Central Aim
Communication Theory à la Grice
The GM Model
Attempts at Communication and Illocutionary Acts
Attempts at Communication and Gricy Actions
References
Although it is not easy to obtain a full overview of the literature on speech
act theory (SAT) in philosophy of language and linguistics, there are only a
few contributions which belong explicitly to its theoretical foundations. In
particular, the task of reconstructing SAT with the help of communication
theory à la Grice (1957) is still to be further developed. Schiffer
(1972) was the first who tried to differentiate different kinds of speech acts
by using an enriched version of Grice’s theory. His account was picked up and
modified by Holdcroft (1978) and Bach & Harnish (1979); and it is the
central aim of our project to make further progress by imbedding SAT in a
general formalized theory of communicative actions to determine more exactly
the notions and differentations of SAT.
Top
The questions and concept formations of SAT were primarily concerned with
different usages of linguistic expressions – supposing a meaning already given.
That means, SAT is traditionally restricted to actions performed by uttering
words
with a certain
conventional
meaning. Especially in that respect, the intentionalistic communication theory
on which our project is based differs from SAT. According to that account, the
conventional meaning of actions and expressions is to be explicated on the
basis of general notions of communicative actions such as
S attempts to communicate something by doing x
or, in Grice’s terms,
S means
NN
something by x
. These notions neither presuppose the use of words nor the use of other signs
with a conventional meaning because an attempt to communicate can be performed
by acting in a way which is not at all conventionalized.
Unfortunately, Grice’s original account can only be taken as a starting point.
It is, in his own words (1968, 117), merely a "sketch of what might [
¼
] prove to be a viable theory". First of all, it is inadequate because it does
not take into account that communicative actions aim at being understood by the
addressee. Although Schiffer (1972) tried to improve Grice’s ideas with respect
to this constraint, he does not attach enough importance to it. Secondly,
neither Grice nor Schiffer made use of formal methods. This leads to very
complicated natural language formulations which are hard to grasp, let alone to
be examined in detail. Formal aids seem to be unavoidable if one wants to get a
grip on the complex intentions and beliefs involved in the rational
reconstruction of communicative actions.
Top
To render more precisely Grice’s ideas, Kemmerling (1979) and Meggle (1981)
for the first time used elements of a logic of actions. Meggle’s account
provides an idealized model for reconstructing, and improving, Grice’s theory
in a formal way. In this model (the GM model), the notion of doing, or acting,
is treated as a basic concept whose connection with wanting and believing is
determined by meaning postulates such as: if S does something, then S believes
that she does it; D(S,x)
É
B(S,D(S,x)). The notion of believing used in the GM model is the notion of
strong belief introduced by v. Kutschera (1972). It conforms to such principles
as: if S believes that p, then S does not believe that ¬p; B(S,p)
É
¬B(S,¬p). The notion of wanting is as well idealized and subject to
analogous principles (cf. Kutschera 1980). It implies that S only wants what
seems to her to be the best possible, and that S believes that what she wants
can be realized.
The GM model contains a series of explicit definitions showing how the
concepts of Grice’s theory can be gained on the basis of concepts of a general
theory of communicative actions. Furthermore, it precisely represents the
essential connections between these concepts by theorems. An attempt to
communicate an addressee to do something – an action of requesting – is
initially defined as follows:
AC(S,H,x,r) :=
(a) D(S,x) &
(b) W
0
(S,D
1
(H,r)) &
(c) B
0
(S,D(S,x)
®
B
1
(H,I(S,x,D
1
(H,r)))) &
(d) B
0
(S,D
1
(H,r)
«
B
1
(H,I(S,x,D
1
(H,r)))).
S’s doing x (in t) is an attempt to communicate H to do r :=
(a) S does x (in t) &
(b) S wants (in t
0
< t) H to do r (in t
1
> t) &
(c) S believes (in t
0
< t) that, if S will do x (in t), then H will believe (in t
1
> t) that S did x (in t) with the intention to achieve that H will do r (in
t
1
> t) &
(d) S believes (in t
0
< t) that H will do r (in t
1
> t) if and only if H will believe (in t
1
> t) that S did x (in t) with the intention to achieve that H will do r (in
t
1
> t).
The connectives "
®
" and "
«
" are to be understood according to v. Kutschera’s conditional logic (1976).
