Grice has a two-stage project: (a) explain speaker meaning in terms of other mental states, especially intentions and beliefs; (b) explain sentence meaning in terms of speaker meaning.
A. Grice's definition (as simplified by Lycan):
By uttering x (to an audience A), S meant that P if and only if:
- S uttered x intending that A form the belief that P, and
- S intended that A recognize that (1), and
- S intended that A form the belief that P (in part) because of A's recognition that (1)
B. Objections that the Conditions are Not Necessary
Objection 1: No Audience Required
It seems that I can meaningfully utter a sentence without having an audience. Examples:
- talking to yourself (I do this in the car all the time)
Grice's response: counterfactual intentions. S utters x intending that if there were an audience A, they would form the belief that P; they would recognize that S intends this; and their forming P would be because they recognize this.
(One might suggest that in these cases S = A, but it doesn't seem that in most cases S intends to produce beliefs in him- or herself.)
Objection 2: No Intention to Produce Belief by Means of Intention
- Intend to produce belief, but not because of this very intention: conclusion of argument. I present an argument with conclusion C. I want A to form the belief that C, so condition 1 is satisfied. Probably condition 2 is also satisfied. But condition 3 seems not to be satisfied. I don't want my audience to believe C because they recognize my intention that they come to have this belief; rather, I want them to come to believe it because the argument is so compelling.
- Don't intend to produce belief at all: answering questions during an oral exam. The examiner already has the relevant beliefs, so there is no intention to produce them. In this case condition 1 is not met.
Response: weaken the conditions again. Perhaps S intends only that A should form the belief that S believes P.
C. Objections that the Conditions are Not Sufficient
Objection 3: Uttering Nonsense
Ziff: "Ugh blugh blugh ugh blugh." Uttered to express contempt. S intends that A form the belief that S feels contempt; S intends that A recognize that S has this intention; and S intends A to form the belief that S feels contempt because they recognize S's intention that they form this belief. Nevertheless, it doesn't seem that S's utterance means that S feels contempt.
(You could say that S is communicating his or her contempt, but not by saying something that means "I feel contempt." Compare: I hit my thumb with a hammer and say "ouch" I may communicate that I'm in pain, but "ouch" does not mean "I'm in pain": it expresses pain, it doesn't state that I'm in pain.)
Objection 4: Kennst du das Land . . .
The American soldier who tries to convince his Italian captors that he is German by uttering the only sentence of German he knows. Satisfies the conditions (with P = "I am a German soldier") but he doesn't seem to mean that he is a German soldier.
General Issue about the Definition of Speaker Meaning
The general idea that what a speaker means must be explainable in terms of the speaker's intentions and beliefs seems plausible (to me, anyway!). What is strange about Grice's view is the attempt to explain speaker meaning without any reference to the semantic meaning of the sentences used.
Of course, that's because Grice wants to explain the linguistic meaning of sentences in terms of speaker meaning, so to avoid circularity he can't use linguistic meaning in his explanation of speaker meaning. But isn't there something backward about this? Well, let's see . . . on to the definition of sentence meaning!
A. First attempt at a definition:
A sentence E of a natural language L means that P if and only if, when speakers of L utter E, they normally speaker-mean that P.
B. Obstacles to the Definition
Obstacle 1: Glyting elly beleg and Gleeg gleeg gleeg
Q: what's the point of this "obstacle"? What Lycan seems to have in mind is that in fact sentence meaning constrains speaker meaning: I can't express the proposition that P by uttering sentence S unless S actually means that P. (I may be able to express or communicate P by using S, but P still won't be what S means when I utter it.)
How is this a criticism of the above definition of sentence meaning? I think L's point is that it seems backward: it seems that sentence-meaning constrains what speakers can mean; if so, then it looks as though trying to define sentence-meaning in terms of speaker-meaning goes the wrong way around, or is circular.
Obstacle 2: Sentences that are Never Uttered
Most sentences of English, for example, never have been and never will be uttered. So there's no way to define their meaning in terms of what people usually speaker-mean by them.
Obstacle 3: Novel Sentences
The first time a sentence is uttered, its meaning is already determined: we don't need to wait to find out what speaker-meaning is statistically most often associated with it.
(At least, I think this is Lycan's point with this objection.)
Obstacle 4: Nonliteral Uses
Metaphor, colloquialisms, hyperbole, etc. -- we often use sentences to communicate things other than their literal meanings. But then if we tried to analyze sentence meaning as what people usually speaker-mean by a sentence, we might get the result that sentences mean something other than their literal meaning.
C. Revising the Definition
An unstructured expression x means that P in S's idiolect if and only if S has in his or her repertoire the following procedure: to utter x if for some audience A, S intends A to believe that S believes that P.
An unstructured expression x means that P in the dialect of a group G if and only if:
- many members of G have in their repertoires the procedure of uttering x if, for some A, they want A to believe that they believe that P;
- they will continue to retain this procedure in their repertoire only if other members of G have the same procedure.
A structured expression x means that P in the dialect of a group G if and only if . . . ?????
(At this point Grice appeals to a resultant procedure that is constructed somehow out of the basic procedures so far mentioned. But the resultant procedure has to be abstract, because it will never be realized for most sentences of the language.)