This is the third of a series of posts in which I try to make clear the different embedding difficulties that, as a family, are thought to present the most pressing objection to expressivism and to distinguish the different kinds of expressivism toward which each difficulty is most forcefully directed. 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). Let’s begin the discussion of the various embedding difficulties by focusing on a rather straightforward difficulty for expressivism, what I call will call the “Objection from Truth Ascriptions.”
The Objection from Truth Ascriptions arises from the fact that we can (and do) embed ethical sentences like ‘Intentionally flying airplanes into tall buildings is wrong’ into truth and fact ascriptions, such as ‘It is true that ______’ or ‘It is a fact that ____’. (I’ll focus on truth ascriptions.) According to expressivism, a proper literal utterance of an ethical sentence is the performance of either a direct expressive illocutionary act or a direct directive illocutionary act, neither of which is an act of describing or representing the world as being a certain way. Now, on a traditionally accepted robust view of truth, we evaluate something (a belief, sentence, proposition expressed, assertive illocutionary act, etc.) as true or false if and only if it (the belief, sentence, etc.) describes or represents the world as being a certain way. Hence, it looks like neither of these acts are truth-evaluable and, by extension, neither are the ethical sentences conventionally used to perform them. However, when we embed ethical sentences into complex truth ascriptions like ‘It is true that _____’, we certainly seem to be evaluating something–a sentence, belief, etc.–as being true or false. So, it looks like expressivism is committed to something that is false.
This objection is quite powerful only when directed toward SNT-expressivism, like Ayer’s emotivism. It seems the only way expressivist theories like Ayer’s emotivism can avoid this problem while retaining the claim that ethical sentences are not truth-evaluable is to hold that such truth ascription embeddings are really a kind of “pretending,” akin to evaluating something in a movie or novel as being true. For example, we often say things like, “It is true that Romeo dies at the end of Shakespeare’s famous play,” though we do not, of course, think that the sentence ‘Romeo dies at the end of Shakespeare’s famous play’ describes or represents the (actual) world as being a certain way. I just don’t see this kind of strategy as being very appealing, for it just doesn’t look at all like, in the case of embedding ethical sentences in truth ascriptions, that we are pretending in a way that we do when we ascribe truth to something that occurs in a movie or play.
The Objection from Truth Ascriptions is not very powerful when directed toward the other kinds of expressivism, since the others accept that ethical sentences are truth-evaluable and, hence, it is not surprising to find that ethical sentences can be embedded within truth-ascriptions. For example, according to Hare’s prescriptivism and expressive-assertivism, CR-expressivist theories, moral utterances and the ethical sentences conventionally used to perform them are robustly truth-evaluable, since they, in part, describe or represent the world as being a certain way. Furthermore, Blackburn’s projectivism and Stevenson’s emotivism, SM-expressivist and CM-expressivist theories respectively, simply reject a robust theory of truth that the objection assumes, leaving these theories immune from the objection–at least until a verdict is rendered in the dispute between robust and minimalist theories of truth.
The Objection from Truth Ascriptions leaves us, I think, with at least one important moral: if an expressivist theory is to avoid the difficulties arising from the possible embedding of ethical sentences within truth ascriptions, yet is not a complex expressivist theory, then this expressivist theory seems to be forced into accepting a controversial view of truth, namely, truth minimalism (or, more cautiously, at least truth minimalism about ethical sentences)–a fit that may not be very comfortable, absent some compelling independent reasons for holding truth minimalism.
Dan, I’m having a little trouble with the second paragraph of your posting. It looks to me like you might be relying on some kind of infallibility of ordinary speakers — that might be defensible but I don’t think it can be taken for granted. Let me focus on this:
“Now, on a traditionally accepted robust view of truth, we evaluate something … as true or false if and only if it … describes or represents the world as being a certain way.”
It seems a lot more plausible that we evaluate something as true or false only if (not “iff”!) we think it describes or represents.
