[for a festschrift for Larry Horn, edited by Betty Birner and Gregory Ward (John Benjamins 2005)]

The Top 10 Misconceptions about Implicature

Ive known about conversational implicature a lot longer than Ive known Larry. In 1967 I read Grices Logical and Conversation in mimeograph, shortly after his William James lectures, and I read its precursor (Implication), section III of The Causal Theory of Perception, well before that. And Ive thought, read, and written about implicature off and on ever since. Nevertheless, I know a lot less about it than Larry does, and thats not even taking into account everything he has uncovered about what was said on the subject long before Grice, even centuries before. So, now that Ive betrayed my ignorance, Ill display my insolence. Im going to identify the most pervasive and pernicious misconceptions about implicature that Ive noticed over the years. This wont be a natural history of them. I have neither the time, the space, nor the patience for such a scholarly endeavor. It would also be unseemly, as if what I am doing isnt. At any rate, Ill keep things short and to the point (ten points, actually), though this will make me seem a bit glib if not dogmatic. I wont target the sources of these misconceptions, much less delineate their paths of propagation or document the damage theyve wrought. Ill simply identify them and, with the help of a handy distinction or an overlooked possibility, suggest how each might arise. I wont follow David Letterman and present them in reverse order of magnitude. Nor will I present them in order of
1

importance or frequency of manifestation. Rather, Ill put them in an easy to follow sequence. Only the last two or three, I hope, will seem contentious (unless otherwise indicated, by implicature I will always mean conversational implicature). Heres the list: 1. Sentences have implicatures. 2. Implicatures are inferences. 3. Implicatures cant be entailments. 4. Gricean maxims apply only to implicatures. 5. For what is implicated to be figured out, what is said must be determined first. 6. All pragmatic implications are implicatures. 7. Implicatures are not part of the truth-conditional contents of utterances. 8. If something is meant but unsaid, it must be implicated. 9. Scalar implicatures are implicatures. 10. Conventional implicatures are implicatures. These formulations of the top ten misconceptions about implicature will appear as section headings in what follows. Please dont take that for an endorsement of any of them. 1. Sentences have implicatures. It is in uttering sentences that speakers implicate things. Yet for some reason, implicatures are often attributed to sentences themselves. Perhaps thats because implicatures are often illustrated with the help of numbered sentences, which are then confused with utterances, which are then treated as if they are agents rather than as the actions that they are. Anyway, Grice was careful to use the verb implicate, not imply, for

what speakers do, and he coined the term implicature to use instead of implication for what speakers implicate. The difference is fundamental. If a sentence is true, what it implies must be true, whereas a speaker can utter a true sentence and implicate something false. For example, you could say that theres a gas station around the corner and falsely implicate that its open and selling gas (maybe its closed for the night or maybe theres a gasoline shortage). If theres a gas station around the corner, it doesnt follow that the gas station is open and selling gas. But it does follow that the gas station is not directly across the street. This fundamental difference reflects the fact that what a sentence implies depends on its semantic content, while what a speaker implicates is a matter of his communicative intention in uttering the sentence. Thats why implicature is pragmatic in character, hence why in different situations one can utter a given unambiguous sentence and implicate different things. For example, you could say Johns command of English is excellent to implicate, depending on the situation, that John is a mediocre student, that he would make a fine translator, that he understood something he heard, or that he had no excuse for the sloppy paper he wrote. Of course, what a speaker could, in a given situation, plausibly be taken to implicate will be constrained by the semantic content of the sentence -- certainly it matters what the sentence means -- but this doesnt make implicature a property of the sentence itself. The tendency to attribute implicatures to sentences is greatest in the case of generalized conversational implicatures, which do not depend on special features of the conversational situation and thus are more directly associated with sentences themselves

