Recent Posts

The Bertrand Russell Show

Feminist Philosophers

fragments of consciousness

Gender, Race and Philosophy: The Blog


Leiter Reports: A Philosophy Blog

Long Words Bother Me

semantics etc. highlights

Thoughts Arguments and Rants



Thursday, July 20, 2006

Modality and Assertoric Content

In "Names and Rigid Designation" Stanley argues against the thesis that rigid and non-rigid expressions never have the same assertoric content. King ("Tense, Modality, and Semantic Values") construes Stanley's argument as an argument against Dummett's distinction between assertoric content and ingredient sense. Here is a related argument, a reductio on the thesis that there are modal operators in the language, and that assertoric content is absolute (viz. if true, true at all worlds; if false, false at all worlds). Consider:

(1) It is possible to travel faster than light

Given Kaplan-style semantics, (1) is true relative to context c iff the content of the operand sentence in (1), viz. that some object travels faster than light, is true at some world accessible from the world of c. Since different contexts will determine different accessibility relations, let us focus on contexts in which speakers are talking about physical possibility. The content of (1) is then false in any actual context. For, as the actual laws prohibit faster-than-light speeds, there is no accessible world in which some object travels faster than light. But since we are quantifying over all contexts c in which the speaker of c is talking about physical possibility, the actual world need not be accessible from the world of c. The content of (1) is true when (1) is uttered at a world where the laws of physics allow objects to travel faster than light. So the content is false at some worlds (e.g., here) and true at some worlds (i.e., distant worlds); so the content is not an absolute proposition. We can conclude that if we want assertoric content to be absolute, we are required to take the modals to be quantifiers over worlds.


Jason Stanley said...


First, some clarificatory remarks: my argument was supposed to be an instance of Dummett's distinction between assertoric content and ingredient sense. That is, I was arguing that folks like Mark Richard and Nathan Salmon employ the assertoric content/ingredient sense distinction already, by distinguishing between temporal semantic values and eternal propositions -- the former feed into tense operators, but are not propositions. I then argued that the relation between modal contents and propositions is similar to the relation between temporal contents and propositions; just as Richard and Salmon don't assume that what temporal operators operate on aren't propositions, so they shouldn't assume that what modal operators operate on are propositions. So my argument is in fact an application of Dummett's distinction between assertoric content and ingredient sense. Jeff, in his paper, argues that in fact tenses aren't content operators at all, but rather variable binders, and so the analogy doesn't work. More broadly, Jeff is basically arguing that ingredient sense and assertoric content always converge in natural language semantics -- what feeds into *genuine* operators is always a genuine proposition. The alleged counterexamples (e.g. tense) do not actually involve genuine operators.

But the main issue you raise is whether a position like the one I advocate in my papers on that topic (the one you cite, plus "Rigidity and Content" and "Modality and What is Said") is committed to modal operators being quantifiers over worlds. I don't think I'm committed to that. But what I may be committed to is treating assertoric contents -- genuine propositions -- as being world-indexed.

What I argue for in my three papers on the topic is that *if* one thinks that temporal expressions are operators on non-propositional semantic contents, temporal semantic values, then one should also think that modal operators are operators on non-propositional semantic values. That is, given the analogy between tense and modality, Richard and Salmon were wrong to try to forge an asymmetry between the two. But one way of thinking of the relation between temporal semantic values and genuine propositions is that the latter are eternal, viz. indexed to a specific time, whereas the former are not. Given the symmetry claim, it would seem that I'm suggesting that the relation between modal semantic contents and genuine propositions is that the former are not indexed to a specific world, while the latter are indexed to a specific world.

I'm not sure how this relates to the consideration you raise...

Jason Stanley said...

Oops..typo. "just as Richard and Salmon don't assume that what temporal operators operate on aren't propositions, so they shouldn't assume that what modal operators operate on are propositions."

This should be "just as Richard and Salmon don't assume that what temporal operators operate on ARE propositions..."

Brit Brogaard said...

Hi Jason. Thanks for the clarifications! As I see it (though you know best of course), you have two arguments for rejecting the Rigidity thesis in the 1997 papers (i.e., the thesis that if the only difference between an utterance of S1 and an utterance of S2 is that S1 contains a rigid term and S2 contains a non-rigid term, then the utterances of S1 and S2 will have different assertoric contents). One argument is that if one already thinks assertoric content and ingredient sense come apart, then one should be careful when appealing to Kripke's modal argument. For Kripke's modal argument relies on intuitions about modal profiles. Yet it might be that those intuitions track properties of the ingredient senses (I think King makes use of this version of the argument for different purposes).

The other argument, I think, turns on the very plausible assumption that the Expression-Communication Principle is valid (viz. that if two utterances express different propositions, then in normal contexts, they communicate diffferent things). Utterances of 'The actual president is George Bush' and 'the president is George Bush' communicate the same thing in normal contexts. So by ECP, they have the same assertoric content. But 'the actual president' is rigid, whereas 'the president' is not. So ECP (together with a distinction between assertoric content and ingredient sense) casts doubt on the Rigidity Thesis.

If, however, Salmon and others take assertoric content to be eternal, couldn't they just reply to the first argument that since assertoric content have properties like contingency etc, modal operators must operate on assertoric content?

Jason Stanley said...


You are right in your description of my two arguments. As for the reply you mention on behalf of Salmon and others, that "assertoric contents have properties like contingency etc.", I take it that is what is at issue -- whether contingency is a property of assertoric contents or ingredient senses. As I point out in the last section of "Names and Rigid Designation", we also say things like "What John said was true, but isn't true anymore", so it looks that our practice of saying ascriptions suggests that assertoric contents have temporal properties. Since I agree with Richard, Salmon et. al. that non-trivial temporal properties are not true of genuine propositions, which are eternalist, I think what this shows is that our saying ascriptions are rather loose -- we can refer to non-propositions by expressions such as "what is said". Mutatis Mutandis for arguments to the effect that contingency and necessity are properties of assertoric contents. Sure, we speak of what is said as necessary or contingent. But since we also speak of what is said as having been true and not being true anymore, and all parties to the debate agree that assertoric contents are eternalist, nobody can take such ascriptions at face-value, even when they involve modal properties such as contingency and necessity.

So Richard and Salmon and I all agree that propositions are eternalist; that is fixed in the debate. They say that temporal properties hold of non-propositional entities, but assume that modal properties hold of propositions. I say that the only reasons for thinking that modal properties hold of propositions are reasons we also have for thinking that (non-trivial) temporal properties hold of propositions (viz. saying ascriptions). So Richard and Salmon aren't entitled to that assumption.

Brit Brogaard said...

o.k. that's a neat counter-reply. I'll buy it. However, perhaps Salmon and Richard could reply to the other argument (in "Rigidity and Content") as follows. Which (compositional) semantic value a term has must be informed by how people use the term (assuming the term is a term in ordinary English). So, your example shows that 'actual' does not (always) function as an indexical rigidifier in ordinary language, not that the Rigidity Thesis is false. Here is a Cappelen/Lepore-style reason to think that 'actual' is not an indexical rigidifier.

If John said "I am hungry now" and what he asserted was true, then he is hungry now ---- bad

If John had said "I am actually hungry" and what he asserted were true, then he would actually be hungry --- much better

You have some other examples, for instance 'Julius ate breakfast' and 'the inventor of the zip ate breakfast', where 'Julius' is introduced to talk about the actual inventor of the zip. But isn't it a controversial thesis that descriptive proper names (i.e., names that are introduced via description) are rigid terms? If they are not, then they do not create trouble for the Rigidity Thesis.