Thank you all for helping me, in the previous thread, to
clarify what is at stake in being able to provide, for expressivism, a logic
based on the preservation of content rather than one based on the
(ir)rationality of holding certain combinations of attitudes. In this follow up post, I wanted to give some
of the basic ideas of the semantic and logical story for exclamative sentences that
I’m trying to make work for expressivism. I can’t possibly give all the details of the theory, but I’ll try to say
most of the important things and then respond to a few of the most serious
questions or concerns. Even so, however,
I’m afraid this is a rather long post.
Things will probably be clearer if I write out my various
comments and assumptions according to my train of thought. My apologies in advance if this appears
rather <i>Philosophical Investigations</i>-ish.
mood, and illocutionary force.
(1.1) The insights of
the theory originated with Kirk Ludwig. Kirk and I have since developed the theory separately and together.
(1.2) The purpose of
a semantic theory for a language is to explain all that one would have to know
in order to understand every sentence of the language for which the theory is a
theory. So, the purpose of providing a
semantic theory is to explain our understanding of a language.
(1.3) Sentential mood
obviously contributes something to a sentence’s semantics, or alternatively, to
our understanding of a sentence. There
is obviously a semantic difference between, and a difference in our
understanding of, ‘You will go home’, ‘Go home’, and ‘Will you go home?’, even
though all have the same propositional content. Likewise, there is an obvious semantic difference between, and a
difference in our understanding of, ‘If you go to the store, you will buy some
milk’ and ‘If you go home, buy some milk’.
(1.4) There is a very
close connection between the sentential moods (declarative, interrogative,
imperative, and exclamative) and the performance of certain kinds of direct
illocutionary acts. Declarative
sentences are particularly apt for the performance of direct assertive
illocutionary acts; interrogatives and imperatives are particularly apt for the
performance of direct, directive illocutionary acts; exclamative sentences are
particularly apt for the performance of direct, expressive illocutionary
acts.
(1.5) Thus, one
natural candidate for what mood contributes to the semantics of sentences is
illocutionary force. But, as Frege/Geach
point out, this can’t be the right story, since the illocutionary force of a
sentence S appears to be absent from some complex sentences embedding S. Thus, mood cannot be contributing
illocutionary force.
(1.6) So, what is the
semantic contribution that mood makes to a sentence (and to complex sentences
embedding that sentence), and how does this contribution help to explain the
close connection between certain sentence types and the performance of direct
illocutionary acts of a certain type?
(2.0) Success
conditions
(2.1) Sentences of
the various moods, and complex sentences embedding sentences of various moods,
have positive and negative semantic bivalent evaluations.
(2.2) There is something right, semantically, with the
sentence ‘You will go home’ if one’s hearer goes home at a time later than the
time of utterance; there is something wrong, semantically, with the sentence if
that condition fails to obtain. Thus, if
the right condition obtains—call it a "success condition"—its
obtaining would warrant the application of the predicate ‘is true’ (with
appropriate relativizations) to the sentence ‘You will go home.’ Thus, declarative sentences are evaluated as
‘true’ of ‘false’, the former being the sentence’s positive evaluation, the
latter the sentence’s negative evaluation.
(2.3) There is
something right, semantically, with the sentence ‘Wow!’ if the speaker is
excited at the time of utterance; there is something wrong, semantically, with
this sentence if that condition fails to obtain. Thus, if the right condition obtains—call it
a "success condition"—its obtaining would warrant the application of
the predicate ‘is sincere’ (with appropriate relativizations) to the sentence
‘Wow!’; it’s failing to obtain would warrant the application of the predicate
‘is insincere’ (with appropriate relativizations). Thus, exclamatives are evaluated as sincere
or insincere, the former being the sentence’s positive evaluation, the latter
the sentence’s negative evaluation.
(2.4) There is a
similar story with respect to imperatives and interrogatives. Sentences of each type have certain kinds of
compliance conditions. Imperatives are
obeyed or disobeyed, interrogatives are evaluated as answered or unanswered.
