normative grounding of the cooperative … · cooperation is a necessary feature of communication...
TRANSCRIPT
NORMATIVE GROUNDING OF THE COOPERATIVE PRINCIPLE
BRIAN ROBINSON Abstract Cooperation is a necessary feature of communication because it allows others to understand our meaning. Paul Grice drove home this point with his cooperative principle. Though widely accepted by philosopher and linguists, the moral aspect to our cooperation in language has so far been overlooked. In this paper I will argue that observing the cooperative principle is a moral obligation. To this end, I will begin with a review of Grice and then develop a problem with his original formulation of the principle. The problem, I argue, can be solved by appeal to rational virtue theory, which simultaneously will establish the normative of a revised version of the principle that I call the Communicative Cooperative Principle. I then argue that fulfilling this obligation should require plausible utterances instead of true ones. Finally, I conclude with a discussion of the view that emerges – virtue-theoretic semantics – of which a normatively grounded theory of communicative cooperation is the foundation. Keywords: Cooperation, virtue, normativity, plausibility Word Count: 5,362
ROBINSON 1
Since the 1960s, when preliminary versions of “Logic and Conversation” (1975a, 1989)
began circulating as mimeographs among graduate students and professors, Paul Grice’s
Cooperative Principle (and the corresponding conversational maxims) have been the subject of
much debate and discussion, as well as plenty of revisions and alleged refutations. “Make your
conversational contributions 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” (Grice 1989: 26). In that
time, Grice’s Cooperative principle has become a cornerstone for the study of pragmatics by
linguists and philosophers of language. My aim in this paper is to argue that speakers have a
moral obligation to observe this principle. To that end, I will outline a new type of virtue ethics,
based upon some incomplete suggestions left by Grice. The key idea of this theory (for present
purposes at least) is that certain behavioral dispositions (or virtues) that are optimal for everyone
as means for realize their various ends, without requiring that everyone’s ends being identical. By
conceptually beginning with this normative theory (though I’ll initially motivate the issue by
showing a problem with Grice’s theory), I can derive a modified version of the Cooperative
principle, while also correcting a fundamental problem in its original. I then recommend
shifting from a maxim of truthfulness to a maxim of plausibility. Finally, I discuss how a
normative Cooperative principle sets the foundation for a new view that I call virtue-theoretic
semantics.
Conversational Cooperation
Before looking at issues of normativity, it is helpful to review Grice’s theory of
implicature, primarily as articulated in his (1975a). For Grice, the most basic notion of meaning
is that of what a speaker S means by uttering a sentence x on a given occasion. What S means is a
ROBINSON 2
function of certain (rather complex) audience-directed intentions he or she had in making that
utterance (Grice 1957, 1969; Schiffer 1972). But not everything a speaker means is conveyed
directly by the words she or he uses. This fact motivates Grice’s distinction between meaning
and saying. What someone has said (in Grice’s sense of the word) is “closely related to the
conventional meaning of the words (the sentence) he has uttered” (1989: 25). But what someone
has said, on a given occasion, by uttering some sentence x, need not be identical to what he or
she meant by uttering x. That’s because speakers can implicate things they do not literally say,
that is, they may imply, suggest, insinuate or hint at things they do not actually say. In this gap
between what is said and what is meant, Grice interjects implicature, which is what the audience
can take the speaker to have communicated beyond (or instead of) what was said, on the
assumption that the speaker is rational. This rationality requirement is a key component that
most commentators on Grice note in passing but fail to develop.
For Grice, implicatures come in two varieties, but he places the more emphasis on
conversational implicature.1 The audience should be able to work out or derive what a speaker
conversationally implicated on the assumption that the speaker’s utterances have certain features.
Speakers typically don’t make random, disconnected remarks. Rather, they tend to be rational
and cooperate with a common purpose or direction when they speak. On the assumption that
speakers are observing the Cooperative principle, the audience can interpret what was said and
figure out what they conversationally implicated (Grice 1989: 26, 31; Neale 1992: 78). To
observe the Cooperative principle, Grice claims, speakers will either also follow or ostentatiously
float his for conversational maxims of Quality, Quantity, Manner, and Relation, to which the
audience can also make reference when working out what was implicated.
1 Grice (1989) also describes conventional implicature, but that is not relevant here.
ROBINSON 3
There are, I think, two possible versions of the Cooperative principle, the first of which
I’ll introduce now.2 The first is what I am calling Conversational Cooperation, with a
corresponding Conversational Cooperative Principle:
ConvCoPrin Make your conversational contribution such that it is contributes to the realization of the goal(s) roughly shared with your interlocutor(s).
Conversational cooperation is how many have understood Grice’s original Cooperative principle
(cf., Kasher 1976, Neale 1992, Davis 2007, Levinson 1983, for instance). It explicitly requires
that interlocutors must have some roughly common goal(s) in order to have a conversation.
