Saturday, November 24, 2007

Scattered thoughts on a vague distinction in the philosophy of language

This may be quite naive and rambling, but I'll go ahead. There is a difference in views of language that I've been somewhat puzzled by for a while. One approach takes linguistics fairly seriously and focuses more speaker intuitions. These two parts may not be intimately related. Philosophers falling into this camp are, say, Jason Stanley and Francois Recanati. Another camp tends not to pay that much attention to linguistics. There is, maybe, a tendency to view language through the lens of first-order logic, broken into terms and predicates. Philosophers in this camp are, say, Sellars and Dummett. I think Davidson might be read as being on each side at different points in his writings. As it stands, I've drawn nothing resembling a clear line. This may even play out in different approaches to semantics. Why one might concentrate on, say, meaning in terms of reference or meaning in terms of inference. As an aside, I once asked one of my linguistics professors at Stanford what linguists thought something like conceptual role semantics. He said that they didn't really think about it because it was a different sort of project than what they were interested in. He pointed me to this little explanatory piece by Ned Block. That helped put things into perspective, but I digress... I almost want to say that the integration of philosophy of language into other areas of philosophy also cuts along these lines. However, I am fairly sure that is false and more a product of selective memory. People on both sides are interested in integrating philosophy of language into other areas, such as philosophy of mind or of action. In any case, I came across a footnote in van Fraassen's essay on Putnam's paradox which exemplifies one side (It is left as an exercise to the reader to guess which)(The context is a discussion of translation and knowledge of meaning):
"For the sake of example I am here pretending that Dutch and English are two separate languages in actu. I usually think of one's language as everything one has learned to speak, and of natural language as consisting in all the resources we have for speaking and writing."
This, by itself, probably does not force van Fraassen into one or the other grouping, although it seems to lend itself to fitting into one group rather than the other. This idea, it seems, lends itself to viewing language more expansively than usual for linguistic semantics. Interestingly, I think that Lewis in "Language and languages" could be on board with this completely even though he is someone I would usually situate in the first group. I must go ask some linguists what they think of this...

This was a little rambling, but, I am currently stuck on a paper idea. What better way to pass the time than ramble on about approaches to the philosophy of language? Probably doing those model theory proofs now that I think about it... But, I'm pretty sure people that would naturally place themselves in one or the other camp read this, so hopefully someone is interested.


Justin said...

Oh snap, it's coming...

If you're using philosophy of language to work on philosophy of mind, something has gone horribly wrong.

Shawn said...

Oddly enough, I did not see that coming. You know, there is probably a similar distinction in the philosophy of mind. Another way of putting it: those for whom empirical studies matter a lot and those for whom they matter little, if at all. Of course, "a lot" and "little" are relative to philosophers.

Rick Danko said...

I'll just assume that Sellars didn't pay attention to linguistics because it didn't really exist when he was doing his important work. I'm pretty certain EPM came out the same year (1956) Chomsky got his PhD. Now certainly Sellars wrote things after this, but I think it is also safe to say that linguistics didn't really come into its own until the mid to late 70s, when Chomsky developed Principles and Parameters theory and people like Dowty, Gallin, Thomason and Cooper slimmed down "Montague Semantics" to something that people other than Montague himself could understand.

Also, it would be difficult to not view language through the lens of first-order logic when that is the only significant conception of logic that one has. This certainly applies to Davidson, who was under the Quine-inspired view (delusion) that first-order logic was the *only* sort of logic that one should care about (this probably is a caricature of Quine and a bit unfair to Davidson, but it seems that something like this view is on display in Davidson's Inquiries-era stuff, where he is hostile even to "model-theoretic" theories of truth, to say nothing of higher-order logic, as something that can be of use to semanticists (in the intro, he recants on this, but nevertheless still pushes the (wrong) thought that we should try and push first-order analyses of semantics as far as possible because of the "well-known benefits of first-order logic" (or something, I'm paraphrasing).

Anyways, I agree with your idea that Davidson can appear to be on both sides of these camps, but not because there are different "phases" in his writings on semantics and language. The history of Davidson is that of gradual improvement and elaboration, not of one "phase" where he was concerned with one set of problems and thought about things in a certain way and then another "phase" where he began to take a fundamentally different view of things. Granted, this sort of wrong-headed view is encouraged by the fact that two very different sorts of people are interested in Davidson, and are often interested in different aspects of Davidson (think of Ernie Lepore and Bjorn Ramburg (sp?), or Kirk Ludwig and Richard Rorty, or Joe Blow at Rutgers and Herbert F. Codpiece at Chicago) But just because different people find different aspects of Davidson's thought attractive doesn't mean that these aspects don't form a coherent whole. They do.

