effort to pay them – that’s why they are late sometimes.) But it’s easy. I’ve probably harped on this a few times before, but since I got some things wrong last time I did, I’ll do it slowly this time.

I used to think cousin was an unambiguous term. Then I realised it was, at best, the disjunction of male cousin and female cousin. Maybe it isn’t disjunctive though. Or maybe it’s ambiguous between those two meanings. If I read Kripke on descriptions, it seems the way to tell whether a term that applies to Xs and Ys is ambiguous between X and Y is to check whether there are usually different terms for X and Y in other languages, and no term for their disjunction. Well I checked a handful of Western European languages (making sure the sample was small and not random) and found there are different terms for male cousin and female cousin, and no term for their disjunction. I was in a panic.

Fortunately I read Zwicky and Saddock’s “Ambiguity Tests and How to Fail Them” and came to my senses. Here’s a few ways that ‘cousin’ behaves that an ambiguous term, say ‘bat’ does not. (Assume Jack is a male cousin of mine, Jill is a female cousin of mine, Ron is my pet bat, and Rosie is my baseball bat. Assume that it’s not abnormal to name a baseball bat.)

(1) Jack and Jill are my cousins.
(2) *Ron and Rosie are my bats.

(3) Jack is my cousin, and so is Jill.
(4) *Ron is a bat, and so is Rosie.

There’s a few other tests that are (marginally) more subtle, but the general principle is sound. If a term t applies to both the Xs and the Ys, then it is ambiguous only if (5) is bad, where a is an X that isn’t a Y, and b is a Y that isn’t an X.

(5) a and b are t

If that reads badly for some reason, you can always use

(6) a is a t, and so is b.

If t is unambiguous, each of these is fairly natural, unless there is some other defect in the sentence. I’m inclined to think this part of the program really is easy. Whether determining the A and B is easy may be another matter, but there isn’t much of a problem at the ambiguity vs disjunction threshold.

Moving right along, a couple of points about Laura Schroeter’s paper Gruesome Diagonals. I think this is what set of the nit-picking. I see people bashing 2Dism, I have to respond. And I respond in the best way I know how – by ignoring the substance of the argument and looking for the most trivial points to find. (It’s fun blogging some days.)

On page 14, Laura says that ‘renate’ and ‘cordate’ are co-extensive in all biologically similar worlds. I was trying to find a formal definition of these in order to see whether this is right, not realising that the terms are a philosopher’s invention. (Quine’s, to be precise.) Learn something new every day. Though I think they’re in wide enough use now that the OED should include them.

So I don’t know the formal definition, but I take it they are properties of animals, and they are defined such that for all x x is a cordate iff x has a heart, and x is a renate iff x has kidneys. (I think this is the definition.) Well those are clearly not co-extensive if we take the ‘s’ at the end seriously – I know some cordate with only one kidney, who are hence not renates. But that’s getting ridiculous.

Let’s imagine that I, in the interest of science, take a small defenceless rat, and pull out its heart. We’ll now have a dead rat, and a bit of a mess, but I think we’ll still have a renate. The dead rat still has kidneys, no? So there’s another counterexample to the co-extensiveness thesis.

Finally, and this one is meant a little more seriously than the previous two, some patients these days get fitted with artificial hearts. They are renates who are not cordates, are they not?

I can’t remember what point the cordate/renate example was meant to make anyway, so I can’t tell whether I’ve refuted it by these little games. Somehow I doubt that I have refuted anything here, but you never can tell.