The fundamental problem for ConvCoPrin is that there isn’t always any common goal
for a conversation. Sometimes, interlocutors’ agendas for a conversation don’t overlap at all.
Politicians face this from the press: they’d sometimes (often?) rather not be forthright and
forthcoming in their answers to reporters. The journalist and the politician each have distinct
purposes for the conversation. And yet the tradition of interviewing politicians survives.
Cooperation, as Grice conceives of it, is too narrowly defined, and there are countless counter-
examples of non-cooperative conversations. They occur all the time in courtrooms, in interviews,
and even among family members. The interlocutors have radically different agendas, yet they are
still able to understand one another and conversationally implicate.
So I propose to divide all conversations into two categories: coordinated and
uncoordinated. Coordinated conversations are those where the interlocutors have some purpose
or goal in common for what the conversation is about or should accomplish. The goals of each
2 Which of the two is the correct interpretation of Grice and what he really meant, I leave
aside for other Gricean exegetes. Additionally, whether the second type of Cooperation Principle is a modification of Grice’s original or a completely new principle is also an matter of textual interpretation that is nonessential to the task at hand.
ROBINSON 4
interlocutor need not be precisely identical so long as they sufficiently overlap. For instance, I
may want to accomplish goals G in our dialogue, while you want to accomplish G and K; we’ve
enough in common to have a coordinated conversation. How the interlocutors come to have
common goals is of no consequence here. All might begin the conversation with the same ends
in mind. Alternatively, one may adopt the goal of another speaker by choice, trickery, or
coercion. Interlocutors may also develop new goals as a conversation progresses. Uncoordinated
conversations, on the other hand, lack that common goal. Your goal may be only K, while mine
is only G. That would be an example of a non-hostile uncoordinated conversation. A hostile
version would be when you want to accomplish G in our dialogue and I want not-G, and we can’t
both get what we want. Nevertheless, regardless of whether hostile or not, to count as rational
each interlocutor in an uncoordinated conversation must have some goal or reason for engaging
the conversation. That fact holds the hope of dealing with uncoordinated conversations. But
first let me explain why the problem of uncoordinated conversations is in fact a problem for
conversational cooperation and ConvCoPrin.
Consider the following two cases. In the first, A asks B a very embarrassing question.
Rather than respond, B just looks A in the eyes and says nothing. I take it as clear to A (and
anyone watching) that B is implicating he does not want to answer the question and will not do
so. In that case, B implicated this without saying a word.3 B has refused to cooperate with the
purpose of the conversation as established by A’s question. Yet, B nevertheless was able to non-
3 In a face-to-face conversation such as this, it is highly likely that the meaning is partly
conveyed by B’s body language, though I’ve tried to leave that out. To see that the implicature does not depend on the body language, consider the same conversation over instant message, text, or email. If B regularly responds very quickly and A knows this, then no response after even after a few minutes still conveys the implicature.
ROBINSON 5
cooperatively conversationally implicate his meaning. That is, it’s at least non-cooperative as
ConvCoPrin conceives of cooperation.
Second, consider a police interrogation. A police officer and an accused criminal have
radically different goals during an interrogation. (Let’s imagine he’s guilty.) Does this difference
in goals mean that they can’t understand one another’s meaning or implicature? Not at all.
Suppose the cop asks “Where were you on the night of the 25th?” The criminal might answer,
“Gee, I don’t know. The moon, maybe.” He has obviously said something false. But he has
implicated, and expects the interrogator to recognize, that he is implicating, that he is not going
to say where he was. And the cop is perfectly capable of recognizing this conversational
implicature. One might say that in a case like this the accepted and shared purpose of the talk
exchange is at a higher-order level – namely to have a conversation – since that would seem to be
the only goal they have in common. Each interlocutor then makes his or her conversational
contributions as relevant, truthful, etc. as is required given that purpose. Grice endorses this
view, calling this second-order goal “a simulation” and “spurious, apparent rather than real”
(1989: 370).
This second-order approach is entirely wrong for three reasons. First, it is circular.
They’re conversationally cooperating in order to have cooperative conversation. That gets us
nowhere. Second, if asked their intentions or goals, speakers in conversations without a common
goal won’t report that their intention was anything like simply having a conversation. The police
interrogator would say she is trying to get incriminating information (or even better, a
confession), while the criminal is trying to conceal information. Those are their goals for the
conversation, which still managed to occur without a common purpose or goal. Third, such a
vague second-order goal isn’t helpful to the audience when working out what a speaker
ROBINSON 6
conversationally implicated. In order for a hearer H to understand what a speaker S implicated,
H has to make reference to S’s intention(s) that motivated the utterance. Having goals in
common is quite helpful to H, but only so long those goals are sufficiently specific. If the only
goal a speaker appears to have is to have a conversation, then that goal is of little use to H in
determining if S implicated anything, and if so, what. Consider the police interrogation from
before. The criminal has clearly implicated that, contrary to what he said about his whereabouts,
he will not tell where he was that night. And the cop can work out that conversational
implicature, but only because she realizes the criminal’s intention or goal of not revealing the
information she wants. If the cop could only refer to their shared goal to have a conversation,
she wouldn’t be able to work out his meaning. Second-order goals are explanatorily useless.
