Contrary to philosophical and linguistic orthodoxy, Matthew Weiner has recently argued that cancellability is not among the defining features of conversational implicatures (Weiner 2006). I do not think Weiner’s examples and argumentation manage to establish this. Grice’s most explicit commitment to the cancellability of conversational implicatures is found in his “Further Notes on Logic and Conversation”. Grice writes:

(…) a putative conversational implicature that p is explicitly cancelable if, to the form of words the utterance of which putatively implicates that p, it is admissible to add but not p, or I do not mean to imply that p, and it is contextually cancelable if one can find situations in which the utterance of the form of words would simply not carry the implicature (Grice 1989b, p. 44).

We might call this Grice’s Cancellability Test. Grice then goes on to write that even though he thinks that all conversational implicatures are cancellable, cancellability is not sufficient for concluding that we have a conversational implicature (Grice 1989b, p. 44). Weiner does not even think that cancellability is necessary for the presence of a conversational implicature.

A conversational implicature is, according to the standard picture, generated by a speaker taking advantage of the fact that his audience will generally regard him as cooperative, e.g., aiming at speaking truthfully (by not saying what he believes to be false, etc), being as informative as the conversation requires, being relevant, not speaking obscurely, etc., in order to mean or communicate something more than or different from the literal meaning of the words he uttered. The speaker does not make what he means fully explicit by what he says and relies on the audience to fill in the gap in accordance with what it is most reasonable to assume that the speaker means under the assumption that he is cooperating, and given the context of communication. The speaker intends the audience to draw these contextual inferences about what he means and he is accordingly in a position to cancel or defeat any putative conversational implicature. Here is one of Grice’s own famous examples of the phenomenon.

A: I am out of petrol.

B: There is a garage around the corner.

(Gloss: B would be infringing the maxim “Be relevant” unless he thinks, or thinks it possible, that the garage is open, and has petrol to sell; so he implicates that the garage is, or at least may be open, etc) (Grice 1989a, p. 32).

Grice’s Cancellability Test tells us that a conversational implicature is explicitly cancellable if the form of words used in an utterance that generates a putative conversational implicature can be conjoined with a denial of that putative conversational implicature. In the example above it can.

  1. A:

    I am out of petrol.

  2. B:

    There is a garage around the corner, but it’s closed.

An implicature is explicitly cancellable and thus a conversational implicature if the cancellation does not give rise to a contradiction. The form of words “There is a garage around the corner” of the potentially implicature generating utterance can be conjoined with the negation of the putative conversational implicature “but it’s closed” without contradiction.

Weiner takes the following cases to prove this principle of conversational implicatures untenable (Weiner 2006, pp. 128–129).

The Train Case: The two friends Alice and Sarah are in a crowded train. One of them, Alice, occupies two seats, while Sarah is left standing. Sarah addresses Alice with the utterance “I’m curious as to whether it would be physically possible for you to make room for someone else to sit down.” Call this (1). We might assume that (1) can generate a putative conversational implicature, which would be that Alice should make room. Call this (2). Implicature (2) seems to be generated since Sarah seems to flout the first maxim of Quality by saying something that is very likely to be false. But Sarah cannot, according to Weiner, cancel (2) by uttering the sentence “Not that you should make room; I’m just curious.” Grice’s Cancellability Test for conversational implicatures fails.

(1) Implicature generating utterance: “I’m curious as to whether it would be physically possible for you to make room for someone else to sit down”.

(2) Putative conversational implicature: Alice should make room.

(3) Cancellation of putative conversational implicature: “Not that you should make room; I’m just curious.”

The Sex Pistols Case: In 1977, the same year as Queen Elizabeth II’s silver jubilee, the Sex Pistols released their single “God Save the Queen”. In the song, the sentence “God save the Queen” is uttered, while adding, in the song, that “We mean it, man”. Call the first utterance (1). We might assume that (1) generates a putative conversational implicature, which would be something like; down with the Queen. Call this (2). Implicature (2) seems to be generated since the group’s lead singer John Lydon seems to “flout a maxim akin to the first maxim of Quality for expressive utterances, ‘Do not express a sentiment that you do not feel’’ (Weiner 2006: 128). Or perhaps we should say that Lydon flouts a maxim of Manner for expressive utterances, “Do not express a positive sentiment in a negative/aggressive manner”. But Lydon cannot, according to Weiner, cancel (2) by uttering the sentence “We mean it, man.” Call this (3). Grice’s Cancellability Test for conversational implicatures fails.

(1) Implicature generating utterance: “God save the Queen”.

(2) Putative conversational implicature: Down with the Queen

(3) Cancellation of putative conversational implicature: “We mean it, man”

Weiner considers the conversational implicature in the Sex Pistols Case to be a particularized conversational implicature that is non-cancellable because the utterance of (3) “merely intensifies the hostile sentiments expressed, by flouting the same maxim again. It makes more sense to interpret the Sex Pistols as doubly sarcastic rather than patriotic” (Weiner 2006, p. 129). This diagnosis of the case is almost correct, though the step Weiner needs for his thesis; that a sarcastically uttered cancellation of a conversational implicature is to count as an (attempted) cancellation is wrong.

Sarcasm, being a species of irony, is where one pretend/make as if to say something, while one is actually communicating an opposing view, accompanied by an appropriate attitude of contempt or indignation or something along those lines. It would be incorrect to report Lydon’s utterance of (3) as “John Lydon of the Sex Pistols said that his utterance of “God save the Queen” was sincere”. In order to report Lydon’s utterance this way you would have to, as Larson and Segal point out, mimic Lydon’s original sarcastic tone of voice (Larson and Segal 1995, p. 453). Otherwise, (3) could only be correctly reported as “John Lydon of the Sex Pistols said that his utterance of “God save the Queen” was not sincere” as Cappelen and Lepore would argue (Cappelen and Lepore 1997, p. 284). The putative conversational implicature (2) is not cancelled by (3), because what Lydon said in (3) does not count as an (attempted) cancellation of (2), even though (3) has the form of a cancellation. But to cancel a conversational implicature is a speech act and Lydon did not perform that speech act by uttering (3).