(2.5) The success
conditions of sentences of the various moods are precisely the success
conditions of the direct illocutionary acts that sentences of these moods are
conventionally used to perform. The
success condition for ‘You will go home’ is precisely the success condition of
the direct assertive illocutionary act performed in uttering ‘You will go
home.’ The success condition for ‘Wow!’
is precisely the success condition for the direct expressive illocutionary act
performed in uttering ‘Wow!’.
(2.6) It is this
"mapping" of success conditions that explains the tight connection
between the sentences of the various moods and the direct illocutionary acts
they are conventionally used to perform.
(3) Complex sentences
and semantic contribution of moods
(3.1) The same kind
of story can be told about complex sentences.
(3.2) There is
something right, semantically, with the sentence ‘If you won the race, wow!’ if
the speaker’s hearer won the race and the speaker is amazed (plausibly at her
hearer having won the race); there is something wrong, semantically, with this
sentence if this condition fails to obtain. Thus, if the right (success) condition obtains, its obtaining would
warrant the application of the predicate ‘is successful’/’is apt’ (with
appropriate relativizations) to the sentence ‘If you won the race, wow!’; its
failing to obtain would warrant the application of the predicate ‘is
unsuccessful’/’is not apt’. Thus, complex
sentences are evaluated as successful/apt or unsuccessful/not apt, the former
being the sentence’s positive evaluation, the latter the sentence’s negative
evaluation. (Kirk prefers ‘apt’/’not
apt’, I prefer ‘successful’/’unsuccessful’.)
(3.3) Thus, we have
strong evidence that what mood is contributing to complex sentences in which
they are embedded is the sentence’s success conditions—conditions that warrant
the applications of ‘is true’/’is false’ to declarative sentences, ‘is
sincere’/’is insincere’ to exclamative sentences, etc.
(4) Logic
(4.1) That sentences,
including complex sentences, have success conditions explains how sentences can
stand in various logical relations. Logical consequence is explained by the preservation of success, or
alternatively (?) of success conditions.
(4.2) Example 1: If (i) and (ii) are successful, (iii) must be
successful.
(i) If you won the race, wow!
(ii) You won the race
(iii) Wow!
pertaining to Jamie’s ‘Bob is hiyo!’ thought experiment: If (i) and (ii) are successful, (iii) must be
successful. (Assume that the success
condition for ‘Hiyo Bob!’ is a compliance condition, namely, that Bob attends
to the speaker.)
near, hiyo Bob!
(ii) Dingos are near
(iii) Hiyo Bob!
(iii).
the store or damn you!
(ii) You will not go
the store
(iii) Damn you!
(5) Expressivism
(5.1) Expressivists
think that ethical sentences work very much like exclamatives; both are
conventionally used to perform direct expressive illocutionary acts.
(5.2) Thus,
expressivists can use "Success condition semantics" and its resulting
explanation of logical consequence.
(5.3) Example 4: Semantic story for ‘If lying is wrong then
getting one’s little brother to lie is wrong’
(i) ‘If lying is
wrong then getting one’s little brother to lie is wrong’ is successful
(relative to a speaker and time) in English iff:
(ii) If ‘lying is
wrong’ is successful in English then ‘getting one’s little brother to lie is
wrong’ is successful in English;
(iii) ‘Lying is
wrong’ is successful in English iff ‘Lying is wrong’ is sincere in English;
(iv) ‘Lying is wrong’
is sincere in English iff the speaker disapproves at the time of utterance of
lying;
(v) ‘Getting one’s
little brother to lie is wrong’ is successful in English iff ‘Getting one’s
little brother to lie’ is sincere in English;
(vi) ‘Getting one’s
little brother to lie’ is sincere in English iff the speaker disapproves at the
time of utterance of getting one’s little brother to lie;
(vii) Therefore, ‘If lying is wrong then getting one’s
little brother to lie is wrong’ is successful in English iff if the speaker
disapproves at the time of utterance of lying then the speaker disapproves at
the time of utterance of getting one’s little brother to lie
(iii) must be successful
wrong, then getting one’s little brother to lie is wrong.