Without being able to appeal to second-order conversational goals, conversational
cooperation is left untenable. There is a large group of conversations to which it doesn’t apply.
This fact would perhaps not be troubling if conversational implicatures didn’t occur in these
conversations. But they do. So we are in need of a new conversational principle that governs our
rational utterances. Though I’ve set up the issue by focusing on implicature and language, the
problem can only be solved by appeal to a normative ethical theory. Luckily, Grice provides
some aid in developing a new kind of virtue ethics from which a revised and normatively
grounded version of the Cooperative principle can be derived.
Rational Virtue Theory
Grice did not leave behind a theory of ethics. What he did leave were two little-known
articles that separately take up issues in psychology and ethics, “Method in Philosophical
Psychology” (1975b) and “Some Reflections about Ends and Happiness” (2001) respectively. By
ROBINSON 7
combing the ideas in these articles, I can construct a new normative view that I call Rational
Virtue Theory, which is essentially Gricean in tone, if not exactly what he envisioned. For the
present focus on cooperation in language use, we don’t need the entire theory, so I will mention
the basic points relevant to the task at hand.
In his (1975b), Grice develops what he calls the “genitorial position” for positing and
justifying psychological laws. Imagine you are a species-designer (or a “genitor” as Grice calls it)
and are asked to create the psychological laws, concepts, and ends, for, say, a squirrel with an eye
to optimizing its chances for survival. Whatever psychology you postulate is then justified based
on how well it explains and predicts the behavior of real squirrels. For Grice, morality emerges
when we turn to creatures complex and intelligent enough to design themselves. We’ve done
well if the psychology we postulated for them is the same one they would come up with for
themselves. Additionally, by being able to place themselves in the genitorial position, they will
be able to adjust their ends as they see fit. Grice contends that such complex and intelligent
creatures will be capable of producing a general and universal set of ethical guidelines that, when
observed, will optimize the likelihood for all creatures of the same type to realize those ends they
set for themselves. Such self-genitorial creatures, Grice says, “might indeed be ourselves (in our
better moments, of course)” (1975b: 41).
In his posthumously published (2001), Grice presents his unsystemitized views on ethics.
He follows Aristotle in focusing on eudaimonia or happiness.4 He differs from Aristotle in two
4 Grice initially is non-committal on whether the English word ‘happiness’ sufficiently
captures the notion of eudaimonia. However, he spent most of “Reflections” referring to happiness and his student and commentator Richard Warner does the same. So even if there is some difference between the two concepts, Grice’s interest is happiness. As Warner (1986) noted, Richard Kraut provides an argument (“Two Conceptions of Happiness,” The Philosophical Review, vol. 88, no. 3, April 1979) that this is a good translation.
ROBINSON 8
important ways. First, Grice does not assume any virtues from the outset (and then proceed to
argue that they lead to happiness as Aristotle does). Second, while Aristotle regards eudaimonia
as a single, monolithic end (at least as he is often interpreted), Grice considers it to be a set of all
ends that are both maximally final (ends that are desired for their own sakes alone) and self-
sufficient (any end that by itself makes life desirable).5 But the number of ends in this happiness
set are far too numerous and diverse for anyone to realize all or even most of them in a lifetime.
So from this happiness set, each person chooses his or her own goals, a life plan unique to that
person. Some life plans are better than others, however, based on how likely they are to be
realized by an individual. If one’s life plan is too large or diverse – such as, for instance,
becoming a billionaire and having a career as a high-school teacher – then that person has chosen
poorly and isn’t likely to realize enough to consider herself or himself happy.
