Saying what you mean

by Carl Pierer

Beautiful Grice

After dinner and upon noticing a stain, Abelard ejects: “Oh no, I look like a pig.”

Bertha: “Well, and you've spilled sauce down your tie!”

Often an utterance means something over and above of what it literally says. Is it in such cases always possible to return to the “literal” meaning? Is it possible to sincerely answer the question “Do you think I'm fat?” with “You have nice feet” and only mean that the questioner has nice feet?

To capture and analyse what is going on in these cases, H.P. Grice introduces the term “implicature”. Abelard's statement by itself is usually understood metaphorically; it is rather unlikely that he literally looks like a pig. More plausibly, he means that he looks messy – possibly because he spilled some sauce. This is an example of conventional implicature – an implicature associated with certain set phrases, where the implicature is not depended on the context in which the utterance is made. The second sentence, Bertha's utterance, also seems to carry a meaning above what she literally said. It seems that in this context Bertha suggests precisely the literal meaning of Abelard's sentence. This kind of implicature, which depends on the context, Grice calls conversational implicature.

On the face of it, Bertha's utterance is at best redundant or worse, it does not make sense. If Abelard is really implicating: “Oh no, I've spilled sauce down my tie”, Bertha's repetition of this very fact does not add anything to the conversation. Although this happens all too often, it seems reasonable to suppose that usually people try to contribute to the conversation. In a meaningful exchange (whatever that means), people try to be constructive. This idea is captured by Grice in what is called the Cooperative Principle. Wayne Davis concisely puts the cooperative principle thusly:

Cooperative Principle. Contribute what is required by the accepted purpose of the conversation.”

For instance, if asked where to have lunch, a person will usually name a place. Another perfectly common reply is: “I'd like to have some lasagne”. Yet, strictly speaking this does not answer the question. In fact, at first sight, it violates the cooperative principle: the purpose of this exchange is to figure out where to go for lunch, so saying what you want to have for lunch is not directly relevant. And still, it is a meaningful contribution, provided that we understand “I'd like to have some lasagne” as implicating “Somewhere where they serve lasagne”.

Grice suggests that the presence of the implicature is figured out as follows, where S is the utteress of the lasagne-sentence:

(1) S is not opting out of the cooperative principle, her contribution should be understood as meaningful

(2) Only if S implicates “Somewhere where they serve lasagne” is her utterance “I'd like to have some lasagne” consistent with (1)

(3) Therefore, S implicates “Somewhere where they serve lasagne”.

The cooperative principle plays a crucial role in the detection of a conversational implicature. On Grice's view, it has to be supposed that the speaker S is abiding by the cooperative principle, because if (1) does not obtain in the above reasoning, (3) does not follow. But, it is very well possible to opt out of the cooperative principle, so if S makes clear that she is opting out, she should be able to say “I'd like to have some lasagne” without implicating “Somewhere where they serve lasagne”.

Grice calls this further feature of conversational implicatures cancellability. In general, a conversational implicature can be cancelled explicitly or contextually, depending on how it becomes apparent that S is not following the cooperative principle. However, cancellability – Grice thinks – is a necessary but not a sufficient feature of conversational implicatures. Indeed, if they were not cancellable, a different method (maybe without invoking something like the cooperative principle) to determine their presence would be needed.

Of course, it has been suggest that there are some conversational implicatures that are not cancellable. With some ironic statements, it seems that an apparent cancellation does not so much cancel the implicature as to strengthen it. This appears to be the case in the following example, which is a reformulation of the one given by Matthew Weiner:

Bertha and Abelard, audibly enjoying his food, are having lunch. Bertha says: “I'm curious to know if it is possible for you to make even louder noises?”

Abelard would suppose Bertha to be abiding by the Cooperative Principle, that is: she is not simply saying what she believes to be false. This lets Abelard figure out the implicature: “Be less noisy!” But now, the example continues:

Bertha adds: “I don't mean to suggest that you should be less noisy, I'm just curious if you can be even noisier?”