They go beyond material implication insofar as they require a certain
connection with regard to the contents of their parts. The expression "I(S,x,D
1
(H,r))"– "S does x with the intention to achieve that H will do r" – embedded
in (c) and (d) is defined as follows:
I(S,x,D
1
(H,r)) :=
(a) D(S,x) &
(b) W
0
(S,D
1
(H,r)) &
(c) B
0
(S,D(S,x)
®
D
1
(H,r)) &
(d) B
0
(S,D
1
(H,r)
É
D(S,x))
S does x (in t) with the intention to achieve that H will do r (in t
1
> t) :=
(a) S does x (in t) &
(b) S wants (in t
0
< t) H to do r (in t
1
> t) &
(c) S believes (in t
0
< t) that, if S will do x (in t), then H will do r (in t
1
> t) &
(d) S believes (in t
0
< t) that H will do r (in t
1
> t) only if S will do x (in t)
An action of informing, or convincing, is defined analogously. You just have
to replace "D
1
(H,r)" ("H will do r") by "B
1
(H,p)" ("H will believe that p").
Strictly speaking, the initial definition states only necessary conditions. It
is later refined to get a definition which includes the fact that communication
has to be
absolutely open
. An attempt at communication does not only involve the intention that the
addressee will do r (or believe that p), it also involves the intention that
the addressee will
understand
the attempt at communication. That means, the addressee H is to recognize, and
hence to know (K), that the utterer’s action x is an attempt to communicate H
to do r:
AC(S,H,x,r)
É
I(S,x,K
1
(H,AC(S,H,x,r))).
In the GM model, this
reflexivity condition
implies that every condition C which is necessary for an attempt at
communication is something which the addressee is intended to know:
(AC(S,H,x,r)
É
C)
É
(AC(S,H,x,r)
É
I(S,x,K
1
(H,C))).
In other words, an attempt at communication is aimed at letting the addressee
know that the utterer’s action satisfies every necessary condition for such an
attempt. In the end, this leads to the following recursive definition:
(a) AC
1
(S,H,x,r) := I(S,x,D
1
(H,r))
(b) AC
n+1
(S,H,x,r) := I(S,x,B
1
(H,AC
n
(S,H,x,r))) &
B
0
(S,D
1
(H,r)
«
B
1
(H,AC
n
(S,H,x,r))
(c) AC
*
(S,H,x,r) := (
"
n)(AC
n
(S,H,x,r))
(a) On level 1, S’s doing x (in t) is an attempt to communicate H to do r := S
does x (in t) with the intention to achieve that H will do r (in t
1
> t).
(b) On level n+1, S’s doing x (in t) is an attempt to communicate H to do r :=
S does x (in t) with the intention to achieve that H will believe (in t
1
> t) that, on level n, S’s doing x (in t) is an attempt to communicate H to
do r; and S believes (in t
0
< t) that H will do r (in t
1
> t) if and only if H will believe (in t
1
> t) that, on level n, S’s doing x (in t) is an attempt to communicate H to
do r
(c) S’s doing x (in t) is an attempt to communicate H to do r := For all
levels n, S’s doing x (in t) is an attempt to communicate H to do r on level n.
A
successful
attempt at communication is, roughly, defined as an attempt at communication
which is understood by the addressee and leads to the desired reaction because
it is understood. More precisely:
SAC(S,H,x,r) :=
(a) AC(S,H,x,r) &
(b) D(S,x)
®
K
1
(H,AC(S,H,x,r)) &
(c) D
1
(H,r)
«
K
1
(H,AC(S,H,x,r)).
That conception of success is very strong. According to it, successful
communication does not only involve understanding but also the realization of
the intended aim: S’s attempt to communicate H to do r is successful only if H
does r; it does not suffice that H merely understands what she is to do.
However, it is easy to gain a weaker notion of successful communication from
the definition above by dispensing with condition (c). In the weak sense, an
attempt to communicate a certain content is successful if and only if, as a
result of the speaker’s action, the addressee recognizes, and hence knows, that
it is an attempt to communicate that content.
Top
By embedding SAT in the GM model, we hope to render more precisely central
notions of SAT, among them the notions of success and happiness and the notion
of understanding. Furthermore, we hope to get a clear account for delimiting
illocutions from both locutions and perlocutions, and general principles for
classifying illocutionary acts. Apart from these ambitious aims, the GM model
provides formal means by which the proposals of Schiffer (1972), Holdcroft
(1978), Bach & Harnish (1979) and Searle (1983) can be formulated exact
enough to compare them in detail.
To give you an impression of the special ideas and problems we are concerned
with, we present you two examples. The first relates to the question of how
strong the connection is between illocutionary acts and attempts at
communication. The second concerns Kemmerling’s proposal for the essential
characteristic of illocutionary acts. (As you will notice, these questions are
closely connected.)
Top
Attempts at communication, as defined in the GM model, are obviously
illocutionary acts (in the wide sense which does not imply producing linguistic
signs). Now, Meggle’s (1997) and Ulkan’s (1882) idea is to
identify
attempts at communication and illocutionary acts, that is, to claim also, the
other way round, that every illocutionary act is an attempt at communication,
and that illocutionary goals are nothing else than communicative goals. Since
communicative goals, such as producing a belief in the addressee, are
perlocutionary goals, illocutionary goals are only a subclass of perlocutionary
goals.
