Cooperation and implicature

In much of the preceding discussion, we have assumed that speakers and listeners involved in
conversation are generally cooperating with each other. For example, for reference to be successful,
it was proposed that collaboration was a necessary factor. In accepting speakers' presuppositions,
listeners normally have to assume that a speaker who says 'my car' really does have the car that is
mentioned and isn't trying to mislead the listener. This sense of cooperation is simply one in which
people having a conversation are not normally assumed to be trying to confuse, trick, or withhold
relevant information from each other. In most circumstances, this kind of cooperation is only the
starting point for making sense of what is said.
In the middle of their lunch hour, one woman asks another how she likes the hamburger she is
eating, and receives the answer in [1].
[1] A hamburger is a hamburger.
From a purely logical perspective, the reply in [1] seems to have no communicative value since it
expresses something completely obvious. The example in [1] and other apparently pointless
expressions like 'business is business' or 'boys will be boys', are called tautologies. If they are used
in a conversation, clearly the speaker intends to communicate more than is said.
When the listener hears the expression in [1], she first has to assume that the speaker is being
cooperative and intends to communicate something. That something must be more than just what
the words mean. It is an additional conveyed meaning, called an implicature. By stating [1], the
speaker expects that the listener will be able to work out, on the basis of what is already known, the
implicature intended in this context.
Given the opportunity to evaluate the hamburger, the speaker of [i] has responded without an
evaluation, thus one implicature is that she has no opinion, either good or bad, to express.
Depending on other aspects of the context, additional implicatures (for example, the speaker thinks
all hamburgers are the same) might be inferred.
Implicatures are primary examples of more being communicated than is said, but in order for
them to be interpreted, some basic cooperative principle must first be assumed to be in operation.
The cooperative principle
Consider the following scenario. There is a woman sitting on a park bench and a large dog lying on
the ground in front of the bench. A man comes along and sits down on the bench.
[2] Man: Does your dog bite?
Woman: No.
(The man reaches down to pet the dog. The dog bites the man's hand)
Man: Ouch! Hey! You said your dog doesn't bite.
Woman: He doesn't. But that's not my dog.
One of the problems in this scenario has to do with communication. Specifically, it seems to be a
problem caused by the man's assumption that more was communicated than was said. It isn't a
problem with presupposition because the assumption in 'your dog' (i.e. the woman has a dog) is true
for both speakers. The problem is the man's assumption that his question 'Does your dog bite?' and
the woman's answer 'No' both apply to the dog in front of them. From the man's perspective, the
woman's answer provides less information than expected. In other words, she might be expected to
provide the information stated in the last line. Of course, if she had mentioned this information
earlier, the story wouldn't be as funny. For the event to be funny, the woman has to give less
information than is expected.
The concept of there being an expected amount of information provided in conversation is just
one aspect of the more general idea that people involved in a conversation will cooperate with
each other. (Of course, the woman in [2] may actually be indicating that she does not want to take
part in any cooperative interaction with the stranger.) In most circumstances, the assumption of
cooperation is so pervasive that it can be stated as a cooperative principle of conversation and
elaborated in four sub-principles, called maxims, as shown in Table 5.1.
The cooperative principle: Make your conversational contribution such as is required, at the stage at which
it occurs, by the accepted purpose or direction of the talk exchange in which you I are engaged.
The maxims
Quantity
1. Make your contribution as informative as is required (for the current purposes of the exchange).
2. Do not make your contribution more informative than is required.
Quality: Try to make your contribution one that is true.
1. Do not say what you believe to be false.
2. Do not say that for which you lack adequate evidence.
Relation: Be relevant.
Manner: Be perspicuous.
1. Avoid obscurity of expression.
2. Avoid ambiguity.
3. Be brief (avoid unnecessary prolixity).
4. Be orderly.
Table 5.1 The cooperative principle (following Grice 1975)
It is important to recognize these maxims as unstated assumptions we have in conversations. We
assume that people are normally going to provide an appropriate amount of information (unlike the
woman in [2]); we assume that they are telling the I truth, being relevant, and trying to be as clear as
they can. Because these principles are assumed in normal interaction, speakers rarely mention them.
However, there are certain kinds of expressions speakers use to mark that they may be in danger
of not fully adhering to the principles. These kinds of expressions are called hedges.
Hedges
The importance of the maxim of quality for cooperative interaction in English may be best
measured by the number of expressions we use to indicate that what we're saying may not be totally
accurate. The initial phrases in [3a.-c] and the final phrase in [3d.] are notes to the listener regarding
the accuracy of the main statement.
