propositions, analysis and context
Interview by Richard Marshall.
Jeffrey King works in philosophy of language and metaphysics. He has written three books and numerous articles. He very much likes to ski.
3:AM: What made you become a philosopher?
Jeffrey King: I was really into mathematics and physics in high school and when I started college, that’s the sort of stuff I thought I wanted to do. In my sophomore year, I was taking a set theory course and on the last day of the course the professor started talking about axiomatizing set theory as a first order theory. I hadn’t taken any logic and I was having trouble understanding what he was saying. After the class I went up and began asking the professor to explain in more detail what he had been saying. He said that it was sort of a long story but that a professor in the philosophy department was teaching a class on first order predicate logic the next term and that if I took the course I would understand what he had been saying.
I did take the course and fell in love with logic. I then took a course on Frege and Russell from the same professor and was hooked. I planned to major in mathematics and minor in philosophy but the more philosophy I took, the more I loved it. Eventually, I majored in philosophy and minored in mathematics. When time came to think about what to do after my undergraduate education, I realized that I wanted to continue studying philosophy. I decided to go to grad school in philosophy. I knew that the job market in academic philosophy was very bad, so I planned to get my Ph.D. in philosophy and get on with my life after that by pursuing some other career. It didn’t turn out that way.
3:AM: You’re a top philosopher of language. A key area you’ve thought about is the nature of propositions. Before saying what you think they are can you lay out the terrain for us a little? Was it Frege and Russell who placed propositions at the heart of analytic philosophy, and do you see your work as developing from these founding fathers – or is there a deeper, longer history?
JK: Yes, I would say that Frege and (some time slices of) Russell placed propositions/thoughts at the heart of analytic philosophy. And yes, I very much see my own work on the topic as developing out of their views, in part by rejecting central features of their views. I know there is a deeper, longer history but my own thinking about the topic really starts with Frege and Russell. I’d add that my thinking about a lot of issues in philosophy of language starts with Frege and Russell.
3:AM: Robert Stalnaker says that propositions are sets of possible worlds, and Barwise and Perry say that they’re sets of situations but you dismiss these approaches don’t you? Are there stand-alone reasons for rejecting them even before we come to your own theory?
JK: Yes, I dismiss those approaches; and yes, I think there are stand-alone reasons for rejecting them.
The first thing that bothered me about such views, and it has bothered many people right from the start, is the coarse-grained way in which they individuate propositions. Propositions P and Q true at all the same worlds/situations are identified on such views. That has a number of bad consequences including that there is only one proposition true at all worlds/situations. So e.g. if the truths of mathematics are true at all worlds/situations as many think, there is only one mathematical truth. That’s a bit hard to believe. If propositions are the objects of belief, this means that any time I believe a proposition P, I must also believe any proposition Q that is true at all the same worlds/situations as P, since on the view in question P=Q. Again, this is hard to believe. Of course proponents of these views have things to say in response to these worries, but ultimately I am not convinced by the responses. It’s maybe worth adding that I’ve noticed that these consequences of the theories in question seem to be an insuperable difficulty to some people and a well-known difficulty that will be worked out in the end to others. I sometimes think this shows there are simply two species of philosophers!
A second difficulty concerns the fact that propositions are true and false relative to possible worlds/situations—ways things might have been. That’s a strange property to have and lots and lots of things don’t have it (numbers don’t have it, people don’t have it, nor do sets of numbers and people, etc.). I believe a theory of propositions needs to explain why/how propositions have this property. In my 2007 book The Nature and Structure of Content and in various papers since then I criticized virtually all extant theories of propositions for not being able to explain why/how propositions are true or false relative to possible worlds. I think that criticism applies to these theories as well. What would explain why/how a set of worlds or situations is true or false relative to worlds/situations? In general, sets of things aren’t true or false relative to worlds or situations. Why would sets of worlds or situations have this unusual property other sets lack? I just don’t see what the explanation of that would be. Further, why would a set of worlds P be true at world w just in case w belongs to P? Why wouldn’t P be true at w iff w didn’t belong to P instead? Could there really be some explanation of that? It’s very hard to see what it would be.
3:AM: You see propositions as structured entities. Can you say what you mean by this and why you think it a superior approach to its rivals? Is it because it handles propositional unity best?
JK: More or less following Russell in 1903, I take the constituents of propositions to be objects, properties and relations. I take it that for a proposition to be structured is for its constituents to be configured in a certain way in the proposition. On my view, constituents are configured in a certain way in the proposition by being bound together by a complex relation that I call the propositional relation. The (simple) constituents of a proposition are the relata of that relation.
This approach is superior to some of its rivals (e.g. propositions as sets of worlds) because it individuates propositions finely and so yields what I think is a suitably fine-grained account of propositions.
