The concept of conversational implicature is due to the work of Paul Grice, and in particular to his paper "Logic and Conversation," which was delivered in 1967 and instantly became highly influential, although it was not published until 1975. A key goal of this paper was to defend the traditional logical understanding of connectives like and against what he saw as the excesses of ordinary language philosophy. He did this by drawing a sharp distinction between what is strictly speaking said and what is conversationally implicated. Consider sentence (1), below.
(1) Amanda and Beau fell in love and got married.
An utterance of (1) will typically suggest that the falling in love preceded the marriage. However, if and has its bare logical meaning, (1) may be true even if the marriage was initially loveless. According to Grice, (1) might indeed be true under these circumstances—because, strictly speaking, and contributes no more than its logical meaning to what is said. Grice claimed that the extra suggestion of temporal order was a conversational implicature. Conversational implicatures are an important part of communication, but (according to Grice) they have no effect on truth value. This is because they are not a part of what is strictly speaking said.
Grice argued that conversational implicatures arise from our adherence to (and presumption that others will adhere to) what he called the Cooperative Principle: "[m]ake your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged" (1989b, p. 26). (He took this principle to govern conversation, but he also took it that the principle would have correlates in other cooperative endeavors.) In its broadest outline Grice's idea was that we presume that others are being cooperative—and we will generally make whatever supplementary assumptions are required to maintain this presumption. This presumption is what allows for the communication of conversational implicatures.
Grice takes it that we generally follow the Cooperative Principle by following four more specific maxims of cooperation (which, like the Cooperative Principle itself, he takes to have correlates in other endeavors):
Quantity: Make your contribution as informative as (neither more nor less informative than) is required (1989b, p. 26).
Quality: "Try to make your contribution one that is true": "do not say what you believe to be false" and "do not say that for which you lack adequate evidence" (1989b, p. 27).
Relation: "Be relevant" (1989b, p. 27).
Manner: "Be perspicuous": "1. Avoid obscurity of expression; 2. Avoid ambiguity; 3. Be brief (avoid unnecessary prolixity); 4. Be orderly" (1989b, p. 27).
Any of these maxims may play a role in generating conversational implicatures.
Grice characterizes conversational implicature as follows, and most scholars have followed him in this characterization.
A man who, by (in, when) saying (or making as if to say) that p has implicated that q may be said to have conversationally implicated that q, provided that:
(1) he is to be presumed to be following the conversational maxims, or at least the Cooperative Principle;
(2) the supposition that he is aware that, or thinks that, q is required to make his saying or making as if to say p (or doing so in those terms) consistent with this presumption; and
(3) the speaker thinks (and would expect the hearer to think that the speaker thinks) that it is within the competence of the hearer to work out, or grasp intuitively, that the supposition mentioned in (2) is required.
(grice 1989, pp. 30–31)
To see how all of this machinery works in the generation of a conversational implicature, one must return to an utterance of (1).
(1) Amanda and Beau fell in love and got married.
Typically, the audience will assume that the speaker is being cooperative (so condition 1 is met). A cooperative speaker will follow the maxim of Manner, and the maxim of Manner dictates orderly presentation. If one is being orderly, one will generally present events in the order in which they occurred, so the audience must assume that the speaker thinks that Amanda and Beau's love predated their marriage. (Condition 2 is met). The speaker surely realizes that the audience is capable of working this out, so condition 3 is met. A typical utterance of (1), then, will implicate something like (1*).
(1*) Amanda and Beau fell in love and then got married.
There are a variety of ways that conversational implicatures may be generated. The above mechanism does not rely on the speaker's utterance being one whose explicit content would be uncooperative, but this latter sort of implicature (created by violating or flouting a maxim) is also an important one. It is crucial, for example, to irony, hyperbole, and understatement.
