G. Pynn on Grice

From his online document, on

"Grice’s Theory of Implicature"


What is said vs. what is meant. Grice’s example:

A: How is C getting on in his new job?
B: Oh quite well, I think; he likes his colleagues, and he hasn’t been to prison yet.

Difference between what A says (that C hasn’t been to prison yet) and what A meant (could be: that C is likely to yield to temptation, that C’s colleagues are unpleasant and treacherous people, etc.).

Grice’s what is said is closely tied to truth-conditions. B’s assertion is true only if C hasn’t been to prison yet.

Everything B meant but did not say could be false without impugning the truth of B’s assertion.

Some more examples:

A: I’m out of petrol. B: There’s a garage around the corner.

Recommendation letter for a philosophy student:

Mr. X’s command of English is excellent, and his attendance at tutorials has been regular. Yours, etc.

Implicature is Grice’s term for the things a speaker means but does not say.

Conventional vs. conversational implicature.

A conversational implicature is generated by general rules of conversation, as applied to a particular conversational circumstance.

A conventional implicature is generated by meanings of words used (and so is a semantic, not a pragmatic, phenomenon).

So, e.g., Grice thought that assertions of

“A and B” and
“A but B” say the same thing (one is true iff the other is), but the latter and not the former conventionally implicates that there is some contrast between A and B.

The existence of conventional implicature is controversial.


The Cooperative Principle: “Make your conversational contribution such as is required, at the stage of the
conversation at which it occurs, by the accepted purpose of direction of the talk exchange in which you find
yourself.” Grice thinks CP governs our conversational practice. He tentatively suggests but does not endorse
an explanation (that observance of CP is “a quasi-contractual, with parallels outside the realm of discourse”).

CP gives rise to a number of maxims:

Maxims of Quantity

Make your contribution as informative as required.

Don’t make your contribution more informative than is required. (Very tentative.)

Maxims of Quality

Don’t say what you believe to be false.

Don’t say that for which you lack adequate evidence.

Maxim of Relation

Be relevant.

(Grice admits that this terse formulation “conceals a number of problems that exercise me a great deal.”)

Maxims of Manner (be perspicuous, sic)

Avoid obscurity and ambiguity.

Be brief and orderly.


Conversational implicatures are generated by a speaker’s presumed adherence to CP.

Specifically, S’s saying that

p conversationally implicates that q when attributing to her a belief that q is required if we are to treat her as
adhering to CP, and where S takes it that her listeners are capable of figuring this out. So take, e.g.,

A: I’m out of petrol.
B: There’s a garage around the corner.

B’s assertion conversationally implicates that the garage is open, because it’s only by attributing the belief that
the garage is open to A that we can regard her as obeying CP (in particular here as adhering to the Maxim of

Relation: if B didn’t think the garage was open, what she said would be irrelevant).

The following are characteristic of conversational implicatures (as opposed to conventional implicatures, or
parts of what is said):

Calculability. It’s important to Grice that listeners be capable of working out the implicature for themselves.
Otherwise he thinks it would be a conventional implicature.

Cancellability. You can explicitly “cancel” a conversational implicature; suppose B follows up her remark
about the garage by saying, “...though I don’t know whether it’s open.”

Nondetachability. The same implicature would be generated if the speaker said the same thing in a different
way (obviously this doesn’t apply to implicatures that rely on Maxim of Manner).

Grice’s theory provides a very powerful framework for articulating and defending lots of diverse philosophical and semantic claims.

I’ll give three examples:

(a) Objections to the causal theory of perception.

(b) Scalar implicature

(c) Indicative conditionals

In discussing this stuff it’s convenient to appeal to a conversational rule that’s not explicitly among Grice’s Maxims, but combines elements of Quantity and Quality:

Assert the Stronger
Say the most informative thing for which you have adequate evidence.

Grice defended a causal theory of perception: that S sees a ϕ iff (i) it appears to S as if there is a ϕ, and (2) the presence of a ϕ is causally responsible for (i). One objection to this theory was that someone who says “That
appears red” would generally believe that the thing in question is not, in fact, red, or at least that there was some doubt as to whether it was, in fact, red. So it can seem, on the face of it, that condition (i) of the causal theory is
incompatible with S’s seeing a ϕ. Grice’s response is to acknowledge that someone who says “That appears red” conversationally implicates that it’s not red, or that he has doubts about whether it’s red. (You can see how this
would work via Assert the Stronger.) But appearing red and being red are compatible.

Scalar implicature. I tell you that some of the students got A’s. If all students got A’s, is what I’ve said false? Gricean answer: no. I’ve implicated that not all students got A’s (again, via Assert the Stronger), but what I’ve
said is true even if all students got A’s. In general, when we can arrange a series of statements along a “scale” from strongest to weakest, someone who asserts something on the scale will conversationally implicate (but not
say!) that the stronger statements are false. (E.g.: [one, some, many, all], [one, two, three, ...], etc.)

Indicative conditionals. If ‘If A, then C’ means ‘A C’, then ‘If A, then C’ is true whenever A is false. But
this seems wrong. Grice’s reply: if you assert A C you implicate (via Assert the Stronger) that you aren’t in a
position to assert :A. So when you know that A is false, asserting A C is misleading. (Many problems remain.
Most painful for Grice is that if this is the explanation, asserting ‘A or C’ when you know :A should always be
misleading, but it’s not. Frank Jackson turns to conventional implicature for a solution: roughly, ‘If A, then C’
conventionally implicates that Pr(CjA) is high. Most others have abandoned the idea that ‘If A, then C’ means
‘A C’. "For much more, enroll in my seminar on conditionals in the fall!", he implicates.