And yes, I think at least my version of the view promises to handle propositional unity better than its rivals. By this I mean it promises to explain what complex, including its constituents, the proposition is. And more importantly, it promises to explain how complexes of this sort can have truth conditions and so represent the world as being a certain way.
3:AM: Propositions aren’t the sentences or the diagrams explaining them are they, so do propositions exist in the way that numbers exist? If not, what kind of existence do they have?
JK: Well, I’m not exactly sure how numbers exist so I don’t know what to say about that (I’m a Platonist about numbers only every other day). On my view, the propositions expressed by the sentences of natural languages come into existence when languages do. I’ll cryptically add that I’m increasingly sympathetic to the Chomskyan idea that languages should be understood as I-languages (and not anything like public languages), so that propositions come into existence when I-languages do.
3:AM: If a proposition doesn’t exist until it’s in words can I first have a thought and then try and put it into words afterwards, or is this an illusion?
JK: On my view, once languages exist the propositions expressed by sentences of those languages exist, even those expressed by sentences that have never been put into words or tokened in any other way.
Further, a speaker of a language can entertain a proposition expressed by a sentence of her language without first articulating that sentence (either by tokening it or “thinking the sentence”).
3:AM: Does your position mean that non-linguistic creatures can’t have propositional thoughts?
JK: There are various things with contents—and probably truth-evaluable contents—other than sentences of natural languages: pictures, maps and perceptual experiences, for example. In each case, I claim that there will be a theory of these contents in the spirit of my theory of the contents of sentences of natural language. On a view like mine, it is an open empirical question whether the contents of any of these content-bearing entities will be expressible by sentences of natural languages. For example, it is an open question whether the content of a map can be expressed by English sentences. Whether this is so or not will depend on the details of the proper accounts of the contents of these entities. But in any case, nonlinguistic creatures can have attitudes towards the contents of their perceptual experiences whatever those are like exactly. They can have thoughts whose contents are the contents of their perceptual experiences. So when a dog is looking at a red ball, he is having a perceptual experience with a certain content. The dog can e.g. believe that content (which we would express by saying something like that he believes a red ball is in front of him).
3:AM: Are there philosophers who think that propositions don’t exist? Why do you think they are wrong?
JK: There certainly are! I think a lot of philosophers who don’t think that propositions exist are thinking of propositions along the lines of the classical picture of propositions of Frege and Russell. On that view, propositions are these sui generis entities that somehow manage to have truth conditions and so represent the world by their very natures and independently of minds and languages. That sounds very mysterious. No one has ever been able to explain how some sui generis abstract thing could have truth conditions by its very nature and independently of minds and languages. So I think folks are right to reject that view of propositions. But I claim there is a better view of propositions that doesn’t suffer from these defects.
3:AM: You’ve challenged Kaplan’s orthodox position (as it was then in 2001) about complex demonstratives which argued that ‘that phrases’ such as ‘that man over there’ are contextually sensitive devices of direct reference. What’s at stake in this and what was your challenge? Was this the first fruit of your work on the metaphysics of structured propositions?
JK: A standard (but by no means universally held) view about noun phrases is that there are at least two semantic types. First, there are directly referring expressions like names and demonstrative pronouns. These have as their semantic values (relative to a context) an individual. Second, there are quantificational expressions. These have as their semantic values something like relations between properties (so ‘Every A is B’ asserts that the properties A* and B*, expressed by ‘A’ and ‘B’, respectively, are related as follows: the things that have A* form a subset of the things that have B*).
Since Kaplan, the orthodox view was that complex demonstratives are directly referring expressions like names and pronouns (except that like pronouns and unlike names, they are contextually sensitive). I argued that complex demonstratives are quantificational expressions like ‘every student’ (except that the former are contextually sensitive in a way the latter are not). I argued that a quantificational approach could account for a broader range of data than a directly referring approach.
Yes, my account of propositions is one of the things that suggested to me that I should consider whether complex demonstratives are quantificational. On my account of structured propositions at the time, the structure of a proposition is identical to the syntactic structure of the sentence expressing it (that isn’t quite true any more). But then a sentence containing a complex demonstrative would have syntactic structure (that of the complex demonstrative) that would not be reflected in the structure of the proposition it expresses (in a context) if complex demonstrative directly refer. Not so if they are quantificational. So yes, my theory of propositions did suggest to me a quantificational approach to complex demonstratives.
3:AM: You had a new book on propositions come out in 2014 – is this because you perceive a kind of backlash against the project of providing a systematic account of the meaning of natural languages that is aimed generally at philosophy of language and not just your own work? Or is it driven by a much more specific need to answer some of the precise issues that have arisen since your last book about whether propositions exist?