Instead, I submit that the reason Davidson can seem like he "takes linguistics seriously" is because he was the first person to ever seriously argue for a formal approach to semantics in the form we know today, namely the truth-conditional logic based approach. Montague begins "English as a Formal Language" by saying something like "I agree with Donald Davidson that the proper goal of semantics is to provide a theory of truth for a language..." (No one seems to notice this interesting fact (that Montague took himself to be working on the same problem as Davidson) or to care (probably because no one reads Montague anymore, or cares about the problem he took himself to be working on (which is sad))).

Now, the reason he can seem like he also falls into the camp that "doesn't take linguistics seriously" is because no one understood (cared?) *why* he thought a theory of truth for a language actually tell us something about meaning. Montague is actually completely wrong. Davidson didn't think that the proper goal of semantics is to specify a theory of truth for a language. He only thought that the proper goal of semantics *could be realized* by constructing such a theory. Which doesn't rule out the possibility of such a goal being realized in other ways.

This problem was exacerbated by (suprise) Ernie Lepore who argued in Linguistics and Philosophy in the early 80s that the significant difference between "model-theoretic" approaches inspired by Montague and "Davidsonian" approaches was that Davidson appealed to "absolute" theories of truth, whereas "model-theoretic" approaches appealed to "relativized" accounts of truth; that "Davidsonian" approaches could "do something" that model-theoretic approaches "couldn't do" (the title of the paper I'm referring to is literally something like "What model-theoretic semantics can't do"). Obviously, there is some sort of difference between these two "approaches", but it is really a minor difference based on a constellation of confusions on Davidson's part which were subsequently adopted by his readers (Davidson recognized this in the introduction I cited earlier, and made a really big point about it in chapter 3 of Truth and Predication). Arguing that the distinction between model-theoretic and "absolute" characterizations of truth is the *major* difference between Davidson's approach to natural language semantics and the mainstream one inspired by Montague (who in turn seemed to take himself to be working on the same project as Davidson) is "not even wrong." It just misses the major divergence completely.

Unfortunately, this thought, and the way of approaching Davidson's stuff about semantics it flows from, seems to have gained currency, mostly due to the fact that Lepore and Kirk Ludwig have become the world-appointed interpreters and expounders of the writings of Donald Davidson on language. As I said, it is not so much that Lepore and Ludwig get things wrong. It is more that what they end up saying, in many respects, is not even wrong.

(in case you can't tell, I started this comment while I was taking a break from writing a paper on Davidson, but ended up writing a rambling rant-y version of the paper by the end. many apologies.)

Shawn said...

Since your comment is so long, I might need to respond to it in parts. You are probably right about Sellars's productive era and the timing on the birth of generative linguistics. Rich Thomason was at Pitt with Sellars from the early 70's to Sellars's death. I suppose by that time Sellars it would be difficult for the new ideas to work their way into his writing. But, I was throwing Sellars out as an exemplar of this. Dummett does the same and he is younger than Sellars.

On a side note, that is fascinating that Montague said that about Davidson. I've tried to read PTQ, but never the article you cited. I've never had to read Montague for a class, not even the formal semantics class I took. The professor seemed to say that there was nothing in Montague we wouldn't get out of Heim and Kratzer.

I'll agree that it makes some sense for Davidson to view language through first-order logic. Quine gives his reasons for wanting to do this too. Why people still do this, I don't know. Even in Quine's time there were more powerful formalisms, such as type theories. Now there are much more flexible formalisms. HPSG is, essentially, first-order, but the FOL representation of it is terribly unwieldy and opaque, whereas the feature-structures of HPSG are fairly straightforward and nice.

Shawn said...

I said there were phases to Davidson's thought because I'm not sure how they fit together. Almost all of Davidson's philosophy of language I've learned about has been piecemeal on my own over the course of a couple of years. I haven't figured out how it fits together. It is nice to hear that it fits together. Davidson, in the intro to one of the volumes of his essays, indicates he thinks it fits together neatly. I'm still trying to figure out how the early Inquiries material fits together with the stuff from "A Nice Derangement of Epitaphs".

Davidson on semantics is something that I have been uneasy with for a while. You add another piece to the puzzle with the comment that specifying a theory of truth wasn't the goal of semantics for Davidson. (I should get back to thinking about that...) You've also convinced me I need to look at Montague's work.

I haven't read the Lepore and Ludwig book on Davidson's philosophy of language. I wasn't aware that it was the standard way of understanding Davidson. The Lepore paper you mention sounds kind of crazy. I will look into this; maybe get a post out of it in the future.

Justin said...

Just by chance, I found that in 1960, Sellars was referring to the way that philosophy often turns particular subject matters over to specialists outside of philosophy.

"If philosophers did have such a special subject-matter, they could turn it over to a new group of specialists, as they have turned other special subject-matters to non-philosophers over the past 2500 years, first with mathematics, more recently psychology and sociology, and, currently, certain aspects of theoretical linguistics."

To what extent that indicates knowledge of ongoing developments, *shrug*. It's from the second page of Philosophy and the Scientific Image of Man.