The evolutionary social psychology of off-record indirect speech acts

Text-only Preview

The evolutionary social psychology
of off-record indirect speech acts
This paper proposes a new analysis of indirect speech in the framework of
game theory, social psychology, and evolutionary psychology. It builds on
the theory of Grice, which tries to ground indirect speech in pure rationality
(the demands of e‰cient communication between two cooperating agents)
and on the Politeness Theory of Brown and Levinson, who proposed that
people cooperate not just in exchanging data but in saving face (both the
speaker’s and the hearer’s). I suggest that these theories need to be supple-
mented because they assume that people in conversation always cooperate.
A reflection on how a pair of talkers may have goals that conflict as well as
coincide requires an examination of the game-theoretic logic of plausible
denial, both in legal contexts, where people’s words may be held against
them, and in everyday life, where the sanctions are social rather than judi-
cial. This in turn requires a theory of the distinct kinds of relationships that
make up human social life, a consideration of a new role for common
knowledge in the use of indirect speech, and ultimately the paradox of ra-
tional ignorance, where we choose not to know something relevant to our
The evolutionary social psychology of o¤-record indirect speech acts
Indirect speech is the phenomenon in which a speaker says something
he doesn’t literally mean, knowing that the hearer will interpret it as he
Would you like to come up and see my etchings? [a sexual come-on].
If you could pass the salt, that would be great [a polite request].
Nice house you got there. Would be a real shame if something hap-
pened to it [a threat].
Intercultural Pragmatics 4-4 (2007), 437–461
DOI 10.1515/IP.2007.023
6 Walter de Gruyter

Steven Pinker
We’re counting on you to show leadership in our Campaign for the
Future [a solicitation of a donation].
Gee, o‰cer, I was wondering whether there might be some way we
could take care of the ticket here [a bribe].
These ‘‘o¤-record indirect speech acts’’ have long been a major topic in
pragmatics, and they have considerable practical importance as well,
including an understanding rhetoric, negotiation and diplomacy, and the
prosecution of extortion, bribery, and sexual harassment. They also pose
important questions about our nature as social beings. This paper,
adapted from a book which uses semantics and pragmatics as a window
into human nature (Pinker 2007), uses indirect speech as a window into
human social relationships. In doing so it seeks to augments the current
understanding of indirect speech with ideas from game theory, evolution-
ary psychology, and social psychology.
Intuitively, the explanation for indirect speech seems obvious: we use it
to escape embarrassment, avoid awkwardness, save face, or reduce social
tension. But as with many aspects of the mind, the danger with common-
sense explanations is that we are trying to explain a puzzle by appealing
to intuitions that themselves need an explanation. In this case, we need to
know what ‘‘face’’ is, and why we have emotions like embarrassment,
tension, and shame that trade in it. Ideally, those enigmas will be ex-
plained in terms of the inherent problems faced by social agents who ex-
change information.
Background: Conversational maxims and the theory of politeness
Any analysis of indirect speech must begin with Grice’s Cooperative Prin-
ciple and the theory of conversational maxims and conversational impli-
cature that flows from it (Grice 1975). Grice proposed that conversation
has a rationality of its own, rooted in the needs of partners to cooperate
to get their messages across. Speakers tacitly adhere to a Cooperative
Principle, tailoring their utterances to the momentary purpose and direc-
tion of the conversation. That requires monitoring the knowledge and
expectations of one’s interlocutor and anticipating her reaction to one’s
words. (Keeping with convention, I will refer to the generic speaker as a
‘‘he’’ and the generic hearer as a ‘‘she.’’) Grice famously fleshed out the
principle in his four conversational ‘‘maxims,’’ quantity (say no more or
less than is required), quality (be truthful), manner (be clear and orderly),
and relevance (be relevant), which are commandments that people tac-
itly follow to further the conversation e‰ciently. Indirect speech may be

