One of the more serious attempts to provide an expressivist
semantics for moral terms (broadly construed) is Gibbard’s.  The basic idea is that they express
plans.  “I should pack” expresses my plan
to pack; “You should pack” expresses a contingency plan for the (unlikely) case
that I am you.  “Journal editors should
move papers along faster” expresses a plan for being a journal editor.  And  so
on.  Call this view the Gibbard
Semantics, or GS.

It seems to me that there are other possible semantics along
this line.  Consider plans for what I
would have someone do, if I were in control. 
For example, if I were in control of the government, I would have the
government give philosophy professors large cash subsidies.  If I were in control of journal editors, I
would have the editors accept all my manuscripts without revision.  Etc. 
Perhaps I could express these plans somehow.  Perhaps like this:  “The government should give philosophy
professors large cash subsidies; editors should accept all my manuscripts
without revision.”  In the normal case, I
am in control of myself, so “I should…” sentences would express normal
plans.  Call this the Nietzsche Semantics
for moral terms, or NS.


Clearly, NS is not equivalent to GS.  One is likely to indulge considerably more
temptation to exploit others under NS than under GS.  For example, if I were in control of my
neighbor, I would have him mow my lawn for me; under NS I would therefore
claim, “My neighbor should mow my lawn.”   On the other hand, if I were my neighbor, I
don’t think I would mow my (that is, HW’s) lawn, so under GS I would not claim,
“My neighbor should mow my lawn.”


Suppose that much is right. 
My next point is that, if expressivism is true, it is an empirical
question which states of mind people are expressing when they make
should-claims.  That is, whether GS or NS
or some other S gives the meaning of ‘should’ and cognate terms, depends on
which mental states correlate with what should-talk. 


However, there is no particular reason to think we’re all on
the same page.  Some people might express
the plans they have for being journal editors; others might express their plans
for journal editors under their control. 
(In general, if you think there is anything to a “hermeneutic of
suspicion” or “ideology” then you should think that people sometimes deploy
NS-ish should-talk.)  Moreover, there is
no reason that an individual person should be consistent with all uses of ‘should’
and cognate terms.  Sometimes he might
express one kind of state, sometimes another. 
And he might have hefty unconscious motives for disguising this fact
from himself.


All that means, I think, that an expressivist semantics
makes it highly empirically likely that we talk past one another, morally
speaking, much of the time.  The journal
editors say, “Journal editors should be impartially rigorous” and the grad
students say, “Journal editors should go easy on young people,” and, if they
are expressing different mental states, there is no disagreement.  Similar problems can be generated for
reasoning, especially public reasoning between two people.  This is Frege-Geach with a vengeance.


I think the source of the problem lies in the fact that we
normally think that meanings are public, or determined by something about a
whole linguistic population, while an expressivist semantics is committed to
the view that meanings are private, determined by what mental states an
individual is expressing.  Given that,
the danger of talking past one another is quite real, whereas we normally would
not think it would be. 

4 Replies to “Nietzschean Expressivist Semantics

  1. Hi Heath,
    Thanks for the interesting post.
    Can you say more about why you think an expressivist is committed to the view that meanings are “determined” by what mental states an individual is expressing? An expressivist like Gibbard does hold that mental states are the semantic values of sentences. But does this commit the expressivist to the view that these semantic values are determined by a speaker? Consider the mental states that would be assigned to sentences like ‘Thank you!’ or ‘Damn you!’. I’m assuming that the semantic values that would be assigned to these sentences–say, gratitude and frustration respectively–are not determined by any particular speaker, but rather by some kind of community convention. That is, it is a community convention that an utterance of ‘Thank you!’ is the expression of gratitude. Couldn’t expressivists say the same thing about the semantic values of normative claims?

  2. I would also recommend reading Mark Schroeder’s ‘Expression for Expressivists’ paper. If I remember this right, Mark argues against the views according to which, for the expressivists, the moral utterances express whatever mental states speakers are actually in. Roughly, the idea is that instead the expressivist should hold that moral utterances express mental states in which the speaker should be in if she were following the norms of assertion of the moral discourse. The psychological part of expressivism then provides an account of what these right kind of mental states are. This is why the expressed mental states would be GS rather than NS even if the speaker were in NS mental states. And, this is also why expressivism makes meaning public rather than something dependent on the individual.

  3. Dan and Jussi,
    I think you both point out a possibility I did not take adequate account of, namely that the semantic values of moral terms might be mental states but that which mental states they are is not determined by what’s in the speaker’s head at the moment. And given the possibility of insincere moral utterance, this seems important and right.
    So here’s perhaps a better way to make the point. I take it that NS is as good a semantics, for some possible kind of discourse, as GS is; NS and GS have the same expressive power. So what determines whether I or you or we are speaking GS or NS when we use moral language? Dan suggests community convention and Jussi suggests norms of assertion and for simplicity I will go with the latter.
    One might start by asking, What determines the norms of assertion? If it’s very democratic and statistical—we count up moral assertions and check which mental states are associated, majority wins—then it’s far from clear to me that GS comes out on top. Even if it did come out on top in “our” community, this would be a function of our sincerity and good-heartedness, and if there were some other more devious community where NS came out on top I’m not sure we could avoid talking past them.
    But that whole line of thought makes no sense on an expressivist understanding. Norms of assertion are just norms, so if I say what the norms of assertion are (e.g. I argue for GS over NS) I am just expressing my plans for using moral terms…or maybe I am expressing my plans for how you are to use moral terms. As soon as we go normative in our expressivist semantics, there will be no fact of the matter what the semantics is, rather the semantics expresses a plan (of some kind) for using moral terms. Which plan that is, is precisely what’s at issue.
    I’m not sure how bad that is, though at the moment I would say it is bad. Gibbard canvasses something like it under the rubric of “norms all the way down.”

  4. Hi, Heath.
    There are excellent reasons to distinguish between so-called ‘political’ and ‘agential’ senses of ‘ought’ or ‘should’. For example, the ‘the meeting ought to start by noon’ is most naturally interpreted as saying that a certain state of affairs ought to be the case – namely, that the meeting starts by noon. But there are good reasons to suspect that ‘Bill ought to start the meeting’ does not just say that it ought to be the case that Bill starts the meeting. For more on this distinction, see Broome on ‘ought’, the relevant chapters of Wedgwood’s The Nature of Normativity, or my ‘Do Oughts Take Propositions?’.
    The question that we should be asking, is which of these, if either, Gibbard’s plan-theory is supposed to account for. I suggest that it is the agential ‘ought’, not the political ‘ought’. Your Nietzschean proposal, in contrast, fits better with the political ‘ought’. For example, one of the tests for the agential ‘ought’ is to compare sentences like ‘Bill ought to kiss Lucy’ and ‘Lucy ought to be kissed by Bill’. Your Nietzschean proposal would predict that someone who utters the former is expressing a plan to, if she had control over Bill’s action, get him to kiss Lucy. It is plausible that it is incoherent to have such a plan, and not also plan, if one had control over Lucy, to get her to be kissed by Bill. So it makes sense, on the Nietzchean approach, for these two sentences to entail each other. In contrast, on the Gibbard approach, it makes perfect sense to endorse one but not the other. You just plan, if you are Bill, to kiss Lucy, but if you are Lucy, to avoid Bill like the plague – nothing incoherent about that.
    In other words, I don’t see why there should be anything problematic about the possibility of a potential ambiguity, here; in real life normative words have different meanings from one another and sometimes are even ambiguous, and these two proposals seem to match up with a real ambiguity.

Comments are closed.