Chapter 9: Pragmatics and Communication

From: Harnad, Stevan (harnad@cogsci.soton.ac.uk)
Date: Fri May 09 1997 - 21:43:30 BST


Chapter 9: Pragmatics and Communication

Pragmatics is the study of "speaker meaning" (or "utterance meaning)"
rather than sentence meaning. As such, it is the study of how speakers
and hearers use CONTEXT.

"All men are mortal"

can be understood if you know what "men" means [people],
and what "mortal" means [do not live forever].

But to understand

"All you ever do is complain"

you need to know more than what the sentence says: You have to know
who "you" is and you have to know that the sentence doesn't LITERALLY
mean that all "you" ever do is complain.

So you need to know the context in which the sentence was spoken --
what was said before, perhaps what was said after -- and also when and
where it was spoken, and by whom is was spoken. You also need to know
what they had in mind in saying that, and what their relationship was
with whoever "you" is. (The same remark can be understood as a gentle
joke or an aggressive barb, depending on who is saying it to whom,
where, and when.)

This is the kind of "mutual knowledge" that human communication depends
on; language itself cannot do all the work. When two people are having a
conversation, each one needs to know what the other one knows, and that
the other one knows that they know, etc.

Not that it's always possible to know what the other one knows you know,
etc.:

You may have heard of the "prisoners' dilemma": Two people are caught
for committing a big crime (murder), and they really are guilty, but
the court only has evidence that they committed a small part of the big
crime (say, breaking and entering, although they did commit the
murder).

The only way they can be convicted for murder is if they confess.
The men know this, but they are separated, and each gets the following
offer:

"We only have evidence to convict you of breaking and entering, for
which the punishment is 3 years in prison. But if you cooperate and
confess that you both did it, and your partner does not confess, then
you will only have to serve 1 year, whereas you partner will get a life
sentence. But we are offering the same deal to your partner, so if he
too confesses, then you will both be imprisoned for 10 years"

So if neither prisoner rats on the other, they each serve just 3 years.
If one rats and the other doesn't, the ratter gets only one year and the
other life. But if both rat, then both get 10 years.

So what do they do? If you put yourself in the position of one of these
prisoners you say to yourself:

"If neither of us rat, we both get three years and that's not so bad,
so I won't rat.

Hang on. Doing only 1 year is even better, and if I rat, I get 1 year
and he gets life. Sounds good. So I'll rat on him.

Hang on. He must be thinking the same thing I am, so he'll rat
on me too, and we'll both get 10 years, which is bad. So I won't rat
on him.

Hang on, he knows that I know this, so that rat will rat on me and I'll
get life and he'll get off with just one year! So I'd better rat on him.
At least that way I'm sure I'll only get 10 years instead of life.

Although the prisoners' dilemma is a situation in which the prisoners
DON'T get to to communicate directly, it illustrates how many levels of
"mutual knowledge" can be involved.

"Children's Theory of Mind"

To have this kind of mutual knowledge about what each speaker has in
mind it is necessary to know that the other speaker has a mind in the
first place. We've talked about the "other minds" problem in this
course, so we all know that we all have minds, and we have a fairly good
idea -- good enough for sensible conversation at least -- of what each
of us is likely to know, expect, want, etc. And we each no that each of
us knows this about us too.

Do all people know that others have minds? Do children? If so, from what
age do they know it, and what is the evidence?

Research on children's' theory of mind actually began with an experiment
by Premack and Woodruff that was published with the title:
"Does the Chimpanzee Have a Theory of Mind?"

This was a bad title, in a way, because "theory of mind" is actually a
branch of philosophy that studies the mind. The title should have been
"Do Chimpanzees know THAT other chimpanzees know and WHAT other
chimpanzees know.

The puppet experiment in Chapter 8 is the kind of experiment that is
used with both chimps and children: For example, a child sees someone
putting an object in one of two boxes. Then the person leaves the room
and another person comes and moves the object to the other box. Then
the first person comes back and the child has to predict which box he
will look in to retrieve the object.

The exact age boundaries are not pinned down yet, up to about 3 years
old, the child thinks the person will look in the box new box,
not the old one. The child is not aware of what the person could or
couldn't know in this situation. The child does not yet have a theory
of mind. Children a little older can give the right answer in this
experiment, but they have to be still older to succeed in an experiment
that is based not just on what a second person would or wouldn't know,
but also what a second person would or wouldn't know about what the
first person, or a third person, would or wouldn't know.

So the mutual knowledge needed to carry on a coherent conversation
requires a lot of additional knowledge of this sort.