The evolutionary social psychology of o¤-record indirect speech acts
explained by the way the maxims are observed in the breach. Speakers
often flout them, counting on their listeners to interpret their intent in a
way that would make it consistent with the Cooperative Principle after
all. That’s why, Grice noted, we would interpret a review that described
a singer as ‘‘producing a series of notes’’ as negative rather than factual.
The reviewer intentionally violated the maxim of Manner (he was not
succinct); readers assume he was providing the kind of information they
seek in a review; the readers conclude that the reviewer was implicating
that the performance was substandard. Grice called this line of reasoning
a conversational implicature.
Grice came to conversation from the bloodless world of logic and said
little about why people bother to implicate their meanings rather than just
blurting them out. We discover the answer when we remember that peo-
ple are not just in the business of downloading information into each
other’s heads but are social animals concerned with the impressions they
make. An implicature involves two meanings: the literal content (some-
times called the sentence meaning) and the intended message (sometimes
called the speaker meaning). The literal sentence meaning must be doing
some work or the speaker would not bother to use it in the first place. In
many implicatures involved in o¤-record indirect speech acts, the in-
tended message is negative but the literal content is positive or neutral.
Perhaps speakers are trying to eat their cake and have it too—they want
to impugn something they dislike while staving o¤ the impression that
they are whiners or malcontents. Dews, Kaplan, and Winner (1995)
showed that people have a better impression of speakers who express a
criticism with sarcasm (‘‘What a great game you just played!’’) than with
direct language (‘‘What a lousy game you just played!’’). The sarcastic
speakers, compared with the blunt ones, are seen as less angry, less criti-
cal, and more in control.
The double message conveyed with an implicature is nowhere put to
greater use than in the commonest kind of indirect speech, politeness. In
their seminal work Politeness: Some Universals in Language Use, Brown
and Levinson (1987b) extended Grice’s theory by showing how people
in many (perhaps all) cultures use politeness to lubricate their social
Politeness Theory begins with Go¤man’s (1967) observation that when
people interact they constantly worry about maintaining a commodity
called ‘‘face’’ (from the idiom ‘‘to save face’’). Go¤man defined face as
a positive social value that a person claims for himself. Brown and Lev-
inson divide it into positive face, the desire to be approved (specifically,
that other people want for you what you want for yourself ), and nega-
tive face, the desire to be unimpeded or autonomous. The terminology

Steven Pinker
points to a fundamental duality in social life which goes by many
names: solidarity and status, connection and autonomy, communion and
agency, intimacy and power, communal sharing and authority ranking
(Fiske 1992, 2004; Haslam 2004; Holtgraves 2002). Later we will see
how these wants come from two of the three major social relations in
human life.
Brown and Levinson argue that Grice’s Cooperative Principle applies
to the maintenance of face as well as to the communication of data. Con-
versationalists work together, each trying to maintain his own face and
the face of his partner. The challenge is that most kinds of speech pose
at least some threat to the face of the hearer. The mere act of initiating
a conversation imposes a demand on the hearer’s time and attention. Is-
suing an imperative challenges her status and autonomy. Making a re-
quest puts her in the position where she might have to refuse, earning
her a reputation as stingy or selfish. Telling something to someone im-
plies that she was ignorant of the fact in the first place. And then there
are criticisms, boasts, interruptions, outbursts, the telling of bad news,
and the broaching of divisive topics, all of which can injure the hearer’s
face directly.
At the same time, people have to get on with the business of life, and
in doing so they have to convey requests and news and complaints. The
solution is to make amends with politeness: the speaker sugarcoats his
utterances with niceties that rea‰rm his concern for the hearer or that
acknowledge her autonomy. Brown and Levinson call the stratagems
positive and negative politeness, though better terms are sympathy and
The essence of politeness-as-sympathy is to simulate a degree of close-
ness by pretending to want what the hearer wants for herself. Two famil-
iar examples are the impotent bidding of good fortune (Be well, Have a
nice day) and the feigned inquiry into the person’s well-being (How are
you?, How’s it going?). One step beyond the ‘‘fictitious benevolence’’ of
politeness is fictitious solidarity. Speakers may address their hearers with
bogus terms of endearment like my friend, mate, buddy, pal, honey, dear,
brother, and fellas; use slang connected to an in-group, as in Lend me two
bucks; or may include the listener in their plans, as in Let’s have another
Politeness-as-deference (negative politeness) is invoked most of all with
commands and requests, which are among the most face-threatening
speech acts because they challenge the hearer’s autonomy by assuming
her readiness to comply. The speaker is ordering the hearer around, or
at least putting her out, something you don’t do to a stranger or a supe-
rior and might even think twice about doing with an intimate. So requests