(but like particularized implicatures, GCIs are cancelable). For example, in uttering Bill is meeting a woman this evening you would normally (in the absence of special circumstances) be implicating that the woman in question is not Bills wife. So it makes sense, without considering actual speakers intentions, to talk about what is likely to be implicated when a certain sentence is uttered. This might suggest that the GCI is a property of the sentence itself, even though GCIs are not semantic in character but are pragmatic regularities. Even so, it is the speaker, not the sentence, that does the implicating. Unfortunately, taking GCIs to be properties of sentences leads to the spurious idea that they comprise some sort of intermediate level of meaning between linguistic meaning and speaker meaning. 2. Implicatures are inferences. For some strange reason, implicatures are often described as inferences. This misdenomer is but a slight variation on the vulgar conflation of implying with inferring. As observed in The American Heritage Book of English Usage,
People sometimes confuse infer with imply, but the distinction is a useful one. When we say that a speaker or sentence implies something, we mean that information is conveyed or suggested without being stated outright. ... Inference, on the other hand, is the activity performed by a reader or interpreter in drawing conclusions that are not explicit in what is said.

Similarly, people sometimes confuse infer with implicate and inference with implicature. Why is the difference important? One obvious reason is that the audience can take the speaker to be implicating something when in fact he isnt. A putative implicature need not be an actual one. Equally obviously, a speaker can implicate something even if the

audience doesnt make the intended inference. Of course, this will not be a case of successfully conveying the implicature, but that doesnt mean the speaker didnt implicate anything, just as a speaker can hint at something without the audience getting the hint. Notice, by the way, that the inference here is not to the truth of the implicature but to its content. Its one thing to recognize what is being implicated and quite another to accept it. 3. Implicatures cant be entailments. It is commonly assumed that what a speaker implicates in uttering a sentence cant be entailed by the sentence itself. To be sure, most implicatures (by speakers) are not entailments (by sentences uttered by speakers), but there are exceptions. For example, suppose someone says to you, Nobody has ever long-jumped over 28 feet. You reply, Whadya mean? Bob Beamon long-jumped over 29 feet way back in 1968. Here you are clearly implicating that somebody has long-jumped over 28 feet. But this is entailed by the fact that Beamon long-jumped over 29 feet. The important point here is why, generally speaking, the truth of an implicature is independent of the truth of what is said. The reason is that its not what the speaker says but that he says it (or even that he puts it a certain way) which carries the implicature. 4. Gricean maxims apply only to implicatures. Grice introduced his maxims of conversation to explain how implicatures get conveyed, but this does not mean, as is often supposed, that theyre idle otherwise. To dispel this misconception we need first to get clear on the character of Grices maxims. They are not sociological generalizations about speech, nor they are moral

prescriptions or proscriptions on what to say or communicate. Although Grice presented them in the form of guidelines for how to communicate successfully, I think they are better construed as presumptions about utterances, presumptions that we as listeners rely on and as speakers exploit. As listeners, we presume that the speaker is being cooperative (at least insofar as he is trying to make his communicative intention evident) and is speaking truthfully, informatively, relevantly, and otherwise appropriately. If an utterance superficially appears not to conform to any of these presumptions, the listener looks for a way of taking it so that it does conform. He does so partly on the supposition that he is intended to. As speakers, in trying to choose words to make our communicative intentions evident, we exploit the fact that our listeners presume these things. These presumptions should not be viewed as delivering a decision procedure for the hearer to figure out what the speaker means (they can clash, after all). Rather, they provide different dimensions of considerations that the speaker, given that hes trying to communicate something, may reasonably be taken to intend the hearer to take into account in figuring out what the speaker means. And speakers implicitly realize this when they choose what to say and how to say it. Its a common misconception that the Gricean maxims, or conversational presumptions, kick in only when the speaker is implicating something (or is speaking figuratively). In fact, they apply equally to completely literal utterances, where the speaker means just what he says. After all, even if what a speaker means consists precisely in the semantic content of the sentence he utters, this still has to be inferred. It might seem that these presumptions play a role only if the speaker is not being perfectly literal and fully explicit. After all, that is when the hearer has to figure out what the

