Content uploaded by Jens Allwood
All content in this area was uploaded by Jens Allwood on Mar 05, 2013
Content may be subject to copyright.
In Dahl, (Ed.) Logic, Pragmatics and Grammar. Lund. Studentlitteatur, pp. 53-99.
A CRITICAL LOOK AT SPEECH ACT THEORY
Dept of Linguistics, Göteborg University
One of the most powerful theoretical conceptions behind current research in pragmatics
is the idea that a theory of linguistic communication is really only a special case of a
general theory of human action. According to this view, the various linguistic
subdisciplines such as phonology, morphology, syntax and semantics should be regarded
as the studies of different abstract aspects of underlying communicative actions.
Explanation of variation within each subdiscipline should preferably be functional, i.e. it
should relate the properties of the phenomenon being examined to the function of a
communicative action as a whole. If this task could be accomplished, the functionalist
claim is that linguistic theory would simultaneously achieve both increased
exhaustiveness and greater internal coherence and simplicity.
A precondition for success in this enterprise is the establishment of a conceptual
framework for the description of action, with enough internal structure to make possible
an account of the relationship between linguistic phenomena and action with sufficient
detail to be convincing. Two of the most important contributions to the creation of such
a framework have been made by Austin 1962 and Searle 1969. In this paper, I want to
discuss some of their respective contributions critically and suggest some ways in which
they might be improved.
I turn first to a discussion of Austin. Austin 1962 is a work which was published
posthumously by Austin's students and can therefore not be regarded as fully
representative of Austin's views had he been given time to finish his work. Austin
proposed a number of distinctions which have inspired a substantial amount of research
on language use during the past 25 years.
See Morris 1938 for the distinction between syntax, semantics and pragmatics.
As far as I can see, the changes made in the new and revised edition of Austin 1962 (ed. by M. Sbisà
and J. O Urmson) do not have any essential implications for the points I wish to raise with regards to
Among other things, Austin suggests that we should distinguish the locutionary aspect
of an utterance from its illocutionary and perlocutionary aspects. To be more precise,
Austin claims that in uttering a sentence, one concomittantly produces three acts: a
locutionary, an illocutionary and a perlocutionary act.
Austin 1962, 108:
'/performing/ a locutionary act ... is roughly equivalent to uttering a certain
sentence with a certain sense and reference, which is again equivalent to meaning in
the traditional sense. Second, we said that we also perform illocutionary acts
such as informing, ordering, warning, undertaking, etc., i.e. utterances which have a
certain conventional force. Thirdly, we may also perform perlocutionary acts:
what we bring about or achieve by saying something, such as convincing,
persuading, deterring and even, say, surprising or misleading'.
My first objection concerns the choice of the term act. This term gives the impression
that Austin is talking about temporally distinct activities rather than simultaneous
aspects of one and the same action.
Further, it is unfortunate that Austin only
discusses the distinction as applied to linguistic utterances. It seems fairly clear that the
distinctions he is after apply to communicative actions in general including non-verbal
ones and not merely to linguistic communicative activity.
Secondly, I would like to discuss in some more detail what aspects of an action Austin's
distinctions really concern. The locutionary aspect seems to correspond to the
conventional content, i.e. to whatever information is tied to an utterance by convention.
The illocutionary aspect is said to be identical with something Austin calls the
conventional force of an utterance.
(1) It's snowing
Thus, in (1), the locutionary aspect of the sentence says something about the weather
conditions while the illocutionary aspect of (1) is its conventional force as a statement.
But what do we do about a sentence like (2) below?
(2) I promise to buy you a drink.
(2) by convention seems to be both a promise and a statement. Furthermore, the fact
that it is a promise seems to be derivable from the conventional content of the word
promise and the fact that (2) is an indicative statement. A speaker is obliged to vouch for
the truth of all statements he makes, so that if he states that he is promising, i.e. states
that he is committing himself to some future action, then he is obliged actually to commit
himself if he is to tell the truth. His statement that he is promising commits him to
promise which in turn means that he has committed himself to carry out a future action.
This point has also been noted by other authors, see e.g. Wunderlich 1974, 324.