Grice’s thoughts, hints, and suggestions serve as a starting point for what I have termed
Rational Virtue Theory, which though based on Grice, stands as a significant advancement
beyond his comments. For the time being, we only need to note a few aspects of this theory that
are relevant to our discussion of the Cooperative principle. Not surprisingly, it is a type of virtue
ethics, though with an important difference from many of others. Instead of beginning with the
virtues and then working to demonstrate their efficacy of realizing happiness, rational virtue
theory derives the virtues. Specifically, we can call whatever dispositions tend to lead to
happiness to be virtues.6 Honesty, for example, may prove to lead to happiness – and so
5 Both finality and self-sufficiency are Aristotlean notions. See Nicomachean Ethics I.7. 6 Julia Driver (2001) has a similar notion. She defines virtue contextually as “character
traits that systematically produce more actual good than not” (2001: 68). Defining virtues in this way goes along way to help rational virtue theory avoid the recent challenges from situationism (cf., Doris 1998, 2002; Harman 1999, 2000, 2001, 2003, 2006; Alfano forthcoming; Vranas 2005). Rational virtue theory need not appeal to global character traits (to which Doris,
ROBINSON 9
legitimately be a virtue – but that can’t be assumed at the outset. For our immediate purposes,
we need to focus specifically on what I call behavioral principles:
Behavioral principle: a ceteris paribus default behavioral strategy (or behavioral disposition) capable of repetition and replication that is stable and rational.7
By repetition, I have in mind a requirement for wide applicability. Behavioral principles don’t
pertain to unique, one-time situations. They cover a multitude of related scenarios in which one
is liable to find oneself. It must also be the case that you are able to repeatedly perform the
action called for by a behavioral principle in all (or most) relevant circumstances. Requiring that
behavioral principles can be replicated means that others, seeing how well possessing a behavioral
disposition works for someone else, are able to replicate the same behavior, deciding to acquire
the disposition for themselves. This idea of replication is based on replicator dynamics, which is
one of the key tools employed by evolutionary game theory (cf. Skyrms 2004.) Essentially, the
idea is that a successful behavioral strategy, one that is stable and increases utility, will spread,
being picked up by others. Success breeds imitation. To say that a behavioral principle is
rational means that performing the action(s) it advocates is in your self-interest, where self-
interest is defined by the ends you choose. Thus, this moral psychology is a form of rational
egoism. Self-interest in rational virtue theory though is largely left to each individual to decide
upon for themselves, so that seemingly altruistic behavior can be countenanced as rational if a
person adopts someone else’s ends as his or her own. For instance, self-sacrifice by a parent for a
child can be rational if the parent has opted to make the well-being of the child a part of his or Harmon, etc. object on empirical grounds). Instead, it countenances local character traits, which remain empirically plausible (cf. Doris 2002: 25). Rejection of Aristotle’s reciprocity thesis (Nicomachean Ethics 1144b30-1145a2) helps avoid this challenge as well.
7 I am here only focusing on behavioral dispositions, not denying other types of dispositions, such as emotional ones.
ROBINSON 10
her life plan, that is a necessary achievement for the parent’s own happiness.8 By stability, finally,
I mean much the same thing as Maynard Smith and Price (1973) did by an evolutionary stable
strategy. A stable behavior principle is one that remains rational even if most or all other agents
also adopt it as well, while simultaneously being successful when confronted with alternative
behavioral strategies.
Communicative Cooperation
Returning back to cooperation and language use, what we’re looking for is some new
version of the cooperative principle that we can derive from my outline of a Gricean moral
psychology that will also be able to handle cases of uncoordinated conversation. When
examining the faults of conversational cooperation, we noticed that in uncoordinated
conversations, though the interlocutors lacked a common purpose between them, each of them
still has his or her own goal(s) or reason(s) for engaging the conversation. With this fact in mind,
I therefore propose a different notion of cooperation in language use, namely Communicative
Cooperation, and with it the Communicative Cooperative Principle:
CommCoPrin Make your conversational contribution such that your interlocutor(s) is sufficiently likely to be able to understand your communicative intentions and meaning, and presume that your interlocutor(s) does so as well.9
8 Accounting for and rationalizing instances of altruism in this manner should answer
objections against egoism offered by Feinberg (1978). 9 While all interlocutors do have the goal of being understood, this isn’t a goal they have
in common; rather our goals are similar. Note the indexicalilty of CommCoPrin. For instance, I have my goal of communicating something to you; you have your goal of communicating something to me. Even if the something we’re both trying to communicate is the same thing, our goals are distinct.
ROBINSON 11
Contrary to the standard Gricean view, the main purpose or goal of each interlocutor is to be
understood. They can have just about any goal they want, e.g., to avoid incriminating oneself.
An interlocutor is only communicatively cooperative in order to satisfy those goals. To support
this notion of cooperation, I need to establish two points: first, that CommCoPrin still counts as
a form of cooperation; and second, that CommCoPrin handles instances of uncoordinated
conversation that Grice’s original Cooperative principle does not.