(ii) Lying is wrong
(iii) Getting one’s
little brother to lie is wrong
(6) Questions/Objections
(6.1) There are
obviously many sentential contexts into which ethical sentences embed
naturally, but exclamatives do not, for example, as antecedents of conditions,
within the scope of negations (*’It is not the case that wow!’), etc. So, ethical sentences appear to stand in many
logical relations that exclamative sentences do not. So, can expressivists (really) profitably use
this semantic theory and its related explanation of logical consequence?
not sure at this point. Merely from the
standpoint of success condition semantics and its explanation of logical
consequence, there seems to be nothing wrong whatsoever with the implication of
the success of (i) and (ii) to the success of (iii).
(i) If wow, then I
would like your autograph;
(ii) Wow!
(iii) I would like
your autograph.
So, from a logical standpoint, expressivists shouldn’t have
a problem here. However, expressivists
certainly have a burden to carry, namely, providing some principled explanation
for why certain kinds of sentences (ethical sentences) that conventionally
express attitudes can embed naturally in some contexts, while other kinds of
sentences (exclamatives) do not embed naturally into those contexts. Complex or hybrid expressivists would appear
to have an easier time meeting this challenge, since complex expressivists would
be able to hold that moral sentences have truth conditions in addition to
sincerity conditions, while exclamative sentences have only sincerity
conditions.
your theory, the success conditions for (i) and (ii) below are the same,
namely, that if the speaker’s hearer won the race, then the speaker is excited
for her hearer for doing so. So, how
exactly do (i) and (ii) differ semantically?
race, congratulations!
(ii) If you won the
race, I’m excited for you for winning the race.
to understand (i), one of the things we have to understand is that the success
condition for ‘Congratulations!’ is a sincerity condition—a condition that
warrants the application of the sincerity predicate. We thus understand that ‘Congratulations!’ is
apt for the performance of a direct expressive illocutionary act. We don’t understand any such thing in coming
to understand (ii). Rather, in coming to
understand (ii), one of the things we have to understand is that the success
condition for ‘I’m excited for you for winning the race’ is that the success
condition for that sentence is a truth condition—a condition that warrants the
application of the truth predicate. We
thus understand that ‘I’m excited for your for winning the race’ is apt for the
performance of a direct assertive illocutionary act. We don’t understand any such thing in coming
to understand (i).
talk about success conditions—I don’t really understand what they are. It sounds like you just made it up, and so
there is really no content to the notion of a "success condition."
am excited about John’s winning the race. If this condition obtains, then it seems that there is something
semantically correct/appropriate/etc. about the following sentences:
about John’s winning the race.
(ii) Congratulations,
John, for winning the race!
(iii) Dan—be excited
about John’s winning the race.
condition that warrants a positive (albeit, different) semantic evaluation for
each of (i)-(iii). So, a success
condition is a condition that must obtain in order to warrant the application
of a sentence’s positive semantic evaluation.
to go with the paper. And, of course, I
would love to hear any comments, questions, objections, or suggestions.
Wow. There is so much going on here and so many things to ask. I’ll start with one. It’s about the example in 5.3. In the midst of the case you have these two claims:
(iii) ‘Lying is wrong’ is successful in English iff ‘Lying is wrong’ is sincere in English
(iv) ‘Lying is wrong’ is sincere in English iff the speaker disapproves at the time of utterance of lying
By transitivity of biconditionals we get:
A) ‘Lying is wrong’ is successful in English iff the speaker disapproves at the time of utterance of lying.
I’m worried that this is just the thing which an Expressivist does not want to say. Blackburn gives endless rants about claims like these. The problem is these claims seem to lead to an awkward normative position from semantics alone. I’m not sure this but your account of succesfulness will probably commit you to this:
(i) lying is wrong iff ‘lying is wrong’ is succesful.
At least, if the succes of assertions is truth, then we get (i) from the disquotational schema in the case of normal assertions. It’s hard to think why things would be different for non-moral ‘assertions’ even when their success is not truth.
The problem is that with (i) and A) above, we get:
B) Lying is wrong iff the speaker disapproves at the time of utterance of lying.