Thus, the actual force of (2) is derivable from the conventions governing the making of
statements in conjunction with the conventional content of the word promise. Therefore,
our knowledge of what kind of communicative action (2) is to count as derives from our
insight into the locutionary dimension of (2). This does not mean that the locutionary
aspect can be equated with the illocutionary aspect, it just means that we often recognize
our utterances as being of a certain type of action through the conventional content of
the words we employ. (2) could, of course, be used for other purposes than promising,
e.g. as an example in a linguistic article. So, recognition of illocutionary force is often a
function of, among other things, comprehension of the locutionary dimension of an
Next I want to turn to the question of in what sense illocutionary forces are
(3) There is a bull in the meadow.
(4) I warn you there is a bull in the meadow.
If we accept what has been said above about sentence (2), it is comparatively clear how
(4) can be taken as a warning. The lexical content of warn in conjunction with the
indicative form of (4) makes it natural to think that (4) has the force of a warning. As far
as I can see, the only conventions that are required to reach this result are the one
connecting indicative form with stating and the one connecting the sound sequence warn
with a content roughly identical with 'making somebody aware of possible danger or
unpleasantness connected with an expected course of action'. No specific conventions
for warning other than the lexical conventions connecting the word warn with its content
are really required. But this lexical type of convention does not seem to be a suitable
candidate for the kind of convention which, according to Austin, is supposed to
constitute acts of warning. It seems that the notion of conventional force
rather explicable in highly institutionalized contexts, like religious or legal activity, as the
set of social consequences of a certain action does not allow an equally neat explication
in less institutionalized situations. Compare the social consequences of an act of baptism
with those of an act of warning. At least in traditional Christian societies, the act of
baptism has been tied to very definite rights and obligations, but it is hard to find any
social consequences connected with a warning over and above those it has by virtue of
being a statement, i.e. a commitment to a certain representation of reality. This becomes
even more obvious if we consider cases such as (3). It is easy to imagine that a sentence
like (3) might function as a warning. But is this due to any special convention for
warnings? Does it make sense to say that (3) has the conventional force of a warning?
What is at stake in the case of (3) seems instead to be the natural connections most of us
can imagine between what is designated by the conventional content of (3) and some
individual being conscious of possible danger connected with a certain course of action.
No other conventions seem to be required.
See Allwood 1976, 122 for further discussion of this point.
Thus, for cases of 'illocutionary force' that are not tied to so-called performatives
occurring in highly institutionalized contexts, we seem to have two cases:
1. The illocutionary force is explicitly indicated by a lexical item. In this case, there are
two types of convention involved, namely:
a. Lexical conventions operating on the locutionary level tying certain lexical items to
certain contents and
b. Modal conventions
connecting what linguists usually call mood (a certain syntactic
arrangement or specific type of verbal affix) with certain communicative actions. In
English, there seems to be at least three such connections: The indicative form is
connected with stating, the interrogative form with questioning, and the imperative
with requesting or ordering
However, neither the lexical conventions nor the mood conventions directly constitute
the illocutionary force of such acts as promising or warning. Rather, these two types of
conventions together make it possible to derive what the force of a certain utterance
containing a mood marker and lexical items like illocutionary verbs reasonably should be.
2. In the second case, such as (3), there is no explicit lexical indication of force. Instead,
what enables us to derive the force of utterances such as (3) seems to be the natural
connections that obtain between the conventional content of such an utterance and a
specific type of communicative action such as warning. Thus, in order to explain that (3)
can function as a warning, no specific conventions for the illocutionary force of warnings
is needed. We only need to understand what it means to warn somebody.
But if what makes (3) or (4) into warnings is not conventional illocutionary force, it is
reasonable to ask the more basic question: What is it that makes a certain utterance or,
more generally, an instance of communicative behavior into a communicative action of a
specific type? The notion of force does not seem to be of much help outside
The answer I would like to suggest to this problem is that the identity of a
communicative action should be determined in exactly the same way as the identity of
other actions. To the extent that our ordinary concept of action suffers from a certain
unclarity, so will our notion of communicative action.
There seems to be at least four common ways of identifying an action. One way is to
identify an action through the effects or results of the action. This first type of
A more thorough discussion of modality is given in Allwood 1976, 124.