JK: My 2007 book The Nature and Structure of Content laid out my view of propositions. Since it was published, certain criticisms of my view emerged that I was eager to respond to and when the book idea came up it seemed like a good venue to do that. I had also changed my mind about certain features of my view and was eager to spell out a new, somewhat different version of my view. (Unfortunately, I have changed my mind a bit again!) But really the clincher was that as Scott Soames, Jeff Speaks and I talked about what the book would be like, I just got really excited about it. The idea of a conversation among the three of us on propositions—both about what we agree on and about where we differ—just sounded like too much fun. I ended up feeling like I couldn’t not do it!
3:AM: I guess what you’ve been doing in all this is offering a philosophical analysis of propositions. But you’ve wondered about what a philosophical analysis is. So what is it, and why don’t you want to say that you’re analyzing concepts, which used to be a common view? Do you think philosophical analysis has changed or just how it is understood?
JK: When philosophers try to give analyses of knowledge, moral goodness, voluntary action, reference or whatever, it seems to me they are trying to say what those things are. I imagine canonical statements of analyses, in this case of voluntary action, to have the following form:
(A) For all x, x is a voluntary action iff C(x)
where ‘C(x)’ is some syntactically complex expression containing only the variable ‘x’ free.
As you say, the first question to ask is: what kinds of things are we purporting to analyze in making statements of this form? The answer I give in my papers on this topic is that we attempt to analyze mind independent properties and relations. Being a voluntary action is a property and the correct version of (A) will tell us what that property is. But what is it to say what a property or relation is? Well, I think that properties and relations can be complex and have other properties and relations as constituents related in certain ways. So, for example, it seems to me plausible that the bachelor property is roughly something like the conjunction of the male and unmarried properties. In analyzing a complex property or relation you are trying to say what its constituents are and how they are combined to form the complex property/relation. If one does that successfully, one has provided an analysis of the property/relation in question. This is a general notion of analysis that is broader than that of a philosophical analysis and includes things like:
(B) Bachelors are unmarried males.
(W) Water is H2O.
In both cases I think one is trying to say what the constituents of a complex property are and how they are combined to form the complex property. I am not sure what to call (B), but I would call (W) a scientific analysis. Neither is a philosophical analysis.
What makes an analysis a philosophical analysis has to do with the epistemic relation typical people bear to the property being analyzed. Very roughly, the property (relation) must be one whose instantiation typical people can more or less reliably detect. But typical people will not know the constituents of the property (relation) and how they are combined to constitute the property (relation). Again, that’s a bit rough, as the exact details would take a while to state properly.
As I said, on this sort of view analyses say what some mind independent complex property or relation really is in the sense of saying what its constituents are and how they combine to form it.
I don’t really have a good feel for the idea of conceptual analysis, but I do worry that viewing philosophical analysis as doing that makes it seem sort of trivial and threatens to make disputes over analyses a matter of talking past each other. My concepts are presumably things in my head. My concepts are in my head and yours are in yours. We each set out to analyze our own concepts and if in the end we disagree about the analysis, it looks like we just have different concepts. Further, what would be the long-term interest of my analyses of my own concepts?
I guess I don’t think philosophical analysis has changed. We just understand it better now, I think. And that’s so even if I am wrong about what analysis is!
3:AM: You’ve put forward a theory about speaker intentions and context. This involves supplementives and coordination and a whole bunch of contextualising stuff. How does it work?
JK: When one looks at the contextually sensitive expressions in natural language, those whose context independent meanings are sufficient by themselves to secure semantic values in context are very much in the minority. Kaplan called such expressions pure indexicals and ‘I’ and ‘yesterday’ may be the only ones (and I worry about ‘yesterday’). Even candidates like ‘here’, ‘now’, and ‘today’ don’t seem to be pure indexicals since e.g. the temporal extent of the semantic value of ‘now’ varies from context to context (‘Let’s leave the party now’ vs. ‘People now watch more content on their computers as opposed to their televisions than they used to’). Similar remarks apply to ‘here’ and ‘today’ (and perhaps ‘yesterday’). So something in addition to context independent meaning has to be brought in to secure a semantic value in context (assuming the context independent meanings of the expressions are the same in the different contexts).
I call contextually sensitive expressions whose context independent meanings don’t by themselves suffice to secure a semantic value in context supplementives. (The expression is meant to suggest that the context independent meanings of these expressions need to be supplemented in context to secure a semantic value.) Demonstrative pronouns like ‘she’ are perhaps the supplementives most studied by philosophers. But there are lots of other supplementives including tense, modals, conditionals, gradable adjectives, quantifiers (via quantifier domain restriction), expressions that take implicit arguments (‘ready’ in ‘John is ready’), possessives (‘Sophie’s skis’), ‘only’ and so on.