And, then we are into an objectionable form of speaker-relativism…
Dan,
Maybe I’m not understanding, but there seems to me to be a big problem here. (It’s basically the problem Jussi mentions, but I think of it a different way.)
From 4.1 we know that logical consequence is necessary success-preservation. (I may have misinterpreted 4.1; if so please correct me.) This means that a sentence’s logical properties will supervene on its success condition.
Suppose we have two different sentences, S1 and S2, and one success condition, C. Suppose S1 has C as its truth condition, while S2 has C as its sincerity condition.
Now, on the one hand, we know that S1 and S2 have very different meanings. But on the other hand, they are going to bear just the same logical relations (to every other sentence in the universe), since logical relations of a sentence all supervene on success condition. That’s hard to believe. (It can happen, obviously, when two sentences are unobviously logically equivalent, so maybe the feature of the semantics and logic that I’m worried about is intentional?)
If I’m right that this is a problem, then I think I know where things went wrong. But tell me first if I’ve misunderstood.
Dan, I’m kind of worried about what Jussi and Jamie seem to be worried about as well, but I’ll leave that to their able hands to pursue. In addition to that issue I have a more amorphous question. The exposition of the project often uses phrases like this:
There is something right, semantically, with the sentence
There are many sentences in the above that say stuff of this general form, and I’m wondering what ‘semantically’ is doing here, because I tend to think that for most of these sentences (not all) when we take it out I agree with what is said.
But I don’t think it is a pre-theoretic datum that the sentences are true or plausible with ‘semantically’ left in place. Obviously the line between semantics and other parts of the theory of language is itself controversial. But on most accounts of the division, we can’t immediately move from a sentence being appropriately used to its being “semantically appropriate”. For one thing, sometimes the appropriateness of using an expression is not (just) a function of its semantic value, even in a context.
So then I think that ‘semantically appropriate’ must be a technical term of sorts in this context, and I wonder what exactly it means. And I also wonder if you could motivate your argument without using the phrase in saying that the something is going right in the examples discussed, but only bringing semantic appropriateness in to explain what you think is going right when the sentences use in the context that you indicate with the descriptions of how they are used.
This question is a bit vague, but it might give me a better sense of the motivations for the theory to hear what you have to say about it.
Jussi, that’s very nice. Let me assume that you don’t want to commit yourself to the semantic theory on offer, but that you think, nonetheless, that exclamatives and other nondeclarative sentences have some kind of semantic bivalent evaluation. So, rather than use ‘is sincere’ or ‘is obeyed’, let me use the neutral ‘has a positive evaluation’. So, tell me, what do you think about this kind of “disquotation” for exclamatives and imperatives:
(1) ‘Wow!’ has a positive evaluation (relative to a speaker and time) in English iff wow!
(2) ‘Go home’ has a positive evaluation (relative to a speaker and time) in English iff go home.
It seems to me that disquotation is out of place in these examples. If so, and if expressivists think ethical sentences work very much like exclamatives, then we shouldn’t expect the Success Condition account to be committed to:
(i) ‘Lying is wrong’ is succesful iff lying is wrong.
If there is in fact a difference between disquotation for nonethical declarative sentences on the one hand and ethical declaratives and nondeclaratives on the other, it must be because disquotation only makes sense when the right-hand side (in this case) of the biconditional describes (rather than expresses or directs) something. Does this response sound plausible to you?
Hi Jamie. I’m somewhat afraid (!) to say that you’ve understood me correctly.
Yes, that’s what I have in mind.
I think you are right about all of this. On my account, ‘Wow!’ and ‘I’m amazed’ will bear all the same logical relations, even to every other sentence in the universe, even though they have very different meanings. (And I agree that this is, at least initially, hard to believe.) This is because logical consequence, as I’m understanding it, is concerned with necessary preservation of success conditions, but there is more to the meaning of—there is more to understanding—a sentence than its success condition. In order to understand a sentence, one must understand not only what its success conditions are, but also what those conditions are conditions *for*–truth, sincerity, obedience, or response. (So, I don’t think that success conditions *are* meanings, whatever that might mean.)