If illocutionary goals are perlocutionary goals, the decisive question is:
What makes the difference between an illocutionary goal and other, merely,
perlocutionary goals? Ulkan’s (1992) answer is: A goal is illocutionary if and
only if one must intend it to perform an attempt at communication. If having a
certain goal is not necessary for a communicative act, then this goal is merely
perlocutionary. Thus, understanding is a general illocutionary goal because one
cannot attempt to communicate something without having the intention that the
addressee understands the attempt. Performing an illocutionary act means,
roughly, to do something with the intention to achieve a certain perlocutionary
effect by achieving the illocutionary effect of understanding. Asking a
question, for example, means to do something with the intention of getting an
answer because the addressee understands that one wants to ask a certain
question.
That account faces some difficulties. First of all, it implies that
illocutionary acts are essentially intentional actions because they involve,
e.g, the intention of letting someone know that one tries to perform a certain
illocutionary act. But what about a person who, prima facie, warns someone by
erecting a sign without caring if anybody will be warned? It seems to be
possible to issue a warning with the sole intention to do what one’s superior
expects from one. However, these cases are not as clear as they seem to be at
first glance. Is it really the person erecting the sign who warns? One can
argue that she is merely the executive body of her superior. It is the superior
who warns, not the person who is instructed to erect the sign. And the
superior, presumably,
has
the intention to warn.
Secondly, as Ulkan (1992, 42) herself recognizes, there is the problem of
catching Austin’s paradigms of illocutionary acts, such as baptizing.
Apparently, baptizing is not communicative because it does not require
addressing one’s utterance to a certain audience (cf. Bach & Harnish 1979,
117; Strawson 1964, 456). One can baptize a child, so it seems, even if one
knows that there is nobody around who might recognize what one is doing. We say
"so it seems" because we are not sure whether baptizing does not, at least,
involve the intention to communicate with God. But even if such illocutionary
acts are in fact not essentially communicative, it is to be hoped that the GM
model helps us to explain the other, communicative, illocutions.
Besides, Bach & Harnish (1979, 116) point to a certain commonality between
communicative and conventional illocutionary acts. In the case of communicative
illocutionary acts, one tries to achieve a goal (e.g. getting an answer) by
doing something of which one thinks that it will bring about the goal because
it helps the addressee to recognize one’s goal. In the case of conventional
illocutionary acts, one tries to achieve a goal (e.g. making a child a
Christian) by doing something of which one thinks that it will bring about the
goal because it conforms to a convention. If this intentionalistic view is
right, illocutionary acts have in common that they involve the intention to
achieve a goal and the belief that the action will bring about the goal because
it has a certain property connected with understanding or conventions. And
though the distinctive properties differ, it is not unreasonable to suppose
that an analysis of communicative illocutionary acts in terms of the GM model
will be helpful in analysing conventional illocutionary acts because they
follow a common pattern.
Thirdly, attempts at communication, as defined in the GM model, involve only
two kinds of perlocutionary goals: actions or beliefs. That raises the question
whether there are illocutionary acts which are performed with the intention to
achieve other kinds of perlocutionary goals. What about Schiffer’s (1972)
example of a man who does the washing-up with the open intention to please his
wife? That would be no problem if one could show that eliciting certain
feelings in his wife is not his primary intention because he wants to produce
that effect by inducing, e.g., the belief in her that he wants to please her.
However, even if there remain illocutionary acts with other types of goals,
that does not mean that the GM model is inappropriate, but only that is has to
be enlarged. It is not difficult to modify the given definitions such that they
cover also attempts at getting other kinds of reactions, apart from action or
belief.
Top
A central problem in SAT is the distinction between locutionary, illocutionary
and perlocutionary acts. Which characteristics make sure that referring to
Bertrand is merely a locutionary act, whereas asking whether Bertrand’s Party
takes place tomorrow belongs to the illocutions? And what makes bringing about
a startled face by this question a perlocutionary act?
Andreas Kemmerling (1997, 1999) tries to answer these questions in an original
way because his account does not presuppose that certain intentions or
conventions are necessary for performing an illocutionary act. In Kemmerling’s
view, the hallmark of illocutions is a certain feature called "griciness" which
is, roughly, defined as follows: An action type g is gricy if and only if
making it clear that one intends to perform an action of type g in doing what
one does is sufficient for therein performing an action of type g. For asking a
question, e.g, it suffices that you make it clear that you intend your action
to be of the type
asking-a-question
. For frightening a person, on the other hand, it is not enough to make it
clear that this is what you want to do by acting in a certain way.