The evolutionary social psychology of o¤-record indirect speech acts
are often accompanied by various forms of groveling, such as question-
ing rather than commanding (e.g., Will you lend me your car?), expressing
pessimism (I don’t suppose you might close the window), and acknowledg-
ing a debt (I’d be eternally grateful if you would . . .).
Politeness, according to Brown and Levinson, is calibrated to the level
of the threat to the hearer’s face. The threat level in turn depends on the
size of the imposition, the social distance from the hearer (the lack of in-
timacy or solidarity), and the power gap between them. People kiss up
more obsequiously when they are asking for a bigger favor, when the
hearer is a stranger, and when the hearer has more status or power. A
fully loaded request like ‘‘I’m terribly sorry to trouble you, and I wouldn’t
ask unless I were desperate, but I’d be eternally grateful if you think you
could possibly . . .’’ would sound smarmy if it were used to ask a stranger
for a small favor like the time, or if it were used to ask a bigger favor (like
the use of a computer) of a spouse or an assistant.
Indirect speech acts, according to the theory, are even higher up the po-
liteness scale than deferential (negative) politeness. In these speech acts, a
request is not stated baldly but conveyed with the help of an implicature.
The result is a ‘‘whimperative’’ like Can you pass the salt? or If you could
pass the salt, that would be great. Taken literally, the first example violates
the maxim of Relevance, because the answer to the question is already
known. The second one violates the maxim of Quality, because the conse-
quent of the conditional is an overstatement. So the hearer interprets
them as requests, while noting from the literal wording that the speaker
was seeking to avoid the appearance of treating her like a flunky.
Because a cliche´d indirect request is recognized as a request by any
competent English speaker, it is e¤ectively ‘‘on the record.’’ A speaker
who says Can you pass the salt? in ordinary dinnertime circumstances
cannot plausibly deny that he has asked for something. But according to
Brown and Levinson, if an indirect speech act is freshly minted rather
than pulled o¤ the shelf, its e¤ect on the hearer is di¤erent. The request
is now ‘‘o¤ the record.’’ When a speaker thinks up a novel indirect re-
quest, like The chowder is pretty bland or They never seem to have enough
salt shakers at this restaurant, the hearer can ignore the comment without
publicly rebu‰ng the request. For this reason, Brown and Levinson argue
that o¤-record indirect speech acts coined for the occasion—hints, under-
statements, idle generalizations, and rhetorical questions—are the politest
forms of all. A speaker can say It’s too dark to read as a way to ask a
hearer to turn on the lights, or The lawn has got to be mowed instead of
‘‘Mow the lawn.’’ According to politeness theory, then, with o¤-record
indirect speech the hearer is implicitly given the opportunity to ignore
the request without a public refusal, which also means that if she complies

Steven Pinker
with the request, it’s not because she’s taking orders. According to Brown
and Levinson, this saves face for both of them, especially the hearer with
her desire for autonomy.
Beyond cooperation
Politeness Theory has been tested in many experiments (see Brown &
Levinson 1987a; Clark & Schunk 1980; Fraser 1990; Holtgraves 2002),
and many of its claims have been confirmed. The use of the proposed po-
liteness strategies indeed makes a request sound more polite; indirect re-
quests sound more polite than direct ones; and the degree of imposition
matters, as does the relative power of the speaker and the hearer.
But according to several literature reviews, one claim has not come out
as well. Brown and Levinson claimed that face threat was a single scale,
the result of adding up the power disparity, the social distance, and the
degree of imposition. They claimed that the three kinds of politeness
were arranged along a scale, too. Sympathy expresses a little bit of polite-
ness, and is suitable for smaller face threats. Deference expresses more,
and is suitable for bigger ones. And o¤-record indirect speech acts (ones
coined for the occasion) express the most politeness, and are suitable for
the biggest threats.
In both cases, Brown and Levinson may have collapsed qualitatively
di¤erent dimensions onto a single scale. Rather than having a single
face-threat meter in their heads, and a single politeness meter that tracks
it, people tend to target certain kinds of face threat with certain kinds
of politeness (Holtgraves 2002). For instance, to criticize a friend (which
threatens solidarity), people tend to emphasize sympathetic politeness
(‘‘Hey, let’s go over this paper and see if we can bring it up to your usual
standards’’). But to ask a big favor (which threatens power), people tend
to emphasize deferential politeness, as in the cringing request to borrow
someone’s computer (‘‘I’m terribly sorry to bother you . . .’’).
Also, o¤-record indirect speech—the topic of this paper—didn’t fit into
the scale at all. Politeness Theory deemed it the politest strategy of all,
but people said it was far less polite than deferential politeness (Holt-
graves, 2002). In fact, in some cases it can be downright rude, like Didn’t
I tell you yesterday to pick up your room? or Shouldn’t you tell me who is
coming to the party? One reason is that if the hearer’s competence and
willingness are questioned too blatantly, it suggests that she is inept or
uncooperative. Another is that an indirect request can make the speaker
sound devious and manipulative, and force the listener to do a lot of men-
tal spadework to figure out what he was trying to say.

