Sentimentalism is an account about the meaning of some evaluative terms in natural language. In general, the sentimentalist about some evaluative term E proposes to analyze E in terms of the notion of fittingness of some emotion. So, for example, ‘x is shameful’ gets analyzed as ‘it is fitting to feel shame at x‘. Fittingness in turn gets analyzed in terms of the truth of the representational content of the relevant emotion. That is to say: shame at x is fitting just in case x-directed shame represents x truthfully. (And, in general, an emotion with representational content p is fitting just in case p.)
The sentimentalist analysis of ‘shameful’ avoids making a mistake that other accounts of the meaning of ‘shameful’ have made. Say that we propose to analyze ‘x is shameful’ as ‘x‘s bearer ought to / should be ashamed of x‘. This is clearly wrong. It is possible that John should be ashamed of smoking — say Osama will blow up India unless John’s ashamed of smoking — even though his smoking is isn’t itself shameful.
This is all seems pretty much correct. Still, the following sentences sound pretty odd to me.
(1) #Smoking is a shameful habit, but it’s not like you ought to be ashamed about it.
(2) #John’s behavior is shameful, and he shouldn’t be ashamed about it.
(3) #That moldy piece of bread is disgusting, but don’t be disgusted by it.
It does not help to observe that it is fairly easy to fill in the conversational context in a way that appears to alleviate the intuitive oddness of these sentences. (For example, it is easy to imagine someone following up an utterance of (1) with the following addition: “After all, shame only makes it harder for you to do what you need to do—quit smoking.”) The oddness is something that needs to be alleviated, and it so it’s still explanandum for the sentimentalist, since his theory seems to predict that a conjunctive utterance like, e.g., (1) should be a perfectly ordinary way of asserting the following claim:
(1’) It’s fitting to feel shame at smoking, but it’s not like you ought to be ashamed about it.
There is no oddness about (1’) once we remember that ‘fittingness’ is a technical notion with a special meaning for the sentimentalist. According to the sentimentalist, it is fitting to feel shame at smoking just in case the representational content of smoking-directed shame is true. It is plausible to understand the ‘ought’ of the second conjunct as something like an all-things-considered, normative-reasons ‘ought’. That is to say: the two conjuncts of (1’) assert different, entirely consistent claims. If we understand the meaning of (1) along the lines of (1’), then we appear to predict that there should be nothing odd about any utterance of (1)—including those that are unaccompanied by some sort of extra “after all…” claim. And this prediction appears to be false.
I think the sentimentalist needs to beat a retreat to the pragmatics to construct a plausible reply. Here’s how that reply might go: the meaning of ‘shameful’ is everywhere what the sentimentalist claims it to be. The oddness of (1) is not a consequence of any sort of lexical or semantic ambiguity, but of a presuppositional confusion generated by the conjunction of the relevant sentences. This is because, in general, an utterance of any claim that invokes any kind of normative language conversationally implicates the relevant all-things-considered normative reasons claim. Conversational implicatures generate pragmatic presuppositions. (Following Stalnaker, it is pragmatically presupposed that p in a conversation C just in case every member of C treats p as true, believes that every member of C treats p as true, and so on.) According to this story, an utterance of (1) first generates a pragmatic presupposition that feeling shame about smoking is appropriate (in the all-things-considered sense), which is immediately canceled by the sub-utterance involving the second conjunct. So the explanation of our intuitive judgment of (1)’s oddness goes like this: though there is, strictly speaking, no contradiction involved in an utterance of (1), it generally generates a pragmatic presupposition that is incompatible with the content of its second conjunct.
I think it is plausible to suppose that there is this sort of conversational implicature in conversations involving the exchange of normative claims. For one, the proposal is able to explain the default intuitive oddness of both (4) and (5). (Note: as with (1)-(3), though it isn’t difficult to alleviate the oddness of (4) and (5), still, something has to be done in order to alleviate it!) [*]
(4) #Yu-chen’s belief that her house is on fire is justified, but she shouldn’t believe that.
(5) #It makes sense for Vanessa to move to New York, but she shouldn’t do it.
For two, there seem to be fairly clear Gricean reasons for supposing that this kind of implicature actually exists. According to Grice’s Maxim of Quantity, a speaker should say no less and no more than required by the demands of conversation. It seems plausible to suppose that, in general, in conversations involving the exchange of normative claims, the conversation demands the dispensation of good evaluative advice by speaker to addressee—that is to say, the best advice that it is possible for the speaker to dispense to his addressee, given his epistemic limitations. A speaker violates the Maxim of Quality by dispensing advice that he does not take to be good advice. If sentimentalism about ‘shameful’ is correct, a speaker’s utterance of ‘smoking is shameful habit’ advises his addressee to feel shame at smoking. If he does not take this to be good advice—perhaps because he thinks that shame only makes it harder for you to do what you need to do—then he should not have dispensed it. In dispensing it, the speaker therefore implicates that he takes it to be good advice. That is to say, he implicates that feeling shame about smoking is appropriate (in the all-things-considered sense).
[*] The nice thing about conversational implicatures is that they are easily cancelable or defeasible. Thus it should be fairly easy to construct a case where, e.g., there is no feeling that an utterance of (1) would be at all odd. And indeed it is fairly easy. For example, wherever it is common ground that the speaker will follow his utterance of (1) with an explanation of why it is all-things-considered bad to be ashamed about one’s smoking—an explanation that does not bear on the fittingness of feeling shame at one’s smoking—it seems plausible to suppose that the relevant implicature is canceled. It is sufficiently clear to every participant in the conversation that the speaker is not advising anyone to be ashamed of smoking.