This idea is very interesting from our point of view because it can be shown
that attempts at communication, as defined in the GM model, have a feature
which comes close to griciness. Kemmerling (1999) emphasizes that "to make it
clear" is to be understood in a factive way: S makes it clear that she has a
certain intention only if S really has that intention. Thus, however making
something clear ought to be reconstructed in the GM model, if doing something
with the intention to perform, by that action, an attempt at communication is
sufficient for performing such an attempt, then attempts at communication are
gricy.
Now, it is not difficult to derive that attempts at communication have that
feature. This results from the following theorems:
I(S,f,AC(S,f))
É
(D(S,f) & B(S,AC(S,f)))
If S does f with the intention to perform, by doing f, an attempt at
communication, then S does f, and S believes that she performs, by doing f, an
attempt at communication.
(D(S,f) & B(S,AC(S,f)))
É
AC(S,f)
If S does f and believes that, by doing f, she performs an attempt at
communication, then S performs an attempt at communication.
Presumably, the second theorem is more interesting. Why does it hold?
Believing that, by doing f, one performs an attempt at communication means to
believe that one’s action f satisfies every condition for such an attempt.
Since these conditions involve, apart from doing f, only beliefs and wants of
the person herself, and since a rational being does not go wrong about her own
beliefs and wants, these conditions are automatically satisfied. In other
words, believing that one does f because of certain beliefs and wants is, in
the case of rational beings, sufficient for doing f because of these
propositional attitudes – given that one knows what one does, i.e. that one’s
action really is an action of type f. Hence, if S does f and believes that her
action satisfies the conditions for an attempt at communication, then it is
such an attempt because these conditions involve no
external
states of affairs with respect to which S might go wrong.
So, by reconstructing griciness within the GM model, we do not only get a clear
definition
of griciness (or something like griciness) but also an
explanation
for the fact that attempts at communication are gricy. They are gricy because
it is very easy to carry them out: you just have to do something with the
intention to perform, and therefore the belief that you perform, an attempt at
communication. There are no external conditions to be satisfied, conditions,
like conforming to a certain convention or having a certain authority, which
are such that you might be mistaken about their satisfaction.
Top
Bach, K. & Harnish, R.M. 1979:
Linguistic Communication and Speech Acts
, Cambridge/M.
Eemeren, F.H. v. & Grootendorst, R. 1983:
Speech Acts in Argumentative Discussions
, Dordrecht.
Grice, H.P. 1957: Meaning, in:
Studies in the Way of Words
, Cambridge/M. 1989, 213–223; or. publ. in:
The Philosophical Review
66, 377–388.
Grice, H.P. 1968: Utterer’s Meaning, Sentence-Meaning, and Word-Meaning, in:
Studies in the Way of Words
, Cambridge/M. 1989, 117–137; or. publ. in:
Foundations of Language
4, 1–18.
Holdcroft, D. 1978:
Words and Deeds. Problems in the Theory of Speech Acts
, Oxford.
Kemmerling, A. 1979: Was Grice mit "meinen" meint, in:
Sprechakttheorie und Semantik
, ed. by G. Grewendorf, Frankfurt/M., 67–118.
Kemmerling, A. 1997: Der bedeutungstheoretisch springende Punkt sprachlicher
Verständigung, in:
Kommunikationsversuche. Theorien der Kommunikation
, ed. by G.-L. Lueken, Leipzig, 60–106.
Kemmerling, A. 1999: Gricy Actions, to be publ. in:
Paul Grice’s Heritage
(working title), ed. by G. Cosenza.
Kutschera, F. v. 1976:
Einführung in die intensionale Semantik
, Berlin & New York.
Kutschera, F. v. 1980: Grundbegriffe der Handlungslogik, in:
Handlungstheorie – interdisziplinär I
, ed. by H. Lenk, Munich.
Meggle, G. 1981:
Grundbegriffe der Kommunikation
, Berlin & New York; 2nd ed. 1997.
Meggle, G. 1997: Theorien der Kommunikation – Eine Einführung, in:
Kommunikationsversuche. Theorien der Kommunikation
, ed. by G.-L. Lueken, Leipzig, 14–40.
Rolf, E. 1983:
Sprachliche Informationshandlungen
, Göppingen.
Schiffer, S. 1972:
Meaning
, Oxford.
Searle, J. 1969:
Speech Acts
, Cambridge.
Searle, J. 1983:
Intentionality. An Essay in the Philosophy of Mind
, Cambridge.
Strawson, P.F. 1964: Intention and Convention in Speech Acts, in:
The Philosophical Review
73, 439–460.
Ulkan, M. 1992:
Zur Klassifikation von Sprechakten
, Tübingen.
|
In Deutsch nicht verfügbar