One might think that communicative cooperation is cooperation in name only. How can
it be cooperative if all interlocutors are working with different purposes? They are not working
together to achieve anything. The reason why communicative cooperation is cooperation can be
seen in considering how Humpty Dumpty wasn’t cooperative. In Through the Looking-Glass,
Carroll presents this conversation between Humpty Dumpty and Alice:
“And only one for birthday presents, you know. There’s glory for you!” “I don’t know what you mean by ‘glory,’ ” Alice said. Humpty Dumpty smiled contemptuously. “Of course you don’t—till I tell you. I meant ‘there’s a nice knock-down argument for you!’” “But ‘glory’ doesn’t mean ‘a nice knock-down argument’,” Alice objected. “When I use a word,” Humpty Dumpty said, in a rather a scornful tone, “it means just what I choose it to mean—neither more nor less.”10
Humpty Dumpty is not being communicatively cooperative. Alice could not have known that
Humpty Dumpty meant ‘a nice knock-down argument’ when he used ‘glory.’ So he is a fool if
he expected Alice to understand what he meant, and blatantly (and annoyingly) uncooperative if
10 In his (1896) Symbolic Logic Part I, Carroll states, “I maintain that any writer of a book is fully authorized in attaching any meaning he likes to a word or phrase he intends to use. If I find an author saying, at the beginning of a book, ‘Let it be understood that by the work “black” I shall always mean “white”, and by the word “white” I shall always mean “black”, I meekly accept his ruling, however injudicious I think it.” The difference from the Humpty Dumpty case is that this hypothetical author tells his audience beforehand. He is at least being somewhat communicatively cooperative, though he could have been more so.
ROBINSON 12
he didn’t but said it anyway.11 Assuming Humpty had any communicative intention, his use of
‘glory’ in such a non-standard way made it far less likely that his intention could be realized by
Alice understanding him. This fact would hold regardless of Alice’s goals and intentions and
how they matched Humpty’s.12 So, communicative cooperation does not require interlocutors to
have sufficiently similar communicative goals. Rather, all that is required is for interlocutors to
recognize and respect the other’s communicative expectations (such as for word meaning and
syntax). The notion of cooperation, then, is expanded beyond our earlier definition. A non-
linguistic analogy is the formation of coalitions between diverse political parties to form a
government. The Conservatives and the Liberal Democrats of Great Britain, who have largely
very different political aims, were nevertheless able to cooperate and form a parliamentary
majority.13 Cooperation can come in many varieties; having shared goals is only the most
common. We can also cooperate with someone’s expectations of us, but solely for self-interested
reasons that the other does not share.
11 The latter option is an interesting possibility. If he meant to say something that he
believed Alice couldn’t have understood, then Humpty Dumpty succeeded in realizing that intention, and did so with being communicatively cooperative. But in that case he didn’t have a communicative intention either.
12 This discussion is also meant to counter the possible objection of Humpty-Dumptyism, which has been raised before against Gricean accounts of meaning based in intention. Just because your intentions determine your meaning doesn’t mean you can successfully mean anything you want with an utterance. It would be irrational to do so, and you’re obligated not to.
13 One might respond by claiming that the leaders of both parties had a second-order goal of forming a government. However, the first two objections I raised earlier to second-order conversational goals still apply to this non-conversational example. The third objection (about working out conversational implicature) wouldn’t directly apply, but a similar line does. If the Liberal Democrats are trying to understand why Conservative PM David Cameron did something (such as voted yea or nay on an important vote), appeal to any second-order goal of forming/maintaining a government is at least less helpful and informative than appealing to his first-order goals of what he wants government to do.
ROBINSON 13
Regardless of whether a conversation is coordinated or not, a speaker has a reason to
observe CommCoPrin. It is applicable in any and all communicative exchange, regardless of
whether or not a shared purpose or direction has been or ever will been set for the conversation,
unlike the conversational cooperative principle. There is no limitation as of yet regarding what
you might intend to communicate. So, it is always applicable anytime you have a communicative
intention. The communicative cooperative principle, therefore, is already superior to Grice’s
original formulation of the Cooperative principle. The real test, however, is whether or not
CommCoPrin is theoretically useful in explaining conversational implicature.
Maxims, Truth, and Plausibility
Grice’s Cooperative principle did not specify how to make your conversational
contribution match what was required. Similarly, CommCoPrin does not explain how to make
your conversational contribution such that it will be sufficiently likely to be understood. To fill
in how to be conversationally cooperative, Grice provides his conversational maxims of Quantity,
Quality, Relation, and Manner, which he thought were derived from the Cooperative principle.
The problem, however, is that faults we found with conversational cooperation appear to filter
down to the conversational maxims. If, in a hostile uncoordinated conversation, I have no reason
to cooperate with you toward your conversational goal, then I have no reason to observe the
maxims either. One is supposed to observe the maxims as a way to be conversationally
cooperative.14
But what if there is no reason to be conversationally cooperative, do you have any reason
to observe the maxims? We can only motivate one to observe the conversational maxims if they
14 One can, of course, ostentatiously violate a maxim and still be conversationally cooperative, as Grice notes. This fact, however, doesn’t undermine the point that the
ROBINSON 14
are derived from CommCoPrin instead. To show that, I will demonstrate that in cases of
uncoordinated conversations, it is still rational for an interlocutor to observe the maxims.