For an argument that 'the exclamative' should be included as the mood for sudden expression of feeling
or attitude see Quirk et al. 1972. Since the subjunctive has such a limited use in English, I find it hard to
understand its precise function. However, I think it can at least partly be analyzed as a marker used to
indicate neutralization of statement status.
identification does not necessarily have anything to do with the kind of action an agent
thinks he is performing, i.e. an agent who is driving a car can be said to be polluting the
air even though this is not his intention in driving the car. This type of identification of
an action therefore depends on the interpretation an observer or the agent himself can
give of the effects of a certain type of behavior, irrespective of whether they correspond
to the agent's intentions or not.
A second type of identification relies on the intentions and purposes an agent connects
with an action, irrespective of whether they are achieved or not. An argument for this
type of identification is that it makes it possible to speak of an agent as performing a
certain action before the purpose of the action has been achieved. For example, it is
possible to say of a person A that he is making a chair before the chair has been finished.
A third type of identification is tied to the overt form of behavior an agent exhibits in
performing an action. This type of identification acquires special strength when some
overt form of behavior is connected by convention with the performance of a certain
action. For example, we would use this criterion if we claimed that a rhetorical question
is indeed a question since it occurs in interrogative form, even though it is intended as a
statement and therefore, by the criterion of intention, should be classed as a statement.
Fourthly, identification can sometimes be accomplished solely on the basis of contextual
extrapolation. Some behavior can be classed as action of a specific type just because it
occurs in a certain context. The communicative action labelled answering at least in the
widest everyday sense of the word seems to be of this type. Any verbal behavior that
follows a question is in this sense an answer. For example (6) below would be counted as
an answer to (5) even though it is irrelevant to the purpose of (5).
(5) What time is it?
(6) My puppy is ill.
It is not clear how Austin's notion of illocutionary force, which primarily seems to be an
extrapolation from institutionalized performatives, can be related to these four criteria.
Can the illocutionary force of a warning be identified with
1. the intention to warn, i.e. with the intention of evoking awareness of danger in
2. some specific type of overt behavior,
3. certain specific contexts or
4. some person actually being warned, i.e. becoming aware of danger connected with his
course of action?
Since I think the answer in all cases is no
for non-institutionalized performatives, it is
difficult to see just what role the notion of illocutionary force can play in a theory of
communication which includes non-institutionalized communicative actions.
Let me now turn to Austin's notion of perlocutionary act. Unfortunately, this notion
retains the ambiguity noted above between intended effect and actually achieved effect.
Austin 1962, 101, writes:
'Saying something will often, or even normally, produce certain consequential
effects upon the feelings, thoughts or actions of other persons: and it may be done
with the design, intention or purpose of producing them; ... We shall call the
performance of an act of this kind the performance of a perlocutionary act or
As we see, this ambiguity is not resolved. Things are made even more difficult by the
fact that Austin included in the illocutionary force of an act its 'pure uptake' by a
receiver. The distinction between illocutionary and perlocutionary acts would have been
easier to maintain if all receiver reactions had been counted as perlocutionary effects.
It should perhaps be noted that the difficulty of distinguishing between intended and
achieved effect would also adhere to an attempt to characterize illocutionary force in
terms of the social consequences of an action, since one should then distinguish between
intended and achieved social consequences.
Thus, while I fully acknowledge Austin's pioneering contribution to the study of the
relationship between action and communication, I am somewhat critical of certain
aspects of his account. Either the concepts he has proposed should be further explicated
or a new terminology and conceptual framework should be adopted. In
what has been said above, I have made some suggestions in the latter direction. In sum, I
would like to propose that the following features of communicative and other actions be
considered as a more suitable conceptual framework for the study of communicative
1. The intention and purpose (intended effects) of a communicative action.
2. The actual overt behavior used to perform the communicative action.
3. The context in which the communicative action is performed.
4. The actually achieved effects in a receiver of the act of communication (these effects
need not be identical with the intended effects).
I am not discussing here formal warnings such as can be given by parking attendants or the police.
These have a conventional, legal force and very often a strict conventional form.
For a proposal of this kind see Wetterström 1975.
For more detail as to what this proposal entails see Allwood 1976, esp. chapters 7, 8, 9, 10 and 14.
5. As an extension of 4, the notion of conventional force, i.e. the social consequences of
a certain communicative action.
In Searle 1969, 24-25, the author presents a theory which is a development of the
account presented in Austin 1962. Searle claims that four acts are characteristically
performed in the utterance of a sentence:
(p.24) a. Uttering words (morphemes, sentences)
= performing utterance acts
b. Referring and predicating
= performing propositional acts
c. Stating, questioning, commanding, promising etc.