speaker means instead of or in addition to what he says. If an utterance appears not to conform to the presumptions, the hearer looks for a way of taking the utterance so that it does conform. But even if it is consistent with the presumptions that the speaker is being literal and means precisely what his words mean, the presumptions still play a role. Obviously, they arent needed to guide the hearer to a plausible candidate for what the speaker means, but taking the utterance just at face value still requires supposing that the speaker is conforming to them. 5. For what is implicated to be figured out, what is said must be determined first. In saying things, people can implicate other things. It might seem, then, that grasping what someone implicates requires first determining what they are saying. However, this is not true and not something that Grice was committed to. Its a mistake to suppose that what is said must be determined first or to suppose that Grice supposed this. This misconception forms the basis for various anti-Gricean arguments based on the premise that, if Gricean pragmatics were correct, a listener would have to determine what is said before inferring anything further as to what a speaker means. Arguments relying on this assumption have been used to defend such claims as that the what is said is a theoretically useless notion or else that what is said is not a purely semantic notion but involves pragmatic intrusion, that the semantic-pragmatic distinction is blurry if not downright bogus, and even that truth-conditional semantics is hopeless and needs to be replaced by something called truth-conditional pragmatics. However, Grice did not even purport to give an account of the psychological processing involved in recognizing an implicature (or in forming the intention to implicate something).

This misconception overlooks the difference between a real-time cognitive process and the information to which that process is sensitive. Grice did not intend his account of how implicatures are recognized as a psychological theory or even as a cognitive model. He intended it as a rational reconstruction. When he illustrated the ingredients involved in recognizing an implicature, he was enumerating the sorts of information that a hearer needs to take into account, at least intuitively, and exhibiting how this information is logically organized. He was not foolishly engaged in psychological speculation about the nature of or even the temporal sequence of the cognitive processes that implements that logic. There are cases in which it is pretty clear to the hearer well before the speaker finishes saying something that he does not mean what he will have said. For example, when the utterance is obviously going to be metaphorical, the hearer does not have to determine first that what the sentence means is not a likely candidate for what the speaker means before figuring out what the speaker does mean. Often that can be done on the fly. For example, if in response to an utterance of No man is an island, someone says Some men are peninsulas, some men are volcanoes, and some men are tornadoes, in order to figure out what the speaker means you do not have to figure out first that he does not mean that some men are peninsulas, some men are volcanoes, and some men are tornadoes. Similarly, if youre discussing a touchy subject with someone and they say, Since it might rain tonight, Id better bring in the laundry, clean out my gutters, and find my umbrella, you could probably figure out before they were finished saying all this that they were implicating that they didnt want to discuss that touchy subject any further. 6. All pragmatic implications are implicatures.
8

I doubt that very many people would own up to this misconception, and my impression of its prevalence may depend more on what people say than on what they actually believe. After all, almost everyone recognizes the difference between implicatures and pragmatic presuppositions. Even so, some people seem to think that anything that may be inferred from the fact that a speaker uttered a certain sentence is an implicature. Yes, such a thing is pragmatic because it is inferred not from the sentences content but from the fact that the speaker uttered the sentence, but that doesnt automatically make it an implicature, contrary to what is sometimes said. For example, there is the claim that if you assert something, you implicate that you believe it, you implicate that your audience should believe it, and you implicate that it is worthy of belief. This claim overlooks, among other things, the distinction between what a speaker means (has a communicative intention to convey), which is the content of an utterance (over and above its semantic content), and what the conditions are for making the utterance felicitously. Also, a speakers saying a certain thing might reveal information about him, such as that he craves attention, that he hates his father and loves his mother, or that he has a certain ulterior motive, but such bits of inferable information arent implicated unless theyre part of what he means. In general, what is meant and in particular what is implicated must be distinguished from anything else that may be inferred from the fact that the speaker made the utterance. 7. Implicatures are not part of the truth-conditional contents of utterances. There is a tendency among those who speak of utterances as having truth-conditional contents to exclude implicatures from these contents. In fact, they even argue that something is an implicature precisely because it is not part of the truth-conditional
9