[3] a. As far as I know, they're married.
b. I may be mistaken, but I thought I saw a wedding ring on her finger.
c. I'm not sure if this is right, but I heard it was a secret ceremony in Hawaii.
d. He couldn't live without her, I guess.
The conversational context for the examples in [3] might be a recent rumor involving a couple
known to the speakers. Cautious notes, or hedges, of this type can also be used to show that the
speaker is conscious of the quantity maxim, as in the initial phrases in [4a.-c], produced in the
course of a speaker's account of her recent vacation.
[4] a. As you probably know, I am terrified of bugs.
b. So, to cut a long story short, we grabbed our stuff and ran.
c. I won't bore you with all the details, but it was an exciting trip.
Markers tied to the expectation of relevance (from the maxim of relation) can be found in the
middle of speakers' talk when they I say things like 'Oh, by the way' and go on to mention some
potentially unconnected information during a conversation. Speakers also seem to use expressions
like 'anyway', or 'well, anyway', to indicate that they may have drifted into a discussion of some
possibly non-relevant material and want to stop. Some expressions which may act as hedges on the
expectation of relevance are shown as the initial phrases in [5a.-c.], from an office meeting.
[5] a. I don't know if this is important, but some of the files are missing.
b. This may sound like a dumb question, but whose hand writing is this?
c. Not to change the subject, but is this related to the budget?
The awareness of the expectations of manner may also lead speakers to produce hedges of the type
shown in the initial phrases in [6a.-c.], heard during an account of a crash.
[6] a. This may be a bit confused, but I remember being in a car.
b. I'm not sure if this makes sense, but the car had no lights.
c. I don't know if this is clear at all, but I think the other car was reversing.
All of these examples of hedges are good indications that the speakers are not only aware of the
maxims, but that they want to show that they are trying to observe them. Perhaps such forms also
communicate the speakers' concern that their listeners judge them to be cooperative conversational
partners.
There are, however, some circumstances where speakers may not follow the expectations of the
cooperative principle. In courtrooms and classrooms, witnesses and students are often called upon
to tell people things which are already well-known to those people (thereby violating the quantity
maxim). Such specialized institutional talk is clearly different from conversation.
However, even in conversation, a speaker may 'opt out' of the maxim expectations by using
expressions like 'No comment' or 'My lips are sealed' in response to a question. An interesting
aspect of such expressions is that, although they are typically not 'as informative as is required' in
the context, they are naturally interpreted as communicating more than is said (i.e. the speaker
knows the answer). This typical reaction (i.e. there must be something special' here) of listeners to
any apparent violation of the maxims is actually the key to the notion of conversational implicature.
Conversational implicature
The basic assumption in conversation is that, unless otherwise indicated, the participants are
adhering to the cooperative principle and the maxims. In example [7], Dexter may appear to be
violating the requirements of the quantity maxim.
[7] Charlene: I hope you brought the bread and the cheese.
Dexter: Ah, I brought the bread.
After hearing Dexter's response in [7], Charlene has to assume that Dexter is cooperating and not
totally unaware of the quantity maxim. But he didn't mention the cheese. If he had brought the
cheese, he would say so, because he would be adhering to the quantity maxim. He must intend that
she infer that what is not mentioned was not brought. In this case, Dexter has conveyed more than
he said via a conversational implicature.
We can represent the structure of what was said, with b (= bread) and c (= cheese) as in [8]. Using
the symbol +> for an implicature, we can also represent the additional conveyed meaning.
[8] Charlene: b &c?
Dexter: b (+>NOTc)
It is important to note that it is speakers who communicate meaning via implicatures and it is
listeners who recognize those communicated meanings via inference. The inferences selected are
those which will preserve the assumption of cooperation.
Generalized conversational implicatures
In the case of example [7], particularly as represented in [8], no special background knowledge of
the context of utterance is required in order to make the necessary inferences. The same process of
calculating the implicature will take place if Doobie asks Mary about inviting her friends Bella
(= b) and Cathy (= c) to a party, as in [9a.], and gets the reply in [9b.]. The context is different from
[7], but the general process of identifying the implicature is the same as in [8].
