The Embedding Objection: Part IV "The Challenge from Incomplete Semantics"

The series on "The Embedding Objection" continues. The first post explained what I take expressivism to be. The second post distinguished four main kinds of expressivism: Simple non-truth-evaluable expressivism (e.g., Ayer's emotivism), Simple minimalist expressivism (e.g., Blackburn's projectivism), Complex minimalist expressivism (e.g., Stevenson's emotivism), and Complex robust expressivism (e.g., Hare's prescriptivism, my Expressive-Assertivism). The third post discussed The Objection from Truth Ascriptions. In this post, I discuss what I used to believe was the most pressing difficulty for expressivists, what I call "The Challenge from Incomplete Semantics."

According to The Challenge from Incomplete Semantics, there is a large gap in many expressivist theories, since many are not "semantically robust" enough to show how we can understand the meanings of complex ethical sentences. The idea is that many, if not most, occurrences of ethical sentences are embedded within complex sentences; thus, any metaethical theory that does not tell us how we can understand the meanings of complex ethical sentences is radically incomplete. Since many expressivist theories do not tell us how we can understand the meanings of complex ethical sentences, it follows that many expressivist theories are radically incomplete. So, expressivist theories must tell us how we can understand the meanings of complex ethical sentences, or they will lack sufficient detail to warrant our rejection of these theories.

This challenge is the central focus of Jamie Dreier's terrific article "Expressivist Embeddings and Minimalist Truth." Dreier's explicit targets there are expressivist theories that accept minimalism about truth (SM- and CM-expressivism), since these theories cannot use a Tarski-style truth theory to explain how we can understand the meanings of complex sentences that embed ethical sentences on the basis of understanding the meanings of their component sentences, logical connectives, and syntactic combination. Hence, these theories are required to provide some other compositional semantic account of how to understand complex ethical sentences. However, this challenge could be raised against any kind expressivism, including SNT- and CR-expressivism, since these theories also cannot make use of a Tarski-style truth theory to explain how we can understand the meanings of complex sentences that embed ethical sentences. (A CR-expressivist could use a Tarski-style truth theory to explain how we can understand the "descriptive" meanings of complex ethical sentences, but would have to supplement it to account for the "expressive" or "prescriptive" meaning of complex ethical sentences.)

I'd like to make two points about this challenge. First, I agree that any expressivist theory that wants to be as complete as possible should offer a semantic theory that explains how we can understand the meanings of complex ethical sentences on the basis of understanding its component sentences, logical connectives, and syntactic combination. Indeed, as Dreier notes, many of the major expressivist theories, including Blackburn's Projectivism, Hare's Prescriptivism, and Gibbard's Norm-Expressivism, all rise to meet this challenge. Thus, it seems to me that this challenge is legitimately directed against all expressivists theories. But, second, it's not clear to me why an expressivist theory's failure to include such a semantic theory warrants rejection of that theory, for I don't see that expressivists have any special burden that others do not also have. Here's an observation: as far as I can tell, exclamations and imperatives appear to be unproblematically embeddable in almost every kind of complex sentence, including mixed mood sentences. Let me use 'Congratulations!' and 'Go home' as my two example sentences:

(1) If you do not want to listen, then go home.
(2) If you won the race, then congratulations!
(3) Go home or did I not make myself clear?
(4) Congratulations on winning the race or am I congratulating you before the fact?
(5) Congratulations on winning the race and rest up during the offseason.
(6) Congratulations and best wishes.
(7) Go home and rest up during the offseason.

Now, if a compositional semantic theory is required in order to explain how to understand the meanings of these complex sentences on the basis of their component sentences (including the component exclamations or imperatives), logical connectives, and syntactic combination, then there must be one (and the burden to uncover it falls on everyone).

Expressivists hold that ethical sentences work very much like exclamations or very much like imperatives. Why could an expressivist not simply hold that the correct compositional semantic theory for complex ethical sentences is (mutatis mutandis) just like that for complex sentences containing exclamations or imperatives--whatever the correct theory is? That is, it seems to me that expressivists have no special difficulty here that others do not also have, namely, the difficulty of providing a compositional semantic theory for complex sentences containing exclamations and imperatives. Once the correct theory is uncovered, an expressivist could just adopt it, with appropriate modifications, as the correct compositional semantic theory for complex ethical sentences. That an expressivist theory may not include such a correct compositional semantic theory is perhaps regrettable, but I don't see why it warrants rejection of that expressivist theory, any more than we would be warranted in rejecting some nonexpressivist theory for not providing a correct compositional semantic theory for complex sentences containing exclamations and imperatives. Expressivism gives rise to no special problem, and so incurs no special obligation to solve the problem.

