Drawing the Line Between Meaning and Implicature - and Relating ...

June 2007



Revised



June 2007















Drawing the Line Between Meaning and Implicature –
and Relating both to Assertion









By
Scott Soames
School of Philosophy
University of Southern California

---

Drawing the Line Between Meaning and Implicature –
and Relating Both to Assertion
Paul Grice’s theory of Conversational Implicature is, by all accounts, one of the great
achievements of the past fifty years -- both of analytic philosophy and of the empirical study of
language. Its guiding idea is that constraints on the use of sentences, and information conveyed by
utterances of them, arise not only from their conventional meanings (the information they semantically
encode) but also from the communicative uses to which they are put. In his view, the overriding goal
of most forms of communication is the cooperative exchange of information -- the pursuit of which
generates norms for its rational and efficient achievement. Among them are Grice’s conversational
maxims.
Maxims of Quantity
1.
Make your conversational contribution as informative as is required (by current
conversational purposes). In other words, don't say too little.
2.
Don't make you conversational contribution more informative than is required.
Don't say too much.
Maxims of Quality
1.
Don't say what you believe to be false.
2.
Don't say that for which you lack adequate evidence.
Maxim of Relevance
Make your conversational contribution relevant to the purpose of the conversation
-- i.e. be relevant.
Maxims of Manner

2

---

1.
Avoid obscurity of expression.
2.
Avoid ambiguity.
3.
Be brief.
4.
Be orderly.
Grice uses these maxims to define conversational implicature. His picture is this. A
person utters a sentence S in a context in which it semantically expresses a certain proposition. In
most cases, this proposition is identified with what the speaker says. Although this proposition is
part of the information conveyed by the utterance, in many cases it does not exhaust this
information. In addition, the speaker also implicates certain things. Grice's notion of
conversational implicature is one particularly important kind. The definition goes roughly as
follows.i

Conversational Implicature
A speaker conversationally implicates q by saying p iff (i) the speaker is presumed to be
observing the conversational maxims, (ii) the supposition that the speaker believes q is
required in order to make his saying p consistent with the presumption that he/she is obeying
the maxims, and (iii) the speaker thinks that the hearers can recognize this requirement, and
also that they can recognize that he/she knows that they can do so.
Grice illustrates the idea in the following remark.
“The presence of a conversational implicature must be capable of being worked out; for even if it
can in fact be intuitively grasped, unless the intuition is replaceable by an argument, the
implicature (if present at all) will not count as a conversational implicature. To work out that a
particular conversational implicature is present, the hearer will rely on the following data: (1)
the conventional meaning of the words used, together with the identity of any references that

3

---

may be involved [in contemporary parlance, the semantic content of the sentence relative to the
context]; (2) the Cooperative Principles and its [conversational] maxims; (3) the context,
linguistic or otherwise of the utterance; (4) other items of background knowledge; and (5) the
fact (or supposed fact) that all relevant items falling under the previous headings are available to
both participants and both participants know or assume this to be the case. A general pattern for
the working out of a conversational implicature might be given as follows: “He has said that p;
there is no reason to suppose that he is not observing the maxims, or at least the Cooperative
Principle; he could not be doing this unless he thought that q; he knows (and knows that I know
that he knows) that I can see that the supposition that he thinks that q is required; he has done
nothing to stop me thinking that q; he intends me to think, or is a least willing to allow me to
think, that q, and so he has implicated that q.”ii (31)
Two points are worth noting. First, on the standard picture, a conversational implicature is a piece of
information conveyed, over and above what is said, or asserted – which is itself closely related to the
meaning of the sentence uttered, and perhaps identical with what would now be called the semantic
content of the sentence in the context.iii Second, in order for a proposition q to count as conversationally
implicated, the conclusion -- that the speaker believes or accepts q, and is inviting his hearers to do the
same – must, in principle, be derivable by an argument of the specified type from information available
to speaker-hearers about the meaning of the sentence uttered, the context of utterance, the conversational
maxims, and other background information in the context. A proper understanding of these two points
is crucial to distinguishing conversational implicature from linguistic meaning, and determining the
extent to which this distinction reflects something psychologically real about how speaker-hearers
process linguistic information, and interpret linguistic performances.