[9] a. Doobie: Did you invite Bella and Cathy? (b & c?)
b. Mary: I invited Bella. (b +>NOT c)
 When no special knowledge is required in the context to calculate the additional conveyed
meaning, as in [7] to [9], it is called a generalized conversational implicature. One common
example in English involves any phrase with an indefinite article of the type 'a/an X', such as 'a
garden' and 'a child' as in [10]. These phrases are typically interpreted according to the generalized
conversational implicature that: an X +> not speaker's X.
[10] I was sitting in a garden one day. A child looked over the fence.
The implicatures in [10], that the garden and the child mentioned are not the speaker's, are
calculated on the principle that if the speaker was capable of being more specific (i.e. more
informative, following the quantity maxim), then he or she would have said 'my garden' and 'my
child'.
A number of other generalized conversational implicatures are commonly communicated on the
basis of a scale of values and are consequently known as scalar implicatures.
Scalar implicatures
Certain information is always communicated by choosing a word which expresses one value from a
scale of values. This is particularly obvious in terms for expressing quantity, as shown in the scales
in [11], where terms are listed from the highest to the lowest value.
[11] < all, most, many, some, few>
< always, often, sometimes>
When producing an utterance, a speaker selects the word from the scale which is the most
informative and truthful (quantity and quality) in the circumstances, as in [12].
[12] I'm studying linguistics and I've completed some of the required courses.
By choosing 'some' in [12], the speaker creates an implicature (+> not all). This is one scalar
implicature of uttering [12]. The basis of scalar implicature is that, when any form in a scale is
asserted, the negative of all forms higher on the scale is implicated. The first scale in [11] had 'all',
'most', and 'many', higher than 'some'. Given the definition of scalar implicature, it should follow
that, in saying 'some of the required courses', the speaker also creates other implicatures (for
example, +> not most, +> not many).
If the speaker goes on to describe those linguistics courses as in [13], then we can identify some
more scalar implicatures.
[13] They're sometimes really interesting.
By using 'sometimes' in [13], the speaker communicates, via implicature, the negative of forms
higher on the scale of frequency (+> not always, +> not often).
There are many scalar implicatures produced by the use of expressions that we may not
immediately consider to be part of any scale. For example, the utterance of [14a.] will be interpreted
as implicating '+> not certain' as a higher value on the scale of 'likelihood' and [14b.] '+> not must'
on a scale of 'obligation' and '+> not frozen' on a scale of 'coldness'.
[14] a. It's possible that they were delayed,
b. This should be stored in a cool place.
One noticeable feature of scalar implicatures is that when speakers correct themselves on some
detail, as in [15], they typically cancel one of the scalar implicatures.
[15] I got some of this jewelry in Hong Kong—um actually I think I got most of it there.
In [15], the speaker initially implicates '+> not most' by saying 'some', but then corrects herself by
actually asserting 'most'. That final assertion is still likely to be interpreted, however, with a scalar
implicature (+> not all).
Particularized conversational implicatures
In the preceding examples, the implicatures have been calculated without special knowledge of any
particular context. However, most of the time, our conversations take place in very specific contexts
in which locally recognized inferences are assumed. Such inferences are required to work out the
conveyed meanings which result from particularized conversational implicatures. As an illustration,
consider example [16], where Tom's response does not appear on the surface to adhere to relevance.
(A simply relevant answer would be 'Yes' or 'No'.)
[16] Rick: Hey, coming to the wild party tonight ?
Tom: My parents are visiting.
In order to make Tom's response relevant, Rick has to draw on some assumed knowledge that one
college student in this setting expects another to have. Tom will be spending that evening with his
parents, and time spent with parents is quiet (consequently +> Tom not at party).
Because they are by far the most common, particularized conversational implicatures are typically
just called implicatures. A further example, in which the speaker appears not to adhere to (i.e. to
'flout') the maxim of manner, is presented in [17].
[17] Ann: Where are you going with the dog?
Sam: To the V-E-T.
In the local context of these speakers, the dog is known to recognize the word 'vet', and to hate
being taken there, so Sam produces a more elaborate, spelled out (i.e. less brief) version of his
message, implicating that he doesn't want the dog to know the answer to the question just asked.
In [18], Leila has just walked into Mary's office and noticed all the work on her desk. Mary's
response seems to flout the maxim of relevance.
[18] Leila: Whoa! Has your boss gone crazy?
Mary: Let's go get some coffee.
In order to preserve the assumption of cooperation, Leila will have to infer some local reason (for
example, the boss may be nearby) why Mary makes an apparently non-relevant remark. The
implicature here is essentially that Mary cannot answer the question in that context.