Imagine a conversation between a reporter and politician. The politician wants to talk about his
accomplishments in office, while the reporter wants to talk about a breaking scandal. So this is a
uncoordinated, perhaps hostile, conversation. Why should the politician observe the maxim of
Manner? The politician is looking for rational, effective means by which realize his
communicative intentions, i.e., to be communicatively cooperative. That will mean observing at
least three, and often all four, of Grice’s conversational maxims. For the maxim of Manner, it
hardly seems necessary to argue that brevity, order, and lack of ambiguity or obscurity are
efficient means to communicate your intended meaning. For Quantity, you want to give as
much information as is required but not more than that, since giving too little or too much
information can obscure your meaning. We only need to remember that how much information
is the right amount is based on your communicative intention, not necessarily some shared
conversational goal. Likewise with Relation, what you say should be relevant not necessarily to a
common purpose, but to your goal of getting your audience to understand your communicative
intention and meaning.
The one potential exception is the maxim of Quality. People sometimes have the
communicative intention to lie or deceive, as did our criminal. Must they observe Grice’s maxim
of Quality, demanding that they only express their justified true beliefs? Grice apparently
thought so. Campbell (2001: 256) perhaps expressed it best when he said, “The maxims
presuppose an almost Utopian level of gentlemanly conduct on the part of a speaker and an old-
fashioned standard of truthfulness that George Washington might have found irksome. They
remind one of the early Puritanism of The Royal Society.” Communicative Cooperation, on the
ROBINSON 15
other hand, isn’t so strict. When it comes to false communicative intentions, there are three
options. We can appeal to a theory of moral psychology that contends it is irrational to have
such intentions in the first place. This option has potential, though the theory of moral
psychology will have a difficult task. Alternatively, we could throw out the maxim of Quality all
together, claiming that its observance is not requisite for communicative efficiency. This option
does not look promising, since intuitively there is some connection between honesty and
communicative efficiency, even if Grice overstated it.
A final option is for the maxim of Quality can be reworked to emphasize plausibility
instead of justified true beliefs. This, I think, is the most promising option. In general, the
point of a communicative intention is to get your interlocutor to believe what you’re trying to
communicate. If you’re trying to communicate to someone that the sky is falling and that is
highly dubious claim to your interlocutor at the outset of the conversation, it would be foolhardy
to begin the conversation by simply asserting that the sky is falling. You would likely fail to
realize your communicative intention. A more rational strategy would be to start by saying
something what will be found plausible, from which you can then build until your eventual
assertion that the sky is falling has become more believable.
Initially it might seem that judging the quality of an utterance based on its plausibility
rather than truthfulness goes too far. Some might see it as an abandonment of a commitment to
truth in our discourse, which could only serve to undermine our communicative practices. After
all, didn’t Kant show that lying, when universalized, entails a contradiction? In short, this
objection fails to realize the connection between truth and plausibility. The requirement of
plausibility is a ceteris paribus requirement for truthfulness. That is because typically a true
ROBINSON 16
assertion will be more plausible to an interlocutor than a false one. So telling the truth is usually
an efficient means by which to satisfy the obligation to plausibility.
Requiring speakers to be plausible, however, has some real advantages over Grice’s
original requirement for justified true beliefs. It allows for the possibility of the noble lie. The
history of noble lies is long and storied, but the basic idea is that sometimes a falsehood can be
justified because, by lying, you bring about a benefit for someone else. In the Republic, Plato
called the myth of the metals a noble lie, told to induce patriotism and a sense of duty-
according-to-capability in each citizen of the kallipolis (414b). To jump all the way to
contemporary work, Alfano (forthcoming) persuasively argues that though people cannot actually
have virtuous (or vicious) global character traits, we can stimulate factitious virtues in people by
lying to them about how virtuous they are, with the result that those lied to will tend more often
to behave virtuously in certain circumstances. Grice’s “old-fashioned” demand for absolute truth
can’t permit any such noble lie. Furthermore, CommCoPrin, along with a maxim of Plausibility,
can countenance some instances of lying or deception. Sometimes it is moral or rational to lie.
When Kant’s murderer at the door comes knocking, inquiring about the person I’m hiding, I
should lie to him. My communicative intention is to convincingly deceive the murderer (and so
save his would-be victim), and requiring my utterances to be plausible is an efficient way to do
that. So it isn’t that you should always tell the truth. Rather, you should always say something
plausibly true, which typically will mean saying something true.