So, here is my tentative explanation for why it seems difficult to believe that two sentences could bear all the same logical relations but mean something very different. I’m not completely confident that this is right story, but something like it sounds plausible. At the heart of the explanation is this claim, which I think most of us believe, though I think it is misleading:
Claim: The meaning of a declarative sentence can be given by specifying its truth conditions.
We generally think: (i) that logical relations obtain among declarative sentences; (ii) that declarative sentences have truth conditions (and so we generally understand logical relations in terms of these conditions); and (iii) that the meaning of a declarative sentence can be given by specifying its truth conditions. If we think (i)-(iii), then we will tend to think that two sentences that share all of the same logical relations will mean the same thing. But, if we think that there is more to understanding a sentence than understanding what its success conditions are, and if we think that logical consequence is concerned with the preservation of success conditions only, then it will make sense that ‘Wow!’ and ‘I’m amazed’ will bear all of the same logical relations, even though they have very different meanings; for there is more to their meanings than their success conditions.
I’m sure I’ll be sorry in the morning for having wrote that.
I would appreciate any help you can give me, Jamie, believe me. Thank you.
Hi Mark. Geez, I sure hope yours is the last of the hard questions!
Your point is very well-taken. Clearly, not every case of there being “something right” or “something wrong” with a sentence is a result of semantic infelicities. Indeed, I make a big deal elsewhere that sentences like ‘If congratulations, then I’ll buy the beer’ is pragmatically infelicitous, not semantically so.
That the (in)felicities are semantic is a conclusion I’m drawing (I won’t speak for Kirk here) based on three individual considerations, which together, I think are mutually reinforcing. First, while I agree with you that this claim is not pre-theoretically true, I do think it is pre-theoretically plausible. For example, suppose I cross the finish line to win the race, you say to me ‘Boo!’, and I (sincerely) thank you for being excited at my having won the race. In this case, it would be plausible to think that I didn’t understand the sentence ‘Boo!’. Second, that the (in)felicities I mentioned are semantic is even more plausible in light of the absence of alternative, workable candidates for the semantic contents of nondeclaratives and mixed-mood sentences. Third (and this is related to the ‘Boo!’-example I just gave), if the claim about the (in)felicities were true, it would be the best explanation, or so I claim, of the extremely close connection between sentential mood, semantic content, and illocutionary force.
Dan,
This now reminds me a bit of some things Frank Jackson said about the ‘Lockean’ theory of content. The problem with what he said was that things with the same Lockean truth conditions could have quite different logical implications. That’s bad, because truth conditions are supposed to tell us the position of a sentence in logical space.
You want to bite the bullet, I guess, but it’s going to be a very big bullet. You will have to say that this argument is valid:
Now for my diagnosis: sincerity conditions of exclamatives cannot be analogous to truth conditions of indicatives, because they are analogous to sincerity conditions of indicatives.
Hi Jamie,
There are a number of things I want to say about your comments here and in the paper, but I only have time right now for a couple of them.
First, I think I really do want to dig in my heels here, at least for a little while longer. The absurdity of thinking this argument valid arises, it seems to me, only if we think that validity is concerned with the preservation of truth. (Why else would one think it invalid?) But, if we grant that noncognitivists have principled reasons for thinking that ethical sentences lack truth conditions, and grant that some arguments containing ethical sentences are clearly valid, noncognitivist expressivists appear to have a principled reason for thinking that validity is something more/other than the preservation of truth conditions.
Second, about this:
As I’m understanding things, indicative/declarative sentences don’t have sincerity conditions, so sincerity conditions of exclamatives cannot be analogous to sincerity conditions for declaratives. The assertive illocutionary acts that declaratives are conventionally used to perform have sincerity conditions, and these are analogous to the sincerity conditions for expressive illocutionary acts; but the sincerity conditions for assertive illocutionary acts are not also the semantic content of declarative sentences. Does that help me in any way?
Dan,
I’m not sure. I don’t see how disquotation is out of place *on your view* in (1) and (2). Of course, disquotation *seems* out of place there because the sentences are ungrammatical in a bad way. This is the same way as the conditional with an ‘Wow’ as antecedent seems non-sensical because of its ungrammaticality. But, when you give the same account of the disquotation biconditional as a result of two conditionals your account seems to support the disquotation.