Sperber & Wilson's theory of relevance is meant to be
replace Grice's "maxims" about what is needed for
sensible conversations.

Grice had the maxims of

Quantity (say no more than needed to get your information across)

Quality (make sure the message is true)

Relation (make sure your message is relevant)

Manner (make it simple and understandable)

[This might have been used as a definition of "kid-sib" explanations.]

Sperber and Wilson replaced Quantity and Relation by
their own priciples of relevance:

In a given context, information should be relevant.
For them, relevance is the outcome of two opposing variables: To
decide which of two messages is more relevant to a hearer in a specific
context, the more relevant one is the one that IMPLIES more.

What does "imply more" mean?

Well in the context of a Tuesday morning on which you had been thinking
about going out for a picnic, the message that it will rain all day has
more implications than the message that it is Tuesday. Raining all day
implies that you will have to postpone your picnic. On the other hand,
the message that the comet Hale-Bop is headed for earth and will blow
it up in a few hours implies not only that you should postpone your
picnic, but that you might better think of it as canceling your picnic,
and any other plans you might have had.

According to Sperber and Wilson, relevance does not depend only on which
message has the largest number of implications; the number of
implications has to be weighed against its cost in processing
effort.

If the rain message had been sent in as a number of relative humidity
readings that would have to be compared with six months' worth of
weather reports to figure out how likely it is that there will be rain
Tuesday compared to other days with similar humidity readings, you'd
probably decide to forget about the calculations and just go on the
picnic and hope for the best.

But if the comet message came with some code that requires you to
decode a lot of data fast in order to fire off a bomb in the right
direction and soon enough to make Hale-Bop explode safely before it got
too near to earth, you'd probably want to make the effort.

If there is a criticism to be made about:

RELEVANCE = Implications/Effort

it is that it is not obvious that what we usually want is the largest
number of implications. What we want is the largest number of RELEVANT
implications. This suggests that the Relevance Principle may be
circular.

What is language for?

Information = the reduction of uncertainty between alternatives
that MATTER to you.

The 6-choice lunch machine is an example of how any data that could
reduce the uncertainty about which of the 6 windows contains the lunch
is informative.

According to "Speech Act" theory, sentences should not just be thought
of as declarative propositions (Sentences that say so-and-so
is true). A declarative is just one kind of speech act. Others are
imperative (telling someone what to do, rather than what is or is not
true). Questions are another kind, and so on.

But making declarative statements is certainly a large part of what
language is used for. For example, think of things that you
can learn by trial and error experience: There's my favorite example of
mushroom picking. If you had to eat mushrooms to survive because there
was nothing else to eat, nibbling a little and waiting to see whether or
not you got sick would definitely be the hard way of getting your
knowledge about mushrooms. Let's call that way of getting your knowledge
-- by trial and error experience, guided by feedback from the
consequences of what you do -- "honest toil."

Language makes it possible for you to get the same information by a kind
of "theft": If someone who knows the features of the edible and
poisonous mushrooms describes them to you in words, then you have saved
yourself a lot of toil. Nor would the speaker lose anything for having
taught you the features -- unless of course mushrooms were scarce.

But in general, when the information you are given through language is
not about scarce resources for which the speaker and hearer are
competing, linguistic "theft" is a victimless crime. It's more like
barter, in which I tell you the things I know and you don't, and
vice-versa. But it's all about minimising unnecessary toil.

Pragmatic deficits

Given how important mutual information and "theory of mind" is for
communication with language, what would happen if someone COULDN'T
tell what was going on in other people's mind, or didn't even know that
ANYTHING was going on in there?

Children under three are just learning language and from an early age,
children's theory-of-mind ability just about keeps up with their
language ability. Could language have continued to develop in a child
even if the theory-of-mind ability did not?

Some people think that the root, or at least a feature of the disorder
called autism might be a theory-of-mind deficit. Many autistic children
can speak and understand. For example, there is no overall problem with
declarative sentences that do not depend on context.

But, starting from eye-aversion (not looking at the person speaking to
you), there are some problems in communicating with autistic children.
They often speak as if they were the only one in people-free world.

The theory that autism is a theory-of-mind deficit has had some support,
and explains some of the features of autism, but there are
counterexamples. Some highly intelligent autistic people go to
university, and they can even TELL us that the "theory-of-mind"
theory is wrong.

Good books to read on this are:

Nobody, Nowhere by Donna Williams.

and

An Anthropologist from Mars, by Oliver Sacks



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:52 GMT