Normativity
To wrap things up, I would like to return to the issue of the normativity of
communicative cooperation. One might think that by asserting that CommCoPrin places a
ROBINSON 17
moral obligation on speakers, I am merely tilting at windmills, since it is obvious Grice had
moral notions in mind when he introduced the Cooperative principle. Unfortunately, that point
has been anything but obvious to most of Grice’s commentators. Most, when they briefly
mention the Cooperative principle before focusing the majority of their attention on the maxims,
make no mention of morality whatsoever, either to affirm or deny its moral status (cf. Kasher
1976, Neale 1992, Davis 2007, Levinson 1983). Several others flatly deny the Cooperative
principle any moral status. Brown and Levinson (1978), for instance, call it a “socially neutral
(indeed asocial) presumptive framework for communication.” It’s something, they think, that is
simply necessary for communication to occur, but not morally obligatory to observe. Davies
(2007) argues that the cooperative principle is often mistakenly associated with the common
notion of cooperation, which is not central to Grice’s views on language. Bach and Harnish
(1979) regard the cooperative principle merely as sort of default setting for speakers. Horn
(2004) is the chief example of non-moral interpreters of Grice. He follows Bach and Harnish in
thinking of the cooperative principle as expressing a presumption audiences have of speakers, not
because speakers ought to observe it, but because they typically do. He states, “But neither the
cooperative principle nor the attendant maxims are designed as prescriptions for ethical action”
(2004: 8).
Though Grice originally stated the cooperative principle as an imperative (a feature I
preserved in the communicative cooperative principle), most commentators have missed or
denied its normative status.15 Supporting my claim that it should be understood normatively, we
can look at some of Grice’s own comments. In his “Meaning Revisited,” Grice remarks,
15 Saul (2001) tries to introduce a notion of normativity to that of implicature. However,
Saul’s normativity neither works nor correctly understands Grice’s theory. In her flawed view, a
ROBINSON 18
[S]omething has been left out, by me and perhaps by others too, in the analyses, definitions, expansions, and so on, of semantic notions, and particularly various notions of meaning,… the notion of value…. The notion of value is absolutely crucial to the idea of rationality, or of being a rational being…. I have a strong suspicion that the most fruitful idea is the idea that a rational creature is one that evaluates.… Value is in there from the beginning, and one cannot get it out (1989: 297-8).
For Grice and most of the theorists that follow him, cooperation in language is fundamentally a
rational activity. It is not often recognized, however, that Grice goes further by claiming that
rationality fundamentally entails value. And rationality is central to his understanding of
conversations and communication. Returning to “Logic and Conversation,” he states,
So I would like to be able to show that observance of the Cooperative Principle and the maxims is reasonable (rational) along the following lines: that anyone who cares about the goals that central to conversational/communication (such as giving and receiving information, influencing and being influenced by others) must be expected to have an interest, given suitable circumstances, in participations in talk exchanges that will be profitable only on the assumption that they are conducted in general accordance with the Cooperative Principle and the maxims” (1989: 29-30, emphasis mine).
For Grice, in our use of language we cooperate with one another, and we do so because it is
rational. And rationality entails value. So Grice anticipated this connection between rationality
and normativity. It is also important to note again that the notion of rationality that Grice adopts
in “Logic and Conversation” that he doesn’t develop there. To understand what he means by
rationality, we have look to his (1975b) and (2001), which we did above. That led us to a
Gricean theory of moral psychology, from which CommCoPrin was derived. Conversation –
how we communicate with each other – is not some special, amoral type of rational activity. The
failure to fulfill your normative obligation to make what you meant clear to your audience entails that you didn’t mean anything. This is a complete misunderstand of Grice’s intention based theory of meaning, where the meaning of your utterance is determine solely by what you intended it to mean. Davis (2007) follows Saul in this error.
ROBINSON 19
demands of ethics are just as real there as in the rest of what we do as rational, value-driven
people.
There remains a final point to mention about normativity. With communicative
cooperation, I’ve argued interlocutors are morally obligated to cooperate with one another. This
normative grounding lays the foundation for a new view I refer to as Virtue-theoretic
Semantics.16 The view is a complex one, but I can here summarize four key elements of the view.
First, philosophy of language is fundamentally a normative discipline and must be understood as
such. Second, there is such a thing as semantic value, which is a property of utterances, and
interlocutors are its source. Third, semantic evaluation must focus on semantic value, which is
judged based on the effectiveness of realizing a speaker’s communicative intentions (which are in
turn based on the speaker’s values). Fourth, there are semantic virtues which, when observed,
raise an utterances semantic value. As a final point, virtue-theoretic semantics is only possible if
normativity is built into our use of language a fundamental level. Communicative cooperation is
such a level, for communication couldn’t occur if we failed to meet this obligation.
Bibliography
Alfano, M. (forthcoming). Factitious Virtue. Cambridge: Cambridge University Press. Aristotle. (1925). Nicomachean Ethics. (D. Ross, Trans., J. L. Ackrill and J. O. Urmson, Revised).