content of an utterance. This is particularly common in connection with claims about conventional implicatures. Yet there is something rather strange about this way of talking. After all, implicatures are capable of being true or false. To be sure, if what a speaker says is true and what he implicates is false, we might still tend to judge his utterance as true. For example, if he accurately says that he saw Bill with a woman and falsely implicates that Bill was not with his wife, we might judge him to be speaking truly but misleadingly. If he were a witness in a divorce proceeding, he might be innocent of perjury. Even so, what he implicated is part of the total truth-conditional content of his utterance. So why do people talk as if an implicature is not part of the truth-conditional content of an utterance? I think theres an easy explanation. What they actually mean is that an implicature carried by an utterance of a sentence is not part of the semantic content of the sentence, or is not part of what is said by the speaker in uttering the sentence. Thats fine, but it does not suggest that the implicature isnt part of the truth-conditional content of the utterance, if by that we mean not the sentence but the act of uttering it. A possible source of confusion here is an often overlooked ambiguity involving the phrase utterance interpretation. Sometimes it is used to mean the psychological process whereby listeners figure out what speakers are trying to communicate, and sometimes it is used in a strict semantic sense to mean something more abstract, a mapping from syntactic structure to semantic contents. When these are confused, utterances are treated as if they are linguistic objects and yet whose interpretation is a matter of discerning speakers intentions.

10

Probably the wisest course is not attribute contents to utterances at all. The only sense in which an utterance has content over and above that of the uttered sentence is as an intentional act performed by a speaker. But in that sense, the content of an utterance is nothing more than the content of the speakers communicative intention in making the utterance. Utterances (considered as acts as opposed to sentences) dont really have contents in their own right, independently of that intention. There is no independent, objective content beyond that. There is what the hearer takes to be the utterance content (i.e., the content of the speakers communicative intention), there is what the hearer could, in the conversational situation, reasonably take that content to be, and there is what the speaker could reasonably expect it to be taken to be, but thats that. 8. If something is meant but unsaid, it must be implicated. One very common assumption is that what a speaker means can be divided exhaustively into what is said and what is implicated. Yet what a speaker means can go beyond what he says without being implicated. Speaking figuratively or obliquely are two familiar ways in which what you mean can depart from the semantic content of the sentence you utter. Either way, to succeed in communicating you exploit the listeners expectation that you are speaking truthfully, informatively, relevantly, and otherwise appropriately. Unfortunately, Grice classified figurative statements, such as You are the cream in my coffee, as implicatures. Given his assumption that to say something is to mean it, this forced him to suppose that to speak nonliterally is merely to make as if to say something. But it seems obvious that in speaking figuratively one really is saying something (but meaning something else

11

instead). At any rate, there is a different kind of case, which Grice seems not to have taken into account. We often use sentences in a way that is not strictly determined by their meanings but is not figurative or oblique (implicature-producing) either. There are other ways of not meaning just what you say. For example, if your spouse says I will be home later she is likely to mean that she will be home later that night, not merely at some time in the future. In such cases what one means is an expansion (as I call it) of what one says, in that adding more words (tonight, in the example) would have made what was meant more fully explicit. In other cases, such as Fanny has finished and Lanny is late, the sentence does not express a complete proposition. There must be something which Fanny is being claimed to have finished and something which Lanny is being claimed to be late for (or late to). In these cases what one means is a completion of what one says. In both sorts of case, no particular word or phrase is being used nonliterally and there is no indirection. Both exemplify what I call conversational impliciture since part of what is meant is communicated not explicitly but implicitly, by way of completion and expansion. Completion and expansion are both processes whereby the hearer supplies missing portions of what is otherwise being expressed explicitly. With completion a propositional schema is filled in, and with expansion a complete but skeletal proposition is fleshed out. Either way, what the speaker means is built up from what the speaker has made explicit. This is different from both figurative utterances and implicatures (and indirect speech acts generally), since the speaker builds directly on what he has made explicit. What he means is an embellished version of what he says.