So, I think that The Challenge from Incomplete Semantics is legitimately directed toward all expressivist theories, and expressivist theories should at least try to meet it. However, I no longer think that an expressivist theory's failure to meet this challenge warrants a rejection of that theory.

(As far as I can tell, the only grammatical contexts in which exclamations and imperatives are problematic are antecedents of conditionals and complement clauses in truth ascriptions ('It is true that ____'), modal claims ('It is possible that ____'), and indirect contexts ('John believes that ____'). So, an expressivist who wants to adopt the correct compositional semantic theory for exclamations and imperatives would have another burden, namely, to explain why ethical sentences are, but exclamations and imperatives are not, so embeddable. I discussed truth ascriptions in the last post. I'll discuss antecedents of conditionals and indirect contexts in future posts.)

Comments

[Jamie]

Interesting.

Here are a few random thoughts.

I guess imperatives (etc.) don't go into 'that' clauses in general. You mention 'true that S', 'believes that S', 'possible that S' -- aside from antecedents, is the imperative-resistant category just the category of 'that S' contexts?

I think imperatives also fail to embed in 'because S' constructions. I say 'I think' because I used to think that interrogatives also fail that embedding, but in fact 'because Q' is fairly common in ordinary English.

I'll be interested to hear why you think imperatives can't be antecedents. My tentative view is that this is a syntactic fact, not a semantic one, but I've recently thought of some evidence that it might be more semantic. (The evidence is that there is a kind of construction in English in which an imperative sentence is made to behave like an antecedent, but then in the resulting complex the imperative looks like it has the semantic value of a declarative sentence.)

[Dan Boisvert]

Hi Jamie. It does indeed look like all 'that S' contexts are resistant to imperatives, exclamations, and interrogatives. For example, it also doesn't look like imperatives or exclamations are embeddable as complement clauses in 'seems that ___', 'said that ___', 'conjectures that ___', and so on. (It also looks like we can add 'wonders whether ____'.) And I think you're right that 'Because ___' is imperative and exclamation resistant, as is '___ so'. So it looks like an expressivist who wanted to adopt the correct compositional semantic theory for complex sentences containing exclamations and imperatives has a pretty hefty burden to explain why ethical sentences, but neither exclamation nor imperatives, are embeddable in these contexts.

On your last point, my view is that pragmatic factors prevent imperatives and exclamations from being antecedents of conditionals, and not semantic and or syntactic factors. (Or, perhaps what might amount to the same thing, I take the pragmatic factor to be the underlying reason that conditionals with imperatives or exclamations as antecedents are syntactically incorrect.) The pragmatic factor is, basically, that having a grammatical construction of a conditional with imperatives or exclamations (or interrogatives) could serve no useful conversational purpose, so we should not expect to find such constructions developing in natural languages. I think I'll make this my next post. For if I am right about this, then it gives rise to what I now think is the most pressing kind of embedding difficulty for expressivists, including my own kind of complex expressivism.

I'm also interested to hear more about your evidence for the reason being semantic, though it may be best to pursue this off-blog. Do you have in mind constructions like 'Play first base and you'll see a lot of action', 'Be good and I'll bring you something special when I return', and 'Give me your money or you'll be sorry'?

[Jamie]

Yes, those last examples are exactly the kind I had in mind. Also interesting about those 'and'-conditionals is that some usual strictures on imperatives are relaxed for the ones embedded there. For example, in general it sounds odd and wrong (not exactly to say ungrammatical, but in some way rejectable) for an imperative to cover something that is obviously not under anyone's control (*"Grow an extra limb"); but this doesn't seem to be a problem in the 'and'-conditional context ("Grow two feet taller and you could be a first rate power forward").

As to "that S" contexts:
I thought 'seems that' might be a propositional attitude construction, and the others are indirect discourse, which maybe you'd already mentioned. But since then I've thought of 'probable that' and 'remarkable that', which aren't attitudes or indirect discourse. (There is a whole family of 'that S' constructions with a kind of dummy subject 'it'.)
My untutored hunch is that 'whether' complements are different -- they seem to me to be rearranged questions (I probably mean interrogatives) that just look like declarative sentences.

I said that my view is that the resistance of antecedents to imperatives is syntactic. I should add that I can't think of any good explanation for why syntax should impose this constraint. So, I am looking forward to your pragmatic explanation.