4

---

The relationship between assertion and implicature is illustrated by Grice’s own examples. Here
are two involving the first maxim of quantity – Don’t say too little.
(i)
A is planning to drive up the coast with B. Both know that A wants to see C, if doing so is
feasible.
A: Where does C live?
B: Somewhere in Northern California
B conversationally implicates that he doesn’t know where in Northern California C lives – since if
he did, he would be violating the first maxim of quantity by not giving the location. Here, the
presumption that he is not violating the maxims requires one to conclude that he does not take
himself to know where C lives. (Adapted from Grice 1989, 32-33.)
(ii)
A is writing a recommendation for a student who is a candidate for a philosophy job, and his letter
reads as follows: “Dear Sir, Mr. X's command of English is excellent, and his attendance in class has
been regular. Yours, truly.” A implicates that his student is no good at philosophy. Why? A knows
that a more informative letter is desired. Since he is also in a position to provide the needed
information, there must be a reason he hasn’t put it in the letter. There is no reason to think he is
being uncooperative, since if he were, he wouldn’t have written. Given that he is being cooperative,
he would surely give us a positive evaluation if he had one to give. (People are more reluctant to
state negative evaluations than positive ones, especially in writing.) Thus, his evaluation must be
negative, and he simply doesn’t want to explicitly say so. Hence, he must think the student is no
good. (Adapted from Grice 1989, 33)
In these cases the proposition conversationally implicated is clearly not asserted. In (i), the proposition B
asserts – that C lives somewhere in Northern California – directly advances the purpose of the interchange,
whereas the proposition that B conversationally implicates – that B doesn’t know where in Northern

5

---

California C lives – is merely collateral information explaining why B chose to say what he did. This fits
the classic Gricean pattern. Assertion is conceptually prior to conversational implicature, which adds
further information about why the particular assertion, rather than some other, was made. This pattern
applies to (ii) as well, even though the reasoning generating the implicature is more convoluted, and the
proposition implicated – that the job candidate is no good – is the real point of the writer’s remark.
Although this may tempt one to identify the implicature as the writer’s “real assertion,” the temptation
should be resisted -- since the whole purpose of using indirect means to convey this information was to
avoid having to state it. In cases like this, the speaker or writer exploits shared knowledge of the
conversational maxims to convey crucial information without asserting it.

This idea is taken a step further by examples involving the first maxim of quality – Don’t say
what you believe to be false.
(iii)
Irony: X, with whom A has been on close terms until now, has betrayed a secret to A’s rival, a fact
known to both A and his hearers. A says X is a fine friend. Since this remark is obviously false,
either A has violated the first maxim of quality, or he means something else by his remark. Since
there is no reason to think that he is opting out of the cooperative principle, he must be intending to
convey something else – most likely, the opposite. Note, A hasn’t asserted the proposition that X
is a fine friend, and implicated something further. He hasn't asserted the proposition expressed by
his sentence at all. Rather, the presumption that the conversational maxims have been obeyed
defeats the normal presumption that the semantic content of the sentence is included in what the
speaker says.
(iv) Metaphor: The speaker says You are the cream in my coffee. Here, the proposition literally
expressed by the sentence is false, since you refers to a person. As a result, it is obvious that
the speaker can't be trying to convey that proposition. (Quality) Nor is he trying to convey its

6

---

negation, the truth of which is too obvious to need conveying. (Quantity) Consequently, he
must be trying to convey an idea naturally suggested by the sentence – something along the
lines that the person to whom he is speaking is as crucial to him as cream is to his coffee.
((iii) and (iv) adapted from Grice 1989, p. 34.)
In both of these cases, Gricean maxims affect what is asserted. However, it is not clear that they generate
implicatures that constitute all, or part, of the speaker’s assertion. What the speaker conversationally
implicates is that he is not asserting what his sentence literally means (its semantic content in the
context), but rather something related to it – which it is up to his audience to discern. In the case of live
metaphors – in contrast with the hackneyed example in (iv) – there is often a tantalizing openness in the
content of the assertion, lending it a desired depth that challenges listeners in ways that go beyond the
banality of standard conversational implicature.iv Cases in which there is an obvious and definite
candidate for the proposition asserted – different from the one literally expressed by the sentence uttered
– should, I think, be seen as limiting cases of this process. Thus, I don’t classify the propositions asserted
in (iii) and (iv) as conversational implicatures, even though the explanation of how they come to be
asserted relies in part on Gricean maxims. Even if one objects, however, and does view them as asserted
implicatures, the way in which implicatures contribute to assertion remains highly restricted. They can
defeat the presumption that the semantic content of the sentence uttered is asserted, thereby forcing a
reinterpretation that gives rise to non-literal assertion. However, on the classical Gricean story, they do
not enter into the assertive contents of normal, literal uses of language.v

This is related to the second point noted earlier – namely the requirement that conversational
implicatures be derivable by speaker-hearers, using the standard Gricean form of argument, from their
knowledge of the meaning of the sentence uttered, and their recognition of what, literally interpreted, the
utterance can, or would, be taken to assert (plus their awareness of the conversational maxims and the