The Sex Pistols Case in not a case where Grice’s Cancellability Test fails, rather it is just a case of an utterance having the form of a cancellation of a conversational implicature, while it is not. What Lydon is doing with his utterance of (3), it seems, is reinforcing the conversational implicature (2). It is not redundant to utter the sentence “There is a garage around the corner” together with one of its conversational implicatures: “There is a garage around the corner and it’s open”. Similarly, Lydon can non-redundantly reinforce the conversational implicature generated by (1) by sarcastically adding “We mean it, man”. The example teaches us that we can reinforce a conversational implicature by generating another conversational implicature.

Weiner considers the conversational implicature in the Train Case to be a generalized conversational implicature. According to Grice you have a generalized conversational implicature when “the use of a certain form of words in an utterance would normally (in the absence of special circumstances) carry such-and-such an implicature or type of implicature” (Grice 1989a, p. 37). But the utterance of (3) does not, according to Weiner, cancel the conversational implicature (2).

Because it is so unlikely that Sarah is just curious about whether Alice is physically capable of moving, Sarah’s second remark flouts the first maxim of Quality again; she has again uttered something that she plainly believes to be false (Weiner 2006, p. 128).

Most likely, a real-life instance of the Train Case would look similar to the Sex Pistols Case. Sarah utters (1) with an ironic tone of voice and reinforces the implicature (2) by ironically adding that she is not of the opinion that Alice should make room. But as with the Sex Pistols Case, it carries no weight against Grice’s Cancellability Test.

What you would need in order for the case to, at least prima facie, challenge Grice’s Cancellability Test, is having Sarah delivering (1) in a deadpanned style and to have her add (3) in a similar style. This might truly leave Sarah’s audience bewildered: What on earth is Sarah trying to achieve with her conversational contributions? (Anyone familiar with main stream American situation comedies would know that this sort of scenario can be used for comic effect.) Would that mean that the conversational implicature was not cancelled? If we understand the original case as Weiner does, then Sarah’s cancellation is just a as-if cancellation and the conversational implicature is, of course, not cancelled. But the case is no threat to Grice’s Cancellability Test, since the case does not show that the speaker could not cancel the putative conversational implicature in this context, only that uttering the form of words which usually would count as (at least) an attempted cancellation could be used to reinforce the conversational implicature in question. But what about the revised version of the case? Grice claimed that a generalized conversational implicature is generated in the absence of special circumstances and then, of course, the presence of special circumstances might contextually cancel any putative conversational implicature. Weiner realizes this when he admits that if the speaker in the Train Case is a mad woman or a philosopher concerned with free will, then that would constitute special circumstances in which the putative conversational implicature would be cancelled (Weiner 2006, p. 129). Does cancelling a generalized conversation implicature in a deadpanned style in circumstances that seems normal, as in the Revised Train Case, in itself constitute special circumstances?

I am not sure if there is an exact answer to that question. Fortunately, the locus of the confusion that the audience experiences in the Revised Train Case is not Grice’s Cancellability Test. That a putative generalized conversational implicature can be shown to be cancellable in special circumstances validates Grice’s Cancellability Test. The question of whether you can make someone believe that you sincerely cancelled a putative conversational implicature in some particular context is different from the question of whether you can cancel it at all. The former is about what John Austin called “uptake”. According to Austin, a speech act like cancelling a putative conversational implicature is not happy (not successfully done) unless the speaker secures not only that the audience understands the meaning of the utterance but also its force; that the cancellation was sincere (Austin 1962, pp. 115–116). In the case of cancelling a putative generalized conversational implicature, and also often with a putative particularized conversational implicature, we provide a reason(s) for why we do not intend to communicate a particular conversational implicature, if context does not provide such a reason(s). This reason(s) gives the audience the means for providing an alternative interpretation of what the speaker meant, when the conversational implicature interpretation has been proven wrong (by the cancellation).

Most generally, I think that the literal meaning of the form of words that were uttered in a speech act functions as the default interpretation for the audience when the speaker cancels all potential conversational implicatures, but if a speaker has uttered something, which in context seems blatantly false or highly unlikely, and furthermore has cancelled all potential conversational implicatures but failed to provide the audience with any clues as to what would be a reasonable interpretation of what he meant or how we can reasonably interpret him as only meaning what the words he uttered literally mean, then the speaker is at fault. His utterances are gravely infelicitous because they have deprived us of any venue of interpretation by which we can make sense of, not what the words he uttered literary meant, but what he meant by uttering them. The locus of our confusion is how to understand speaker’s meaning. The cancellations of the putative conversational implicatures in Weiner’s cases can go through; indeed it is only because they can go through that audience will be confused or puzzled as is the case in the Revised Train Case. That the locus of confusion/puzzlement in the cases Weiner introduces, if they can be made to work as intended, lies with the notion of speaker’s meaning should come as no surprise, since conversational implicatures are a form of speaker’s meaning. Grice’s theory of meaning and theory of conversation are intimately connected. They are in many respects the mirror images of each other and any further illumination of the problems brought to our attention by Weiner’s paper ought to pay careful attention to the connections between the two theories.Footnote 1 Suffice to say for now is that Weiner has not provided us with reasons for doubting Grice’s Cancellability Test for conversational implicatures.