Kinds of Conversational Implicature
Grice distinguished two main kinds of conversational implicature, generalized and particularized. Generalized conversational implicatures are ones that are usually carried by a certain word or phrase, while particularized conversational implicatures depend far more heavily on context. (Grice also allowed for nonconversational implicatures, such as conventional implicatures, with no role whatsoever for context. These are not of concern here.) The example discussed so far, involving (1), is that of a generalized conversational implicature: utterances of sentences involving and will usually carry the suggestion of temporal order that it is present in this example. Particularized implicatures depend far more heavily on context. Suppose you are hiring for a philosophy job, you ask me what I think of my student Charla, and I reply with nothing but (2).
(2) Charla reads a lot.
(2) is clearly not adequately informative, given your question—philosophy jobs require a great deal more than reading. By violating the maxim of Quantity in this way, I conversationally implicate that Charla is not a good philosopher. But in a different context, there would be no such implicature (consider, for example, a context in which I was asked for names of people who might like to join a book group). Because the implicature depends so heavily on context, it is particularized rather than generalized.
A great deal of work has been done, especially in linguistics, on generalized conversational implicatures, and various mechanisms for their generation. Some of the most important work on this topic is by Laurence Horn (1972) and Stephen Levinson (2000). Generalized conversational implicatures have also played a particularly important role in philosophy. For Grice, these implicatures were especially important for their role in explaining certain problematic intuitions—like the intuition that (1) is false if the marriage preceded the love. He argued that generalized conversational implicatures are especially difficult to distinguish from what is strictly speaking said and that they may therefore give rise to mistaken intuitions. Philosophers since Grice have followed up on this thought by using generalized conversational implicatures in explanations of recalcitrant intuitions in a wide variety of areas.
It is worth noting that many cases that Grice took to be ones of generalized conversational implicature are very much disputed. For example, Robyn Carston (1991) has argued for a notion of saying (or, in her preferred terminology, explicating ) on which the meaning of and is just what Grice would have taken it to be, yet nonetheless an utterance of a sentence like (1) says (rather than implicates) that Amanda and Beau's love preceded their marriage.
(1) Amanda and Beau fell in love and got married.
Carston's work on such examples is a part of a broader debate on the notion of what is said, which is not addressed here. But for some other approaches that also result in examples like the above counting as said, see Jeffrey King and Jason Stanley (2005), François Recanati (1989), and Dan Sperber and Dierdre Wilson (1986). For objections to reconstruing such generalized conversational implicatures as a part of what is said, see Kent Bach (2001), Laurence Horn (1992), Stephen Levinson (2000), Michael O'Rourke (2003), and Jennifer Saul (2002b).
Testing for Implicature
Grice does not offer necessary and sufficient conditions that would allow one to test conclusively whether a given claim is a conversational implicature. However, he does offer certain necessary conditions for conversational implicature that can provide partial tests, and these have been widely accepted. Two especially important ones are cancelability and calculability. For more on testing for conversational implicatures, see Jerrold Sadock (1978).
Because all conversational implicatures depend at least to some extent on context, it is always possible to cancel a conversational implicature by indicating either explicitly or implicitly that the implicatures should not be taken as present. For example, one might utter (1C):
(1C) Amanda and Beau fell in love and got married, but not in that order. Because (1C) contains an explicit cancellation of the conversational implicature standardly carried by (1), that implicature will not be carried by an utterance of (1C).
This contrasts with the case of saying. An attempt to "cancel" something that is said results only in a contradiction. To see this, consider an utterance of (1C*):
(1C*) Amanda and Beau fell in love and got married, but they didn't get married.
Applying this test shows us that the claim that Amanda and Beau got married is definitely not a conversational implicature, while the claim that their love preceded their marriage may well be.
While failure of the cancelability test does indeed indicate that one is not dealing with a conversational implicature, passing the cancelability test cannot be taken to decisively established that one is dealing with a conversational implicature. There are at least two reasons for this. First, a case of disambiguation may resemble one of cancellation, as in (3):
(3) He is in the grip of a vice, but not the mechanical kind.