For a given supplementive, a crucial question is: what sort of supplementation does it require in context to secure a semantic value? Or better, what is the mechanism by means of which a given supplementive secures a semantic value in context?
In Demonstratives Kaplan famously claimed that in the case of demonstratives (and demonstrative pronouns) it is the demonstration (typically a pointing) that secures the semantic value. I say ‘she’ pointing at Annie, and she becomes the semantic value of the pronoun in that context. In Afterthoughts, Kaplan famously rejected that view and claimed that it was the “directing intention” of the speaker that did the job.
In my recent work on this topic I begin by arguing against both of these views. In their place I put what I call the coordination account. Very roughly, it says that the semantic value of a use of a demonstrative pronoun by a speaker S in context c is that object (if any) o such that S intends o to be the semantic value in c of the use of the pronoun and an idealized hearer would know that S intends o to be the semantic value in c of the use of the pronoun (I say something about ϕ-features too but I’ll suppress that here). I then spell out in some detail what an idealized hearer is. The idea is that successfully securing a semantic value for a pronoun in context requires the speaker to successfully reveal to an idealized hearer what she intends to refer to. There are many motivations for appealing to an idealized hearer instead of your actual audience, but a simple one is that I should not fail to secure a semantic value for my pronoun simply because you decided to ignore me and so didn’t know what I intended to refer to. So one property ideal hearers have is that they are paying attention to the speaker.
Assuming I’ve made a good case that the coordination account is the correct account of how demonstratives and demonstrative pronouns secure semantic values in context, I think the really big question here is: does the coordination account apply to all supplementives or are there several different mechanisms by means of which supplementives secure semantic values in context? I have been pushing for the former. I think this is just good methodology. Something like the coordination account is needed for demonstratives and demonstrative pronouns. Given that it is needed anyway, one should try to extend it to as many supplementives as possible and only posit an additional mechanism when it can be shown that the coordination account can’t work for some supplementives. I haven’t yet seen that this is so.
3:AM: Peter Unger has said this type of philosophy is ’empty’, and John Heil in his book on metaphysics says that there are no such things as propositions. What’s your push back?
JK: On the first point, I agree with what Timothy Williamson said about it in his review in The Times Literary Supplement. Basically, as far as I or Williamson can tell, Unger characterizes ‘empty’ (or ‘concretely empty’) in such a way that to say that metaphysicians’ claims are empty is simply to say that they don’t purport to provide any contingent information about concrete reality. Metaphysicians who knowingly make such claims should happily agree that they are doing so and note that the fact that a claim is empty in Unger’s sense is no reason for rejecting it or not taking it seriously. As Williamson points out, using a different example, what Unger does here is rather like my defining ‘bad book’ to mean book not written by me and then expecting it to sting when I tell you that your new book is bad.
I’d add that a lot of what philosophers of language say won’t be empty even in Unger’s sense, since to say things about what human languages are like is often to say something contingent about concrete reality.
As to Heil, it seems that the notion of a proposition he rejects is the classical notion. I looked up his 3am interview and he talked of “mysterious timeless entities with built-in meanings”. As I said earlier, I reject that view of propositions too. I just think there is a better view.
3:AM: Ernie Lepore claims that philosophy of language is going through a mini renaissance. Do you agree and if so, why do you say this?
JK: Oh yes, I agree completely. Simply put, the main reason is that there are just a lot of younger, very talented philosophers of language. From the level of newly minted full professors down to students working on dissertations, I just see a lot of talent out there. When I go to specialty conferences in philosophy of language and look around at the people there, I always think to myself “philosophy of language is in very good shape”. It gives me a really good feeling about the field. I’d name names, but I would inevitably forget people and then feel horrible about it.
3:AM: And finally are there five books you could recommend to the readers here at 3:AM Magazine who want to go further into your philosophical world?
JK: I’ll stick to old school backgroundy stuff since anything a lot more recent would presuppose a lot of that stuff anyway.
The Semantics of Definite and Indefinite Noun Phrases, Irene Heim
Papers on Philosophical Logic, David Lewis (especially Chapters1-3)
Context and Content, Robert Stalnaker
Demonstratives, David Kaplan (ok, not literally a book, but very book-like)
Modals and Conditionals, Angelika Kratzer (ok, a relatively recent book, but it collects old classics)
I’ll end with a shout out to Grice’s Studies in the Way of Words!
ABOUT THE INTERVIEWER
Richard Marshall is still biding his time.
Buy his book here to keep him biding!
First published in 3:AM Magazine: Sunday, April 19th, 2015.