In addition to these fairly prosaic examples of implicatures, there are other more entertaining
examples, as in [19] and [20], where the responses initially appear to flout relevance.
[19] Bert: Do you like ice-cream?
Ernie: Is the Pope Catholic?
[20] Bert: Do vegetarians eat hamburgers?
Ernie: Do chickens have lips?
In [19], Ernie's response does not provide a 'yes' or 'no' answer. Bert must assume that Ernie is
being cooperative, so he considers Ernie's 'Pope' question and clearly the answer is 'Yes'. So, the
answer is known, but the nature of Ernie's response also implicates that the answer to the question
was 'Obviously, yes!'. An additional conveyed meaning in such a case is that, because the answer
was so obvious, the question did not need to be asked. Example [20] provides the same type of
inferencing with an answer 'Of course not!' as part of the implicature.
Properties of conversational implicatures
So far, all the implicatures we have considered have been situated within conversation, with the
inferences being made by people who hear the utterances and attempt to maintain the assumption of
cooperative interaction. Because these implicatures are part of what is communicated and not said,
speakers can always deny that they intended to communicate such meanings. Conversational
implicatures are deniable. They can be explicitly denied (or alternatively, reinforced) in different
ways. To take a simple example, there is a standard implicature associated with stating a number,
that the speaker means only that number, as shown in [21].
[21] You have won five dollars! (+> ONLY five)
As shown in [22], however, it is quite easy for a speaker to suspend the implicature (+> only) using
the expression 'at least' [22a.], or to cancel the implicature by adding further information, often
following the expression 'in fact' [22b.], or to reinforce the implicature with additional information,
as in [22c.].
[22] a. You've won at least five dollars!
 b. You've won five dollars, in fact, you've won ten!
c. You've won five dollars, that's four more than one!
We have already noted with many of the previous examples that implicatures can be calculated by
the listeners via inference. In terms of their defining properties, then, conversational implicatures
can be calculated, suspended, cancelled, and reinforced. None of these properties apply to
conventional implicatures.
Conventional implicatures
In contrast to all the conversational implicatures discussed so far, conventional implicatures are
not based on the cooperative principle or the maxims. They don't have to occur in conversation, and
they don't depend on special contexts for their interpretation. Not unlike lexical presuppositions,
conventional implicatures are associated with specific words and result in additional conveyed
meanings when those words are used. The English conjunction 'but' is one of these words. The
interpretation of any utterance of the type p but q will be based on the conjunction p & q plus an
implicature of 'contrast' between the information in p and the information in q. In [23], the fact that
'Mary suggested black' (= p) is contrasted, via the conventional implicature of 'but', with my
choosing white (= q).
[23] a. Mary suggested black, but I chose white,
b. p & q (+>p is in contrast to q)
Other English words such as 'even' and 'yet' also have conventional implicatures. When 'even' is
included in any sentence describing an event, there is an implicature of 'contrary to expectation'.
Thus, in [24] there are two events reported (i.e. John's coming and John's helping) with the
conventional implicature of 'even' adding a 'contrary to expectation' interpretation of those events.
[24] a. Even John came to the party.
b. He even helped tidy up afterwards.
The conventional implicature of 'yet' is that the present situation is expected to be different, or
perhaps the opposite, at a later time. In uttering the statement in [25a.], the speaker produces an
implicature that she expects the statement 'Dennis is here' (= p) to be true later, as indicated in
[25b.].
[25] a. Dennis isn't here yet. (=NOT p)
b. NOT p is true (+> p expected to be true later)
It may be possible to treat the so-called different 'meanings' of 'and' in English as instances of
conventional implicature in different structures. When two statements containing static information
are joined by 'and', as in [26a.], the implicature is simply 'in addition' or 'plus'. When the two
statements contain dynamic, action-related information, as in [26b.], the implicature of 'and' is 'and
then' indicating sequence.
[26] a. Yesterday, Mary was happy
and ready to work. (p & q, +> p plus q)
b. She put on her clothes and left
the house. (p & q, +> q after p)
Because of the different implicatures, the two parts of [26a.] can be reversed with little difference in
meaning, but there is a big change in meaning if the two parts of [26b.] are reversed.
For many linguists, the notion of 'implicature' is one of the central concepts in pragmatics. An
implicature is certainly a prime example of more being communicated than is said. For those same
linguists, another central concept in pragmatics is the observation that utterances perform actions,
generally known as 'speech acts'.