Second, speaking loosely may result in an appearance of cancellation. Grice's own example (1989a, p. 44) concerns the fact that one may acceptably say, "Macbeth saw Banquo, even though Banquo was not there to be seen," even if it is known by all that Banquo merely hallucinated. Because one might be using the verb "see" in a loose way, this apparent cancellation does not indicate that utterances involving "see" merely implicate that what is seen exists.
According to Grice, a putative conversational implicature is not a conversational implicature unless it is possible for audience to work out that the presence of the implicature is required in order to understand the speaker as cooperative. This calculation is meant to draw on knowledge of the linguistic meaning of the sentence uttered, the maxims of conversation, relevant background information, and the specific context. If no explanation can be given of how an audience would perform a calculation like this, a hypothesis that a particular conversational implicature is present must be rejected.
This necessary condition is also widely accepted. But its exact interpretation is a matter of some controversy. In particular, there is disagreement over what it requires psychologically on the part of the hearer: must the hearer actually have distinct conscious representations of what is said and what is conversationally implicated (as argued by François Recanati in his work (1989)? Or are the requirements much more minimal (as argued by Kent Bach (2001), Manuel Garcia-Carpintero (2001), and Kenneth Taylor (2001)? The calculability requirement has proved very important: it has been used, for example, to argue for a more expansive conception of what is said (as in Recanati's work, as well as Robyn Carston's ); to argue for and to object to particular invocations of conversational implicature; and (as in Wayne Davis's ) to raise quite general concerns about the viability of Grice's theory of implicature.
It is very common to maintain that speaker meaning must divide exhaustively into what is said and what is implicated. Thus, any claim that the speaker means but does not say must be an implicature. (It need not, however, be a conversational implicature, since it could be a conventional implicature.) It is not entirely clear, however, that this is the right way to understand relationship between speaker meaning and implicature. For objections to this view of the relationship, see Kent Bach (1994) and Jennifer Saul (2002b).
Bach, Kent. "Conversational Impliciture." Mind and Language 9 (2) (1994): 124–161.
Bach, Kent. "You Don't Say?" Synthese 128 (2001): 15–44.
Davis, Wayne. Implicature: Intention, Convention, and Principle in the Failure of Gricean Theory. Cambridge, U.K.: Cambridge University Press, 1998.
Garcia-Carpintero, Manuel. "Gricean Rational Reconstructions and the Semantics/Pragmatics Distinction." Synthese 128 (2001): 93–131.
Grice, Paul. "Further Notes on Logic and Conversation." In Studies in the Way of Words, 41–57. Cambridge, MA: Harvard University Press, 1989a.
Grice, Paul. "Logic and Conversation." In Studies in the Way of Words, 3–21. Cambridge, MA: Harvard University Press, 1989b.
Horn, Laurence. "The Said and the Unsaid." In SALT II: Proceedings of the Second Conference on Semantics and Linguistic Theory, 163–192. Columbus, OH: Ohio State University Department of Linguistics, 1992.
Levinson, Stephen C. Pragmatics. Cambridge, U.K.: Cambridge University Press, 1983.
Levinson, Stephen C. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press, 2000.
O'Rourke, Michael. "The Scope Argument." Journal of Philosophy 100 (2003): 264–288.
Recanati, François. "The Pragmatics of What Is Said." Mind and Language 4 (1989): 295–329.
Sadock, Jerrold. "On Testing for Conversational Implicature." In Syntax and Semantics. Vol. 9: Pragmatics, edited by Peter Cole and Jerry Morgan. New York: Academic Press, 1978.
Saul, Jennifer. "Speaker Meaning, What Is Said, and What Is Implicated." Noûs 36 (2) (2002a): 228–248.
Saul, Jennifer. "What Is Said and Psychological Reality: Grice's Project and Relevance Theorists' Criticisms." Linguistics and Philosophy 25 (2002b): 347–372.
Sperber, Dan, and Dierdre Wilson. Relevance: Communication and Cognition. Cambridge, MA: Harvard University Press, 1986.
Taylor, Kenneth. "Sex, Breakfast, and Descriptus Interruptus." Synthese 128 (2001): 45–61.
Jennifer Saul (2005)