12

Confusion in this area has arisen because of a common but bewildering use of the phrase explicit content. Some people classify as part of the explicit content of an utterance elements that are not explicit, that is, elements that do not correspond to any syntactic constituents of the uttered sentence. So, on this usage, when Yogi Berra said No one goes there any more -- its too crowded, the explicit content of his utterance was that no one important goes to Ruggerios (in St. Louis) any more -- its too crowded. To confuse the issue even further, some people use the neologism explicature for this embellished content, as if there is no difference between somethings being explicit and its being made explicit (explicated). 9. Scalar implicatures are implicatures. People wouldnt call so-called scalar implicatures implicatures unless they thought that thats what they are. But generally theyre not -- theyre mostly implic-i-tures. Why do I say that? Well, consider a simple example. A typical claim is that in uttering Some of the boys went to the party, the speaker implicates that not all of the boys went to the party. But this assumes that the speaker means not one but two things, that some of the boys went to the party and that not all of them did. Really, though, the speaker means only one thing, that some but not all of the boys went to the party. He could have spelled this out by including but not all after some. Similarly, if you say, I have two TV sets, you do not mean both that you have two TV sets and that you dont have more than two, you mean that you have exactly two, which you could have made explicit by putting exactly before two. Lest this seem like a merely terminological point, notice that there are special cases in which the speaker does mean two things. For example, suppose youre asked whether all
13

the boys went to the party. You reply, Some of the boys went to the party. In this case, you are saying that some went and implicating that not all did. Similarly, suppose youre asked if you have three TV sets and you say I have two. In this case you are saying that you have two and implicating that you dont have three. These really are cases of implicature, unlike the above previous cases of impliciture. I am not suggesting that reclassifying most scalar implicatures as implicitures would have any great consequence for Larrys or anyone elses work on the subject, but I see no reason to keep calling them implic-a-tures. 10. Conventional implicatures are implicatures. As I see it, the category of conventional implicature needlessly complicates Grices distinction between what is said and what is implicated. Even worse, it includes under one heading two quite different phenomena, each of which is really a case of something else. The first involves expressions like but and still. Grice claimed, as had Frege long before him, that the conventional meanings of such terms make contributions to the total import of a sentence without bearing on its truth or falsity. In She is poor but honest, for example, the contrast between being poor and being honest due to the presence of but is supposedly implicated but not stated. Grice based his claim primarily on the intuition that one would be speaking truly even if the contrast does not hold, provided the conjunction does hold. But this implies that you would have said nothing less with She is poor and honest than with She is poor but honest. To me, thats counterintuitive. Grice observed that conventional implicatures are detachable but not cancelable, but thats no argument for them. It does distinguish them (if they are there to distinguish)
14

from conversational implicatures, which are cancelable but not detachable (except for those induced by exploiting the maxim of manner, which are not detachable because they depend on how one puts what one says), and from entailments, which are neither cancelable nor detachable. However, detachability is not an independent test. If a putative implicature really were part of what is said, one could not leave it out and still say the same thing. To use and rather than but, for example, really would be to say less. And thats how it seems to me. To say that she is poor and honest is to say less than that she is poor but honest. Similarly, to say that conventional implicature is widely accepted is to say less than that conventional implicature is still widely accepted. The second kind of case is connected to Grices suggestion that conventional implicature is involved in the performance of noncentral speech acts. He had in mind uses of such expressions as after all, by the way, for example, frankly, furthermore, in conclusion, in other words, and to digress to comment on the very utterance in which they occur -- its point, character, or place in the discourse. However, the second-order speech acts these utterance modifiers are used to perform dont seem to be mere implicatures. For example, in uttering Frankly, Dr. Payne is a quack, you are not implicating but explicitly indicating that you are speaking frankly. The Obvious Conclusion Ive enumerated and briefly explained what I take to be ten misconceptions about implicature. I havent tried to give full-blown arguments for why they are misconceptions let alone to spell out their insidious consequences (but see the papers in the SemanticPragmatics Series at http://online.sfsu.edu/~kbach). Each depends on overlooking a fairly

15

obvious distinction or possibility, at least obvious once you notice it. These misconceptions are best avoided by keeping those distinctions and possibilities in mind.

16