According to Grice, this addition should cancel the initial implicature of “Be less noisy”, as Bertha explicitly denies the presumed implicature. However, in any normal context, Weiner writes, Abelard is very likely to understand Bertha's addition as implicating “Be less noisy” even more strongly. The reason is: if Abelard supposed (justifiably) that Bertha was not interested in his noise-making abilities in the first place, there is no reason for him to think that she is now. Given the nature of the Gricean calculus, this circle can be reiterated: If Abelard thinks that the implicature “Be less noisy” is required to make Bertha's utterance consistent with the assumption that she is abiding by the Cooperative Principle, then there is nothing Bertha can say to convince him otherwise – hence there is no way she could cancel that implicature. In this case, it just seems highly implausible that Bertha is sincere, rather than ironic. If Weiner is right, then – at least sometimes – it is impossible to renounce an implicature and there is no possibility of getting back to the “literal” meaning of the utterance.

Blome-Tillmann acknowledges that in this picture the implicature is not cancellable in the Grice's sense. Yet, he suggests, this is not a problem if only the criteria for determining cancelability are modified. By changing the context of the exchange, it can be shown both that there is a situation where the implicature is cancelled contextually and one where the implicature can be cancelled explicitly. The story is retold:

(i) Abelard has been involved in a fight and is severely injured, so that he can barely move his jaw. Bertha is a physician and interested in Abelard's injuries. During their lunch she examines Abelard's wounds, then asks Abelard: “I'm curious to know if it is possible for you to make even louder noises?”

(ii) The same situation, but Abelard does not know that Bertha is a physician. Answering her question, he says: “Should I be less noisy?” Bertha replies: “I don't mean to suggest that. As a physician, I'd like to understand your injuries. I'm just curious if you can be even noisier?”

In (i) the context is such that “Be less noisy” is not required to make Bertha's utterance consistent with the assumption that she is being cooperative. In (ii), Bertha can successfully cancel the implication, it is not implausible for Abelard to believe her. But if this is the case, then the implicature presented by Weiner is both contextually cancellable (in (i)) and explicitly cancellable (in (ii)), albeit in a weakened sense. This weakened sense is that for a given utterance carrying a certain implicature, there exists a (possibly fictional) context in which a parallel utterance conveys the same implicature. Further, denying the parallel utterance conveys a cancellation of the implicature. Lastly, denying the parallel utterance while at the same time holding the initial utterance makes sense in this fictional context. So, provided that we can tell the same story in a different context in which the cancellation seems plausible, an implicature is said to be explicitly cancellable in the weak sense.

Reflecting on the initial exchange about looking like a pig, the story could be retold in this way:

Abelard just had Spaghetti Bolognese for dinner. Because he is not very skilled at eating long pasta noodles, he spilled sauce on his clothes. Noticing a stain, he says: “Oh no, I look like a pig”. Bertha replies: “And you've spilled sauce down your tie, as well!” Abelard smiles wryly. Bertha adds: “No, I'm not making fun of you, I thought you just noticed the stain on your trousers.”

It appears that here the context is such that the cancellation is plausible. Blome-Tillmann suggests that any “uncontroversial” conversational implicature is explicitly cancellable in the weak sense. This seems at least likely, as with the infinite resources of providing fictional background information, it should be possible to paint a picture in which the cancellation is plausible.

The comforting answer is then that conversational implicatures are explicitly cancellable, in the weak sense. The less comforting answer is that in a particular situation cancellation may be impossible. So the implication “Yes, I do” of replying “You have nice feet” to the question “Do you think I'm fat?” could be cancelled in a (fictional) context. Unfortunately, in the particular and relevant situation this might not be an option.



Blome-Tillmann, M. (2008). Conversational implicature and the cancellability test. Analysis, 68(2), 156-160.

Davis, W. A. (1998). Implicature: Intention, convention, and principle in the failure of Gricean theory. Cambridge: Cambridge University Press.

Grice, H. P. (1991). Studies in the way of words. Cambridge, MA: Harvard University Press.

Weiner, M. (2006). Are all conversational implicatures cancellable? Analysis, 66(290), 127-130.