Oxford: Oxford University Press. Bach, K., and Harnish, R. (1979). Linguistic Communication and Speech Acts, Cambridge, MA:
MIT Press. Brown, P., and Levinson, S. C. (1978). Universals in language use: Politeness phenomena, in E.
Goody (ed.), Questions and Politeness, pp. 56-311, Cambridge: Cambridge University Press.
16 Though there will be some overlap with discourse ethics (cf. Habermas 1990, 1993),
important and fundamental differences will remain. To briefly note two, the first and most basic is that mine is a virtue theory, while discourse ethics is Kantian. A second major difference is the function and method of universalizing normative claims to all speakers.
ROBINSON 20
Campbell, J. (2001). The Liar’s Tale: A History of Falsehood. New York: W. W. Norton & Company.
Davies, B. (2007). Grice’s Cooperative Principle: Meaning and Rationality. Journal of Pragmatics, 39: 2308-2331.
Davis, W. (2007). How Normative is Implicature. Journal of Pragmatics, 39: 1655-1672. Doris, J. M. (1998). Persons, situations, and virtue ethics. Nous, 32:4, 504-540. Doris, J. (2002). Lack of Character: Personality and Moral Behavior. Cambridge: Cambridge
University Press. Driver, J. (2001). Uneasy Virtue. Cambridge: Cambridge University Press. Feinberg, J. (1978). Psychological Egoism. In Joel Feinberg (Ed.), Reason and Responsibility,
fourth edition (and other editions), Belmont: Wadsworth. Grice, P. (1957). Meaning. Philosophical Review, 66: 377-88. Grice, P. (1969). Utterer's meaning and intentions. Philosophical Review, 78: 147-77. Grice, P. (1975a). Logic and Conversation. In D. Davidson & G. Harmon (Eds.), The Logic of
Grammar, Encino, California: Dickinson Publishing, 64-75. Grice, P. (1975b). Method in Philosophical Psychology (From the Banal to the Bizarre),
Proceedings and Addresses of the American Philosophical Association, 48: 23-53. Grice, P. (1989). Studies in the Ways of Words. Cambridge, Massachusetts: Harvard University
Press. Grice, P. (2001). Aspects of Reason. (R. Warner, Ed.). Oxford: Clarendon Press. Habermas, J. (1990). Moral Consciousness and Communicative Action (C. Lenhardt & S. W.
Nicholsen, Trans.). Cambridge: MIT Press. Habermas, J. (1993). Justification and Application (C. P. Cronin, Trans.). Cambridge, MA: MIT
Press. Harman, G. (1999). Moral philosophy meets social psychology: Virtue ethics and the
fundamental attribution error. Proceedings of the Aristotelian Society, New Series 119, 316-331.
Harman, G. (2000). The Nonexistence of Character Traits. Proceedings of the Aristotelian Society 100. 223–226.
Harman, G. (2001). Virtue ethics without character traits, in Byrne, Stalnaker, & Wedgwood, eds., Fact and Value, 117-127. Cambridge: MIT Press.
Harman, G. (2003). No character or personality. Business Ethics Quarterly, 13:1, 87-94. Harman, G. (2006). Three trends in moral and political philosophy. The Journal of Value Inquiry,
37. Horn, L. R. (2004). Implicature. In L. R. Horn & G. Ward (Eds.), The Handbook of Pragmatics,
Oxford: Blackwell Publishing, 3–28. Kasher, A. (1976). Conversational maxims and rationality. In: Asa Kasher (Ed.), Language in
Focus: Foundations, Methods and Systems, Dordrecht, Reidel, 197-216. Kraut, R. (1979). Two Conceptions of Happiness. The Philosophical Review, 88, 167-197. Levinson, S. (1983). Pragmatics. Cambridge: Cambridge University Press. Lewis, C. (1871). Through the Looking-Glass. London: Macmillian. Lewis, C. (1896). Symbolic Logic Part I. London: Macmillian. Maynard Smith, J. & Price, G. R. (1973). The Logic of Animal Conflict. Nature, 246. 15-18. Neale, S. (1992). Paul Grice and the Philosophy of Language. Linguistics and Philosophy, 15,
509-559.
ROBINSON 21
Saul, J. (2001). Review of Implicature: Intention, Convention, and Principle in the Failure of Gricean Theory by Wayne Davis. Noûs, 35: 630-41.
Schiffer, S. (1972). Meaning. Oxford: Clarendon Press. Skyrms, B. (2004). The Stag Hunt and the Evolution of Social Structure. Cambridge: Cambridge
University Press. Vranas, P., 2005, “The Indeterminacy Paradox: Character Evaluations and Human
Psychology,” Noûs 39: 1–42. Warner, R. (1986). Grice on Happiness. In R. E. Grandy & R. Warner (Eds.), Philosophical
Grounds of Rationality: Intentions, Categories, Ends. (475-493). Oxford: Clarendon Press.