Even that weaker claim is far from obvious to me. Any competent English speaker will be willing to call such sentences as “Whoa, that tie is awesome!” true and false. (True, anyway. People almost never use false in ordinary spoken English, as Brian Weatherson pointed out to me.) Unless we already accept some kind of truth minimalism, we aren’t going to think that the ordinary language considerations demand that we come up with a theory according to which calling something “awesome” describes it as being a certain way.
Hi Jamie. Thank you very much for the helpful comments. You are absolutely right that my original claim is far too strong. Moreover, I created needless confusion by slipping into talking about *speakers* evaluating something as true or false rather than *sentences*, etc. being truth evaluable.
I should have said this:
On a traditionally accepted robust view of truth, something (a belief, sentence, proposition expressed, assertive illocutionary act, etc.) is truth evaluable only if it (the belief, sentence, etc.) describes or represents the world as being a certain way. Hence, it looks like neither direct expressives nor direct directives are truth-evaluable and, by extension, neither are the ethical sentences conventionally used to perform them. However, since ethical sentences are embeddable in complex truth ascriptions like ‘It is true that _____’, it certainly seems that the embedded ethical sentence is evaluable as true or false. So, it looks like expressivism is committed to something that is false.
This version is clearly a more accurate representation of the intended objection, so I thank you for forcing me to revise my original. I also think that this revised version has the same implications for the different kinds of expressivism that I mentioned in the last two paragraphs of the post: it spells trouble for SNT-expressivism, little trouble for SM- and CM-expressivism (since each rejects the view of truth assumed in the objection), little trouble for CR-expressivism (since it is amenable to ethical sentences having robust truth conditions), and one important lesson to take from this objection is that if an expressivist theory is to avoid the difficulties raised by the objection, yet is not a complex expressivist theory, then that theory seems to be *forced* into accepting truth minimalism (at least for ethical sentences), an uncomfortable result if one is not already sold on the independent reasons for accepting truth minimalism.
Good.
I am still worried about something similar, though.
However, since ethical sentences are embeddable in complex truth ascriptions like ‘It is true that _____’, it certainly seems that the embedded ethical sentence is evaluable as true or false.
Would someone who accepts what you call a ‘robust view of truth’ really admit that any sentence embeddable in complex truth ascriptions are truth evaluable? I would have thought that it was symptomatic of a deflationary conception to draw that inference.
Should even typical deflationists draw that inference? This seems more like a commitment to a deflationary conception of “true” or truth-aptitude than to a deflationary conception of truth.
Someone could have a deflationary conception of truth, but reject the deflationary theory of truth-aptitude. This person may think that our use of the truth predicate is captured adequately by the equivalence schema, but still think that it takes more than a sentence having the right sort of syntax for it to be truth-apt — perhaps it must be descriptive and not evaluative for it to be truth-apt. This person has a deflationary conception of truth, but a more robust conception of truth-aptitude. This seems to be Ayer’s view. He presents his emotivism in chapter VI of LT&L, right after chapter V where he defends the deflationary theory of truth.
Likewise (as Jamie suggests?), someone could have a robust view of truth and of truth-aptitude. The point is that it doesn’t seem like the deflationary theory of truth has anything to do with different views about what sorts of sentences are actually truth evaluable.
Hi Jamie and Kyle. Thank you again for your comments, which, as I read them, suggest to me that the Objection from Truth Ascriptions is even weaker than I thought. Another possibility, I suppose, is that I am just not understanding the objection. But I don’t see what else the objection could be. It seems clear that the objection wants to say that expressivism entails that ethical sentences are not truth evaluable, since they have no descriptive or representational content, but that this entailment must be false, since ethical sentences are embeddable in truth ascriptions. So, let’s spell out in more detail what we can of this argument
Strong Version
1. Assume that expressivism is true
2. Assume the following robust view of truth evaluability: something is truth evaluable only if it describes or represents the world
3. If expressivism is true and the robust view of truth evaluability is true, then ethical sentences are not truth evaluable
4. Therefore, ethical sentences are not truth evaluable (1-3)
5. Ethical sentences can be embedded within truth ascriptions
6. If a sentence can be embedded within truth ascriptions, that sentence is truth evaluable
7. Therefore, ethical sentences are truth evaluable (4, 5)
8. We shouldn’t give up our robust view of truth (presumably some independent arguments)
9. Therefore, expressivism is false (1, 2, 4, 7, 8)
Here are the defenses for expressivism, the first being the reason why I read your comments as suggesting that the objection is worse than I thought.
