Jason Bridges

University of Chicago

Phil340, W02-lecture 12 notes

Schiffer’s conventionalist account of language



I. Setting the stage

Kripke’s topic was how a semantic property, namely the reference, of a particular type of expression is determined.

Today we going to begin going to step back and ask a more sweeping question.  What kind of object is a natural language?

On the syllabus, I call this topic the ‘ontological status of natural languages’ because ‘ontological status’ is a fancy way of talking about questions about the kind of object a given item is.  Ontology is the study of what kind of objects there are.  Here our concern is more specific: what kind of object is English, or French, or Mandarin?

One possible answer is that they aren’t real objects at all, that strictly speaking, there isn’t such a thing as English.  As will see, there is not simply one way of opposing this nihilistic position.  Various views agree that languages exist, but differ over what precisely they are.


For the moment, we’ll keep this topic in the background.  I will not place what I say about Schiffer today explicitly in relation to the ontological question.  When we get to Wiggins Davidson, we will be in a position to think about Schiffer’s view from this angle.


For now, I’ll discuss Schiffer’s view from his own perspective—that is, from the perspective of the particular problem which, given his larger commitments, he is attempting to solve.


Schiffer’s project: To construct an account of utterance meaning (and ultimately sentence and word meaning) on the basis of a Gricean account of utterer’s meaning


Remember the distinction between what a speaker or utterer means and what the words he uses mean (the latter is sometimes identified, as we noted earlier, by talk of what words literally or explicitly mean.)

Recall our example: In answering the question, “How do you like my play?” by saying, “That actor has a good walk,” I mean both that the actor has a good walk and that the play stinks.  But what the sentence itself means is only that the actor has a good walk.

Schiffer’s going to take the Gricean account of utterer’s meaning and build upon it an account of an utterance’s meaning, e.g., a sentence’s meaning


Before we get into the nitty-gritty, a couple of points.

Recall that Grice uses the notion of an utterance in an extended sense.  Ordinarily we think of an utterance as involving the production of a sentence in a language.  But Grice employs the following technical meaning:


An utterance is any action in the doing of which the agent meansnn something.

[Note that ‘nn’ subscript means ‘non-natural’, and say a bit to bring back to mind the natural/non-natural distinction.]


As we’ve discussed, almost any type of action, not just the production of a sentence, can meannn something.  Even not doing anything can mean something.  If you ask me what I think of the play and I just stare at you in grim silence, my meaning is clear.


Now Schiffer thinks that the difference between what a speaker means by an utterance of a sentence and what the sentence means is one aspect of a broader distinction that applies to the whole broad Gricean category of utterance.

Just as “That actor has a good walk” means that that actor has a good walk whatever a speaker means by a particular utterance of it, so, say, giving someone the finger means what it does regardless of what a speaker means by a particular instance of it, and so, say, waving a white flag in a certain context means, ‘I surrender’, regardless of what else the speaker might mean by it.  And so on.


So we have to distinguish quite generally:

Utterer’s meaning: What a person means by an utterance.

Utterance meaning: The meaning of an utterance-type.

[Say a bit about type/token ambiguity of “utterance”, and why it is types that are the primary bearers of utterance meaning.]


The following are simply special instances of these categories:

Speaker’s meaning: What a person means by an utterance of a sentence

Sentence meaning: The meaning of a sentence

[The word “sentence” is generally read as applying to types, so we don’t need to add ‘-type’ here.]

Word meaning: The meaning of a word.


We’re only going to discuss one piece of Schiffer’s complex account of meaning, namely, the account of utterance meaning he constructs using his account of utterer’s meaning.

He follows this by giving an account of sentence meaning in particular, and from there gets to word meaning.  But we’re not going to talk about these parts of his view.  We’re only looking at one aspect of the story: his account of what he calls non-composite whole-utterance types, that is, his account of how utterances that lack a compositional structure get their meaning.  Think of one-word sentences or signals or gestures.



II. The account

So Schiffer’s aim is to give an account of utterance meaning on the basis of utterer’s meaning.  A special case of this will be to give an account of sentence meaning (sentences comprising a subset of utterances) on the basis of speaker’s meaning (speaker’s meaning just being utterer’s meaning when what is uttered is a sentence rather than a gesture or signal or other non-linguistic item).