= performing illocutionary acts
(P.25)d ."To these three notions, I now wish to add Austin's notion of the
perlocutionary act. Correlated with the notion of illocutionary act is the
notion of the consequences or effects such acts have on the actions, thoughts
or beliefs etc. of hearers. For example, by arguing I may persuade or
convince someone, by warning him I may scare or alarm him, by making
a request I may get him to do something, by informing him I may
convince him, enlighten, edify, inspire him (get him to realize). The
italicized expressions above denote perlocutionary acts'.
If we substitute 'utterance act and propositional act' for locutionary act, I believe most of
the problems discussed above with Austin's theory also apply to Searle's theory. Searle
himself points to the difficulties with the act terminology (Searle 1969, 24), but retains it
all the same. In fact, Searle makes things a little worse since he introduces the term
speech act where act must stand for a whole which is composed by the locutionary,
illocutionary and perlocutionary acts. Furthermore, the term speech act is unfortunate
since it would seem to restrict Searle's claims to communication via speech, whereas in
fact most of his arguments are equally valid for non-verbal warnings, refusals, denials and
The difficulties found in Austin's notion of illocutionary force are also present in Searle's
treatment. It remains unclear to what extent illocutionary forces are conventional,
intentional or tied to achieved effects, overt behavior or context. Searle also gives a
number of conditions for certain particular speech acts that he believes should be met if
the act is to be felicitous. I think these conditions are generally correct but misleading in
that, as a rule, they make general features of action and communication into particular
features of particular speech acts.
I will now summarize these conditions as they
appear in Searle's analysis of promising in chapter 3, Searle 1969, and then comment on
1. Normal input and output conditions:
Communication is supposed to be literal and serious and to take place between a
sender and a receiver who are physically and physiologically able to communicate.
Comment: This seems to be a general requirement on communication (and this is
mentioned by Searle). The only objection I have to this condition is that Searle assumes
that verbal communication normally is literal. As far as I can see, this is the exception
rather than the rule.
2 and 3. Propositional content conditions:
2. The utterance act should express a proposition.
3. The proposition predicates a future action of the speaker
Comment: Searle uses the term proposition in such a way that questions and
imperatives also can be said to express propositions. While I agree that there are
similarities between the contents of questions, imperatives and statements, I do not
think the best way to capture this similarity is through the notion of proposition, which
has a relatively clear explication in relation to statements, but not in relation to questions
The predication of a future action of the speaker is a consequence of the meaning of the
word promise and the fact that the speaker is making a statement about himself, a
statement that he is promising, that is, undertaking an obligation to perform the action
indicated in the subordinate clause. Since there are no time-machines, the action will have
to be a future action. Similar restrictions on the content will be derivable from the
meaning of other speech act verbs and general considerations of the nature of the world
4 and 5. Preparatory conditions:
4. The speaker assumes that the hearer wants him to perform this action; and the
hearer actually does want him to perform it.
5. It is not obvious to both speaker and hearer that the speaker will perform the
See Grice 1967, lecture 1, for criticism in a similar vein.
See Allwood 1976, chapter 7.
Comment: 4 does seem to bring out a trait which is specific to promising, i.e. that
promising, as opposed to threatening or scaring, does seem to be a communicative action
which requires the speaker's ethical consideration of the receiver.
Condition 5, however, is due to a much more general tendency in human behavior,
namely that one's behavior should have purpose and point and that one should not
spend more energy than necessary on anything.
6. The sincerity condition:
The speaker intends to perform the action his expressed proposition predicates of
Comment: This again seems to be a very general condition having to do with the fact that
most communicative acts by convention convey information about a certain attitude of
mind or motive that is supposed to be expressed by the communicative act in question.
Further, it is bad to be unethical: not having the attitudes one by convention is supposed
to have in performing a certain communicative act would be to deceive, and deception is
unethical. Thus, condition 6 is really a general ethical requirement on communication to
the effect that it should not be deceptive.
7. The essential condition:
The speaker intends that his utterance should place him under an obligation to
perform the action in question.