• This objection equivocates between a robust and minimalist theory of truth evaluability. If we hold constant for a robust theory of truth evaluability, then (6) is false, since (6) can only be true on a certain kind of minimalist theory of truth evaluability (syntacticism). However, if we hold constant for this kind of a minimalist theory of truth evaluability, then (6) is true, but we cannot then assume (2) or (8). So the objection fails even to get off the ground. And there are other possible defenses for SM-, CM-, and CR-expressivism.
• It is open to SM- and CM-expressivists to reject either (2) or (6). They can reject (2) if they want to accept a minimalist theory of truth-evaluability, and they can also reject (6) if they accept a certain kind of minimalist theory of truth evaluability (disciplined syntacticism).
• I don’t see how a CR-expressivist, or at least one who accepts some kind of correspondence theory of truth, can reject (2), since it seems that, if truth is some kind of correspondence between the world and some item, the only kinds of things that can be truth evaluable are those things that describe or represent a possibly corresponding world. However, a CR-expressivist would clearly reject (3), since, according to CR-expressivism, ethical sentences are, in part, descriptive.
Perhaps the objection can be made stronger by weakening (6), (7), and (9) in the following ways:
6′. A sentence’s being embeddable within truth ascriptions is good evidence that that sentence is truth evaluable—after all, the ascription ‘it is true that’ certainly seems to be evaluating the embedded sentence (or the proposition or belief that it expresses) as true.
7′. Therefore, it is very likely that ethical sentences are truth evaluable
9′. Therefore, it is very likely that expressivism is false or that a robust view of truth evaluability is false (probable incompatibility between (4) and (7′).
This weaker form of the objection is not incoherent, so I think this formulation of the objection is probably the most charitable. Still, it looks like CR-expressivists can still reject (3), and SM- or CM-expressivists can still reject (2). But SNT-expressivism, since it entails that ethical sentences are not even truth evaluable, has to meet a challenge, namely, that truth ascriptions appear to be evaluating the embedded sentence (or the proposition or belief it expresses) as true—so what is going on in these cases, if in fact the embedded sentence (or the proposition or belief that it expresses) is not truth evaluable? I can think of only three responses, none of which seems very satisfactory. The first is the “pretending” answer that I described above. The second is that ‘it is true that’ is ambiguous in English, at times evaluating a truth evaluable item as true, at other times (when the embedding sentence is an ethical sentence or a sentence like Jamie’s ‘Your tie is awesome’) doing something else, like registering agreement with an attitude, or something of that sort. The third is to claim that ‘it is true that’ is univocal but, rather than evaluating the embedded item as true, it does something more general, like evaluating the embedded item as having its positive evaluation, whether that be fulfillment, felicitousness, or whatever kind of more general evaluation would subsume ‘true’.
So, I think the strong version of the objection poses no threat to any kind of expressivism, since it equivocates between a robust and minimalist account of truth evaluability. SM-, CM-, and CR-expressivism also have other replies at their disposal. The weak version of the objection poses a threat for SNT-expressivisms, like Ayer’s and Gibbard’s (as articulated in WCAF). Whether it poses a threat for SM- or CM-expressivism depends on whether minimalism about truth evaluability is plausible. I don’t see how it poses a threat at all for CR-expressivism.
Thanks again for helping me to think through these issues.
Oops. Disregard the ‘or that a robust view of truth evaluability is false’ in (9′) above.