7

---

shared assumptions in the context). For Grice, the purpose of the derivability requirement is (i) to
distinguish conversational implicatures from other information, either present in the speech situation (e.g.
that someone is speaking) or conventionally encoded by the sentence uttered, (ii) to establish
conversational implicature as a normative notion, rather than a purely descriptive one about the
psychological processes at work in the real-time extraction of information from utterances, and (iii) to
ground this normative notion in knowledge that speaker-hearers actually have. Although (i) and (ii) are
well-motivated and, in my opinion, quite correct, there is a legitimate worry that the knowledge required
by (iii) may exceed what competent speaker-hearers can reasonably be expected to have.
The standard Gricean story tacitly assumes that meaning and assertion are basically transparent –
in the sense of being reliably identifiable independent of whatever further, non-semantic, nonassertive
content may be carried by an utterance. The Gricean argument grounding every conversational
implicature starts with the presumed recognition by speaker-hearers of meaning and assertive content –
what is said. From there it proceeds via the cooperative principle and conversational maxims, plus (in
some cases) special features of the context, to the conclusion that the speaker must be relying upon, and
intending to communicate, some further proposition, acceptance of which is required by adherence to
the maxims. Although it would be unrealistic to suppose that speaker-hearers actually construct
derivations of this sort every time an utterance conversationally implicates something, Grice, quite
rightly, doesn’t require this. What he does require is that every implicature be, in principle, derivable
from information that speaker-hearers really possess. The worry is that they may not always possess the
information about meaning needed to get the required derivations started.
Grice expresses a version of this worry in connection with one of his most successful, and
theoretically informative examples. In logic, a disjunction is true iff either one or both of the disjuncts
are true; it is false otherwise. However, philosophers have sometimes wondered whether or has the

8

---

same meaning in ordinary language as it does in logic.vi One reason for this comes from the fact that
nearly always when we assertively utter A or B, we do so not because we already know that A is
true, or because we already know that B is, but rather because we have reason to believe that it is
unlikely, perhaps impossible, that both A and B will fail to be true. Though we don’t know whether A
is true, and we don’t know whether B is true, we have grounds for thinking that one will be true,
should the other turn out not to be. So, when we assert a disjunctive proposition, we generally have
non-truth-functional grounds for doing so – grounds which are not grounds for asserting the
proposition expressed by either disjunct. This is such a pervasive feature of our ordinary talk that it
might seem to be a special feature of the meaning of or in ordinary language that is not part of its
meaning in formal logic.
Grice argues that this feature of the use of or is matter of conversational implicature, not
meaning. Suppose one assertively utters A or B in a conversation in which one is abiding by the
conversational maxims. Suppose further that or is the logician's or, and so A or B has the standard
truth conditions given in formal logic. Then, the proposition asserted is weaker -- less informative --
than the proposition expressed by A -- as is shown by the fact that it is entailed by that proposition, but
not vice versa. The same is true of B. The first maxim of quantity directs one to make the strongest,
most informative statement one can, provided that it would be relevant, and provided also that doing so
would not violate other maxims -- including the maxim not to assert that for which one lacks adequate
evidence. Since one has asserted the weaker statement – expressed by the disjunction -- rather than the
statement expressed by either disjunct, and since, presumably, if the disjunction is relevant to the
conversation, then each disjunct is too, the presumption that one is obeying the conversational maxims
requires one’s hearers to conclude that one lacks adequate evidence to assert the statement expressed by

9

---

either disjunct. But since one must have adequate evidence to assert the statement expressed by A or
B, one must have non-truth-functional grounds for it.

So, someone who asserts the proposition expressed by A or B typically conversationally
implicates that he/she doesn't know, or have adequate grounds to assert, the proposition expressed by A, or
the proposition expressed by B, but that he/she does have adequate grounds for thinking that both can't be
false. This is the feature of ordinary use that led to the suspicion that or has a sense in ordinary language
different from its sense in logic. What Grice has shown is that this suspicion is unfounded. Even if we
start with the logical or, we will end up concluding that assertive utterances employing it conversationally
implicate precisely what is implicated by utterances of ordinary disjunctions. Thus, we have no reason to
posit a special sense of or in ordinary language, distinct from its sense in logic.

Although this argument is decisive, Grice uses it to raise a larger, more troubling issue. He
imagines someone remaining unconvinced, and maintaining that it is intuitively clear that the sense of or
in ordinary language differs from its sense in logic. Though the particular complaint is unreasonable,
the larger issue is not. How, if Grice is right, could there ever be controversy about what information is
part of meaning, and what is conversationally implicated? The worry is that the Gricean model would
not seem to allow for serious controversy of this sort. It is a basic presupposition of the model that the
generation of conversational implicatures requires competent speaker-hearers to have a clear, reliable,
antecedent grasp of the meaning of the sentence uttered, and of what is said by an utterance of it. But
then, it would seem, we should be able to rely on that grasp to determine whether a given piece of
information conveyed by the utterance is, or is not, conversationally implicated. If it is part of what is
said, this should be recognized by competent speakers, and if it is conversationally implicated, there
should be a canonical derivation demonstrating it to be (as in the case of or just given). How, then,
could matters of this sort ever be controversial?

10

---