This is because the crux of both of those biconditionals is that a necessary and a sufficient condition for being able to *use* the expression on the right side is that that very same expression *mentioned* in the left side has a positive valuation. Denying that seems impossible. The positive valuation of the mentioned expression just seems to be the necessary and sufficient condition for its use.
First, let me echo Jamie’s diagnosis. It connects to the dilemma I tried to pose for Dan in the last thread, in terms of how the success conditions for mixed sentences are to be understood – they’re not truth conditions, and they’re not sincerity conditions, so what are they conditions of? The view that treats all sentences by assigning them sincerity conditions doesn’t have this problem.
Dan says: ‘indicative/declarative sentences don’t have sincerity conditions’. Surely that’s not what he means, though. Surely what he means is that indicative/declarative sentences aren’t assigned sincerity conditions directly by the compositional semantics! They certain do have conditions that you have to be in, in order to be sincere in asserting them, though! If you assert ‘grass is green’ while not believing that grass is green and understanding what it means, then you’re being insincere. No? If that’s right, then I think Jamie’s point stands.
Hi Mark,
They are the conditions of semantic success/semantic aptness/semantic felicity—whatever you want to call the conditions that warrant the most general, positive semantic evaluation of a sentence. (Something tells me, though, you knew I was going to say that.)
I agree that *I* would be insincere, and I certainly agree that the *assertive illocutionary act* I perform in uttering the sentence would be insincere. But evaluating declarative (or interrogative or imperative) sentences as sincere or insincere is, I think, to conflate an evaluation type for illocutionary acts (or speakers) with an evaluation type for the sentences that are conventionally used to perform these acts. (In response to Mark vR’s question, I’ve tried to give a principled reason to think that exclamative sentences are evaluated as sincere/insincere.) Of course, you can, if you want to, assign to sentences of these types the sincerity conditions of the direct illocutionary acts that they are conventionally used to perform. But in doing so, I would say that you are forcing onto these sentence types conditions for a semantic evaluation that I do not think they ordinarily have, just as you think I am forcing onto mixed mood sentences conditions for a semantic evaluation (success/failure) that you do not think they ordinarily have. At any rate, you are certainly correct that, on this account, neither declaratives, imperatives, nor interrogatives are assigned sincerity conditions.
Thank you, everyone, for all of the help. I’m learning a lot from these two posts.
Dan,
I figure we already know which sentences are logical consequences of which. So I don’t agree with this:
My view is that which (simple) arguments are valid is data. The task at hand is to explain the data with some expressivist theory; you aim to do so with a story about what positive semantic value logical consequence preserves.
I guess you are in fact talking about some other relation. I don’t have a problem with your caling it ‘logical consequence’ (relevance logicians call their relation ‘logical consequence’, after all), but it isn’t the intuitive logical consequence relation. Yours is interesting too. (I think your version of ‘validity’ might be what I call ‘Moorean validity’ in the paper I linked to above.)
But, what explains the intuitive logical consequence relation?
Hi Jamie,
There is no question in my mind that what explains what you are calling the *intuitive* logical consequence relation is the preservation of truth.
So, if I’m understanding correctly the implications of this part of the discussion for my overall question of the past two posts, it is the following:
–Noncognitivist expressivists have had a difficult time explaining certain data; I suggested that noncognitivist expressivists could use my general story to explain the data;
–The data that requires explaining is that there are arguments of certain simple forms that we intuitively evaluate as valid, and arguments of certain simple forms that we intuitively evaluate as invalid;
–But, on my general story, some arguments that are intuitively invalid turn out to be valid;
–Therefore, even if my general story is correct, it cannot be used to explain what was supposed to be explained, as I suggested it could.
If this is right, the main question at this point, besides whether the general semantic story is adequate, is whether this general story can be used to explain anything of logical importance (rather than of mere interest), including, perhaps, anything that might be more logically important than the “intuitive logical consequence relation.” I’m way out of my depth here, and it’s not clear to me whether this question even makes sense.