Comment: This condition derives directly from the meaning of the word promise and the
fact that the speaker is making a claim about himself. In general, when a speaker makes a
claim, he is under obligation to vouch for the truth of the claim, which in this case entails
that he regards himself under an obligation. To place oneself under an obligation is a
social phenomenon which is created by convention. But it is a phenomenon which is
much wider than promising. Promising is a communicative action which seems to involve
a combination of (i) ethical considerations of the receiver and (ii) obligation undertaken to
perform an action on behalf of the receiver. The word promise is then by lexical
convention tied to this combination. Thus, the conventions involved are on the one hand
lexical ones and on the other conventions creating the possibility of undertaking
obligations. Only the former are specific to promise. The obligation created by uttering a
sentence like I promise to go are thus due to the conventional force tied to the
declarative mood and the use of the word promise.
8. The non-natural meaning condition:
The speaker intends the hearer to realize that the speaker is placed under an
obligation to perform the action in question by his utterance, and that the hearer's
realization of this should be by virtue of his knowledge of the meaning of the
Comment: As Searle points out, Grice 1971 is the source of condition 8. If Grice is right,
then this condition is a feature of all communication and not specific to promising (this is
also implied by Searle) or for that matter to verbal communication or communication
with conventional content.
9. The defining condition:
The speaker's utterance is a correct and sincere promise if conditions 1-8 obtain.
Comment: This condition is correct. But this is not due to any specific conventions for
promising. Rather, it is due to the lexical meaning of promise in conjunction with certain
general features of action and communication. The only special social conventions that
have to be invoked, besides those of a purely lexical nature, are those that make it
possible for an individual to take responsibility for a certain representation of reality.
Thus, it seems to me that Searle's analysis should be seen as an elegant characterization
of how certain general conditions on action and communication interact with the lexical
content of certain communicative activity verbs and with the conventional force
connected with certain mood markers. Such an analysis is valuable since it is of interest
to find out exactly what combinations of general conditions and idiosyncratic factors that
can be denoted by the various communicative activity terms of a natural language.
Analogously, it is important to characterize exactly the factors that help to constitute
human communicative action.
Finally, a few general remarks. Austin's and Searle's approaches have in common that
they concentrate on single communicative acts. I think there is a certain danger inherent
in this, namely that it is easy to lose perspective on communication as a whole. After all,
communicative acts seldom occur in isolation, but rather sequentially in interaction. It
might therefore be better to study larger chunks of communication, a study for which we
might use Wittgenstein's term language games.
Secondly, there is always a danger that the analysis of communication becomes too
linguistic by focusing on the meaning of certain communicative activity verbs. The
analysis of such meanings should be as clearly separated from the study of actual
communication phenomena as possible. If this can be accomplished, the results will
probably be positive in both directions. Not least linguistically, since we would then get
a better idea of the conceptual parameters that underlie the lexical field of communicative
What then should be salvaged and further developed in the work of Austin and Searle?
At least the following three areas seem to be of some interest:
1. The study of different communicative intentions,
2. The study of the various psychological and behavioral reactions characteristically
evoked in a receiver during communication, and
3. The study of social consequences of acts of communication.
All three factors should then be correlated on the one hand with biological and
psychological factors in human beings and on the other hand with material conditions
and social power and solidarity relations in society at large.
Allwood J. 1976: Linguistic Communication as Action and Cooperation
Gothenburg Monographs in Linguistics 2, Dept. of Linguistics,
University of Göteborg.
Austin J.L. 1962: How to do things with words University Press, Oxford
1975: How to do things with words, new revised edition, M. Sbisà & J.
O. Urmson (eds), Clarendon Press, Oxford.
Grice P. 1967: Logic and Conversation lst lecture William James lectures
1971: 'Meaning' in Steinberg & Jacobovits (eds.) Semantics, Cambridge
University Press, Cambridge.
Morris C.W. 1938: Foundation of the theory of signs. International Encyclopedia of
Unified Science, vol. 1, No. 2, University of Chicago Press.
Quirk R., Greenbaum S., Leech G. & Svartvik J. 1972: A grammar of contemporary
English, Longmans, London.
Searle J.R. 1969: Speech Acts, Cambridge University Press, Cambridge, Mass.
Wetterström T. 1975: Intention and Communication unpubl. thesis, Dept. of
Philosophy, University of Göteborg, Göteborg.
Wunderlich D. 1974: Grundlagen der Linguistik. Studium rororo Rowohlt.