Three terminological points in reading Schiffer:

1.      Schiffer’s account of utterer’s meaning has a strong Gricean flavor , but it’s a bit different.  It’s even more complex than the account Grice ended up with in Studies in the Way of Words.  We’re not going to get into the details of Schiffer’s account of utterer’s meaning here.
      Schiffer uses the phrase, “S M-intends that A believe that p” as equivalent to “S intends that A believe that p, and moreover has the additional package of higher-order intentions necessary and sufficient for S to thereby mean that p.”  For our purposes, we can just plug Grice’s relatively simple “Meaning”-era view in here.

2.      “Activated belief” means belief brought to mind.  It is designed to respond to that old objection that we can’t intend someone to form a belief that she already has.  The thought is that we can intend to make such a belief “active”.

3.      “A and S mutually know* that p” means that A and S both know that p and know that the other knows that p and know that the other knows that she knows that p, ad. infinitum.  This idea extends in a straightforward way to the case of more than two individuals: see the text.  The notion is introduced to deflect one type of counterexample to Gricean accounts of utterer’s meaning, a type we discussed when reading Grice, as well as analogous counterexamples that come up in the context of conventionalist accounts of utterance meaning.


One way to work our way into the motivation for Schiffer’s account is to dwell on the fact, already noted, that some utterance-types have a meaning they carry with them regardless of the intentions of the utterer, and some do not.

If you flip your finger at me, even if you meant that I should have a nice day, you’re still guilty of doing something that carries an offensive connotation.

But take a case in which I stare at you after you ask me how the play was, with the meaning that the play’s no good.  Here we don’t what to say that the looking itself carries some standing meaning.  It’s certainly not the case that there’s some particular thing that’s ‘said’ whenever someone looks at someone.

So what divides these cases?


Schiffer proceeds from the intuitive thought that utterance meaning, as opposed to utterer’s meaning, is conventional.  At this stage, he’s just thinking of the ordinary, unelaborated notion of convention, where a convention is something like a commonly accepted practice or agreed upon pattern of behavior.  For example: there’s a convention in certain upper-crust social strata to have a bunch of different forks at dinner and use this one for eating the salad and this other for eating the entrée.  There used to be, as Schiffer points out, a widespread convention for men to open doors for women.  And there’s a very stable convention in this country to drive on the right.  In England, they have a different convention with respect to this activity.  (The last are not ‘mere’ conventions: they’re governed by law.)

Schiffer’s intuitive thought is that the difference between an utterance-type  that has a meaning (finger-flipping) and an utterance-type  that does not (staring) lies in there being a convention for the former to be used in a certain way and for the latter not.

He wants to say that this applies to linguistic utterances in particular.  It’s a matter of convention that words and sentences mean what they do.  It’s not a matter of convention what a speaker means; this is a matter of the intentions of the moment.  But it is conventional what the uttered sentence means.


(I am going to shave away some of the more complicated details of Schiffer’s presentation in what follows, and reorganize his convoluted presentation.)


First we need an analysis of the notion of convention, a refinement of the ordinary idea:

Schiffer on convention

There exists a convention in a group G to do A iff members of G know* that:

1.      There is a precedent or explicit agreement in G to do A.

2.      Because of 1, members of G expect members of G to do A.

3.      Because of 2, members of G do A.


So let A be driving on the right side of the street.  Then there is a convention to drive on the right side of the street in our community iff members of our community know* that there’s a precedent or agreement to drive on the right side of the street, that members of G expect people to drive on the right side in light of that precedent or agreement, and that people do in fact drive on the right side of the street in light of this expectation.  Does this seem right?  Maybe.

To take another example, let A be greeting someone by waving one’s hand.  Probably there was never an explicit agreement to do so, but by now there is certainly a precedent for doing so.  People have been doing so for a while.  And 2 and 3 seem, prima facie, roughly true, though perhaps it’s hard to verify the claims about the explanatory linkages between 1, 2, and 3..

(Note that this doesn’t mean that there can’t be conventions concerning other ways of greeting.)