The evolutionary social psychology of o¤-record indirect speech acts
The fact that indirect speech acts are not so considerate to the hearer af-
ter all brings up another problem. The examples with which we began—
veiled threats, oblique bribes, sexual come-ons—are hardly examples of
a speaker being polite. A merchant listening to an advisory from the local
racketeer on the many accidents that can befall a store surely doesn’t see
it that way. And the cop with his ticket book, or the woman at the eleva-
tor door, sensing the indecent proposal in the innocent question, could be
forgiven for thinking that the propositioner was looking out for his inter-
ests, not theirs (though as we shall see, there can be complicity in those
cases as well).
A final problem for Politeness Theory is the built-in dilemma in its
treatment of o¤-record requests. If an implicature is too much of a trea-
sure hunt, the speaker will have missed an opportunity. The hearer might
have been perfectly happy to comply with his request, if only she knew he
was making one! (In an episode of Seinfeld, George Costanza turned
down his date’s invitation to come up to her apartment for co¤ee, realiz-
ing too late that, in his words, ‘‘ ‘Co¤ee’ doesn’t mean co¤ee . . . . ‘Co¤ee’
means sex!’’). On the other hand, if the implicature is so easy that the
hearer can figure it out without fail, then it should be obvious enough
for any other intelligent person to figure out, too, so it’s not clear why
the request should be perceived as being ‘‘o¤ the record.’’ Who could
claim to be fooled by the line about the etchings, or about settling the
ticket right then and there?
The Cooperative Principle and Politeness Theory are a good start, but
they are incomplete. Like many good-of-the-group theories in social
science, they assume that the speaker and the hearer are working in per-
fect harmony (Pinker 2002). We need to understand what happens when
the interests of a speaker and a hearer are partly in conflict, as they so
often are in real life. And we need to distinguish the kinds of relationships
people have, and how each is negotiated and maintained, rather than
stringing all forms of face threat into a single scale, and doing the same
with all forms of face saving. Finally, we need a deeper analysis of the
enigmatic commodity called ‘‘face,’’ and how it depends on the equally
elusive ‘‘record’’ such that requests can be ‘‘on’’ it or ‘‘o¤ ’’ it.
Plausible deniability as a strategy of conflict
To get some purchase on nebulous concepts like ‘‘providing an out,’’
‘‘plausible deniability,’’ and ‘‘on the record,’’ let’s begin with a scenario
in which their meanings are clear-cut. Consider a perfect Gricean speaker
who says exactly what he means when he says anything at all. Maxim

Steven Pinker
Man is pulled over for running a red light and is pondering whether to
bribe the o‰cer. Since he obeys the maxims of conversation more assidu-
ously than he obeys the laws of tra‰c or the laws of bribery, the only way
he can bribe the o‰cer is by saying, ‘‘If you let me go without a ticket, I’ll
pay you fifty dollars.’’
Unfortunately, he doesn’t know whether the o‰cer is dishonest and
will accept the bribe or is honest and will arrest him for attempting to
bribe an o‰cer. Any scenario like this in which the best course of action
depends on the choices of another actor is in the province of game theory.
In game theory, the conundrum where one actor does not know the
values of the other has been explored by Thomas Schelling (1960: 139–
142), who calls it the Identification Problem. The payo¤s can be summa-
rized like this, where the rows represent the driver’s choices, the columns
represent the di¤erent kinds of o‰cer he might be facing, and the con-
tents of the squares represent what will happen to the driver:
Don’t bribe
Tra‰c ticket
Tra‰c ticket
Go free
Arrest for bribery
The allure of each choice (row) is determined by the sum of the payo¤s of
the two cells in that row weighted by their probabilities. If the driver
doesn’t try to bribe the o‰cer (first row), then it doesn’t matter how hon-
est the o‰cer is; either way the driver gets a ticket. But if he does o¤er the
bribe (second row), the stakes are much higher either way. If Maxim Man
is lucky and is facing a dishonest cop, the cop will accept the bribe and
send him on his way without a ticket. But if he is unlucky and is facing
an honest cop, he will be handcu¤ed, read his rights, and arrested for
bribery. The rational choice between bribing and not bribing will depend
on the size of the tra‰c fine, the proportion of bad and good cops on the
roads, and the penalties for bribery, but neither choice is appealing.
But now consider a di¤erent driver, Implicature Man, who knows how
to implicate an ambiguous bribe, as in ‘‘So maybe the best thing would
be to take care of it here.’’ Suppose he knows that the o‰cer can work
through the implicature and recognize it as an intended bribe, and he
also knows that the o‰cer knows that he couldn’t make a bribery charge
stick in court because the ambiguous wording would prevent a prosecutor
from proving his guilt beyond a reasonable doubt. Implicature Man now
has a third option:

The evolutionary social psychology of o¤-record indirect speech acts
Don’t bribe
Tra‰c ticket
Tra‰c ticket
Go free
Arrest for bribery
Implicated bribe
Go free
Tra‰c ticket
The payo¤s in this new, third row combine the very big advantage of
bribing a dishonest cop with the relatively small penalty of failing to bribe
an honest one. We have explained the evolution of Implicature Man.
Well, almost. We also have to take the point of view of an honest o‰-
cer and the legal system he serves. Why wouldn’t an honest o‰cer arrest
anyone who o¤ered a veiled bribe? If it’s obvious to him, it might be
obvious to a jury, so he has a chance of putting a bad guy behind bars.
To explain why the o‰cer wouldn’t arrest people at the hint of a bribe,
making implicature as dangerous as naked bribery, we must assume two
things, both reasonable. One is that even if all dishonest drivers o¤er
remarks that can be interpreted (correctly) as implicated bribes, some
honest drivers make those remarks too, as innocent observations. So any
arrest might be a false arrest. The second assumption is that an unsuccess-
ful arrest is costly, exposing the o‰cer to a charge of false arrest and the
police department to punitive damages. Then the o‰cer’s decision matrix
would look like this:
Don’t arrest
Tra‰c ticket
Tra‰c ticket
False arrest
(Of course from his point of view a tra‰c ticket is a good thing, not a bad
thing.) The appeal of arresting the driver will depend on the values of the
outcomes in the four cells and on their probabilities. And those probabil-
ities will depend on the proportion of dishonest and honest drivers who
utter the ambiguous remark, that is, on the ratio of the numbers of events
in the left and right columns. If the remark sounds close enough to an
innocuous remark that plenty of honest drivers might make it (or, at
least, enough of them so that a jury could not convict the speaker for
those words beyond a reasonable doubt), then the odds of a successful

Steven Pinker
conviction go down, the odds of a false arrest go up, and the appeal of
the ‘‘Arrest’’ row would be lowered. And that is how Implicature Man
can force the o‰cer’s hand. He can craft his remark so that a dishonest
o‰cer will detect it as an implicated bribe, but an honest o‰cer can’t be
sure (or at least can’t take the chance) that it is one.
A crucial aspect of this analysis is that indirect speech is not an ex-
ample of pure cooperation. Implicature Man is manipulating an honest
o‰cer’s choices to his own advantage and to the o‰cer’s disadvantage.
Though not fully consistent with the Cooperative Principle, it is consis-
tent with the theory by biologists such as Dawkins and Krebs (1978)
that communication in the animal kingdom can often be a form of ma-
nipulation, not just information-sharing.
Plausible deniability in non-legal contexts
A veiled bribe to a police o‰cer is an example in which a person’s words
are on the record and the stakes are tangible, such as tra‰c ticket or an
arrest for bribery. What about everyday life, where o¤ers and requests
can be tendered without fear of legal penalties? In the give-and-take of or-
dinary conversation one might think that we are free to speak our minds,
without worrying that the way a hearer parses our words could land us in
jail. But in fact when it comes to everyday bribes, threats, and o¤ers, our
own emotions make us watch our words as carefully as if we were in legal
jeopardy, and we all turn into Implicature Man.
When would a law-abiding citizen be tempted to o¤er a bribe? Here is
a real-life example. You want to go to the hottest restaurant in town. You
have no reservation. Why not o¤er fifty dollars to the maitre d’ if he will
seat you immediately? This was the assignment given to the writer Bruce
Feiler by Gourmet magazine (Feiler 2000). The results are eye-opening
for any linguist or psychologist interested in the social psychology of indi-
rect speech.
The first result is predictable to most people who imagine themselves in
Feiler’s shoes: the assignment is terrifying. Though no one has ever been
arrested for bribing a maitre d’, Feiler felt like a grievous sinner:
I am nervous, truly nervous. As the taxi bounces southward through he trendier
neighborhoods of Manhattan—Flatiron, the Village, SoHo—I keep imagining
the possible retorts of some incensed maitre d’.
‘‘What kind of establishment do you think this is?’’
‘‘How dare you insult me?’’
‘‘You think you can get in with that?’’