Now we can give Schiffer’s account of utterance meaning:

An utterance-type x means “p” among the members of a group G if there exists a convention in G to mean that p by uttering x.

(Note that it’s utterance meaning at issue in the consequent and utterer’s meaning in the antecedent, so the condition is not circular.)


A couple of points:

1.      This is a sufficient but not necessary condition, according to Schiffer.  (That’s why I put “if” instead of “iff”).  There don’t have to be full-blooded conventions in his sense in order for an utterance-type to mean something.  He briefly discusses a weaker alternative.  He ends up calling what I’ve put on the board an account of specifically the cases in which an utterance-type means “p” in virtue of its being a ‘conventional means’ of meaning that p.

2.      On the face of it, the account seems like it might fit our intuitive sense of which utterance-types have meaning and which don’t.  Consider various cases.
        A. Is there a convention in Schiffer’s sense to mean, by looking at someone, that the play we’re watching is no good?  Clearly not.  But neither do we take that utterance-type to have that meaning.
        B. Let G be the community of chess players.  Might it not be prima facie plausible that Schiffer’s account explains why knocking one’s king over means “I concede”?
        C. Is there a convention in Schiffer’s sense to mean that that actor has a good walk by uttering the sentence, “That actor has a good walk”?
        This is a tricky case because Schiffer is ultimately going to claim that this sentence’s meaning is built out of conventions governing the usage of the words.  But suppose we pretend this sentence lacks compositional structure, and so that we have a non-composite whole-utterance type in play.  Then we might suppose we could answer, “Yes”.
        D. Is there a convention to mean that the play we’re watching is no good by uttering the sentence, “That actor has a good walk”?  Of course not.  But neither does that sentence have that meaning.
        Thus we see why this approach might be thought to provide an account of sentence meaning on the basis of speaker’s meaning in a manner that accords with our intuitions about the literal meaning of sentences (as opposed to what an utterance of a sentence implies or intimates on a given occasion).

3.      Now, one reason there might be a precedent for doing B by doing A is that there was in the past an explicit agreement to this effect.  This surely happens sometimes.
      But clearly this did not happen with the conventions, such as they may be, governing the use of natural language.  Indeed, if we thought this held generally of linguistic conventions, we’d encounter a regress.  In what language were the first conventions laid down?
      This is why it’s important to Schiffer that his account of convention leaves open that there was no explicit agreement, but rather simply a precedent, and that this precedent became widely recognized and thus led to expectations and thus led to conformity to the precedent.  Conventions can arise organically, as it were, without anyone explicitly laying them down.
      Is it plausible to think that conventions to mean things by utterance-types could arise in this way?  Schiffer thinks so, and he tells a little story to illustrate how.
      The basic idea in his story is that one can start out trying to convey to someone that p by producing something that ordinarily has the natural meaning that p.  Schiffer uses the example of “grr”, which when produced by a dog indicates (meansnn), that the dog is angry.      But let’s take a different example: suppose some predator makes the noise, “Rarr.”  Then if one day I wish to communicate to you that a predator is coming that you’re not yourself in a position to observe, I may myself go, “Rarr”, hoping you’ll make the connection.  If it works, then it’s likely to be used again in the same circumstances.  This time, however, the reason will not be, or at least will not wholly be, that this is the sound that the predator makes, but rather that both parties, you and I, know that there is a precedent for meaning that a predator is coming by going “Rarr”.  We both know, and know that the other knows, that this has worked before.  So we do it again, hoping the other will catch on.  Eventually the natural connection may be forgotten.
      In fact the natural-meaning element of Schiffer’s sort of story seems dispensable.  I may on some occasion say “Yikes” while running past you.  A few moments later it emerges that I’m being chased by a lion.  I said “Yikes” for no reason: it just burst out of me.  It has no natural meaning.  But maybe I know you heard me do it, and so the next time a lion is around I intentionally go, “Yikes,” intending that you believe that there’s a lion around on the basis of your realization that I intend you to realize there’s a lion around.  Thus I mean that there’s a lion around by uttering “Yikes”.  If this works it might become self-perpetuating.