ILLOCUTIONARY ACTS

Item Type text; Dissertation-Reproduction (electronic)

Authors Carr, Charles Raymond, 1945-

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this materialis made possible by the University Libraries, University of Arizona.Further transmission, reproduction or presentation (such aspublic display or performance) of protected items is prohibitedexcept with permission of the author.

Download date 13/03/2021 10:21:54

Link to Item http://hdl.handle.net/10150/289262

INFORMATION TO USERS

This material was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted.

The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction.

1. The sign or "target" for pages apparently Sacking from the document photographed is "Missing Pags(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity.

2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame.

3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. Of necessary, sectioning is continued again — beginning below the first row and continuing on until complete.

4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced.

5. PLEASE NOTE: Some pages may have indistinct print. Filmed as received.

Xerox University Microfilms 300 North Zeeb Road

( Ann Arbor, Michigan 48106

76-1399

CARR, Charles Raymond, II, 1945-ILLOCUTIONARY ACTS.

The University of Arizona, Ph.D., 1975 Philosophy

Xerox University Microfilms, Ann Arbor, Michigan 48106

© 1975

CHARLES RAYMOND CARR, II

ALL RIGHTS RESERVED

THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED.

ILLOCUTIONARY ACTS

by

Charles Raymond Carr, II

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF PHILOSOPHY

In Partial Fulfillment of the Requirements For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

19 7 5

Copyright 1975 Charles Raymond Carr, II

THE UNIVERSITY OF ARIZONA

GRADUATE COLLEGE

I hereby recommend that this dissertation prepared under my

direction by Charles Raymond Carr, II

entitled Illocutionary Acts

be accepted as fulfilling the dissertation requirement of the

degree of Doctor of Philosophy

issertation Director

After inspection of the final copy of the dissertation, the

following members of the Final Examination Committee concur in

its approval and recommend its acceptance:""

73

£L 21.

This approval and acceptance is contingent on the candidate's

adequate performance and defense of this dissertation at the

final oral examination. The inclusion of this sheet bound into the library copy of the dissertation is evidence of satisfactory

performance at the final examination.

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the copyright holder.

SIGNED:

ACKNOWLEDGMENTS

A number of people have made valuable contributions

to this work. I have tried as much as possible to acknowl­

edge individual contributions, but much remains in here for

which people other than myself deserve the credit. A

number of graduate students at The University of Arizona

have patiently listened to, and criticized my arguments and

proposals. The work is improved, not to mention lengthened,

by the suggestions I have incorporated into the study which

were offered by my committee. J.L. Cowan has offered a

number of examples which have forced me to rethink the scope

of my project, and has made suggestions for handling some of

the material. Henry Byerly has corrected a number of mis­

takes in earlier drafts, and offered a number of valuable

proposals. I could not be much more indebted than I am to

R. M. Harnish. It was he who first introduced me to the

subject, and he has overseen the project at each stage of

its development. As much as I do he deserves credit for

whatever is correct about this work, or at least inter­

estingly wrong. I have not yet decided where to assess

blame for the parts that are uninterestingly wrong.

Lastly, I would like to acknowledge the contribu­

tions of my wife. In addition to the encouragement she has

iii

iv

provided along the way she has been willing to miss watching

Kojak in order to proofread typescript.

TABLE OF CONTENTS

Page

LIST OF ILLUSTRATIONS vi

ABSTRACT vii

CHAPTER

I. SPEAKER MEANING AND ILLOCUTIONARY ACTS 1

Grice's Account of Meaning and Illocutionary Acts 12

Schiffer , . 18 Modifications of the Analysis 40 Other Features of Act-Type Identification . 64

II. INSTITUTIONS, CONVENTIONS, AND ILLOCUTIONARY ACTS 76

Institutional Acts 82 Conventional Acts 98 A Taxonomy of Illocutionary Acts 105 Analyses of Some Illocutionary Acts . . . . 1 1 8 Moore's Paradox as a Test of Theories . . . 135

III. UTTERANCE MEANING AND FORCE 139

Cohen 141 Frye 147 Utterance Force 151 Speaker Meaning, Expression Meaning, and Communication ..... 176

IV. ILLOCUTIONARY ACTS AND LANGUAGE 180

Illocutionary Acts and Grammar . 180 Interpretation of Grammars and Speaker Meaning Theories 209

Illocutionary Acts and the Use of Language 217

Performative Utterances t 240

REFERENCES 254

v

LIST OF ILLUSTRATIONS

Figure Page

1. A Chart of Illocutionary Acts 13 6

2. Illocutionary Act Sentence Restrictions .... 202

vi

ABSTRACT

The concept of an illocutionary act has emerged

as a central concept within the theory of language use.

Among attempts to characterize illocutionary acts one that

offers a good deal of promise as a vehicle of inquiry is the

suggestion that a speaker performs an illocutionary act if

and only if he means something. This study attempts to

provide a theory of illocutionary acts within this frame­

work. An account of speaker meaning is developed along the

lines of those which claim that to mean something is to

intend in a certain way to achieve a certain effect. Among

the reasons accounts of speaker meaning within this tradi­

tion fail to provide an adequate basis for the core of an

analysis of illocutionary acts is that the requirement that

the speaker intend to produce a belief or intention in the

audience is too strong. Consequently, an account of speaker

meaning with a weakened intended effect is provided. With

the new analysis of speaker meaning it is reasonable to

suppose that to perform an illocutionary act just is to mean

something. Arguments are given that this holds for the

more "conventional" illocutionary acts like christening

which have been held to resist a speaker-meaning analysis

as well as cases like stating or warning.

vii

viii

The relation between speaker meaning, illocutionary

acts, and expression meaning is the subject of the second

half of the study. One important suggestion is that even

within a specification of what a speaker meant, different

aspects of that specification are being sought in a request

for the meaning than are being sought in a request for the

force. The theory of illocutionary acts developed in the

first part of the dissertation characterizes those different

aspects. Next, an argument is provided that illocutionary

force attaches to expression object types. In the case of

non-performative sentences it is a generic force that is not

incompatible with any specific illocutionary act that the

speaker performs by using that expression. This force is

then characterized.

A certain amount of illocutionary act information

would be represented in a grammar of a language, A formal

means of representing this information is adopted from work

done in semantic theory, and possible interpretations of the

formalism consistent with a theory which makes speaker

meaning basic are discussed. Certain information about

illocutionary acts is such that it would be represented in a

grammar. For example, indirect speech acts cannot be

predicted from a linguistic description of the expression

uttered in their performance. An account of how, in such

cases, a speaker's intentions are realized is offered.

ix

The study concludes with a discussion of whether

performative utterances are capable of assessment in terms

of truth and falsity. A number of arguments to show they

are not are criticized. It is suggested that the issue is

one for a more comprehensive theory of language use to

decide.

CHAPTER I

SPEAKER MEANING AND ILLOCUTIONARY ACTS

The project to be undertaken in this dissertation is

to provide a theoretical account of illocutionary acts, and

to relate the concept of an illocutionary act to other

central concepts within the theory of language use. It is

not too difficult to go through a dictionary and identify

which of the verbs found there are illocutionary verbs.

"Answer," "tell," and "order" are, while "persuade" and

"convince" are not. Austin (1962) said that verbs like the

latter pick out perlocutionary acts. They describe a

particular effect on the hearer of what the speaker said.

Austin never provided an analysis of the concept of

an illocutionary act, but he does offer a number of charac­

teristics which do help to identify the class of illocu­

tionary acts. None of the criteria proposed by Austin,

which Austin admits, provide necessary and sufficient con­

ditions.

The illocutionary acts cannot be identified by the

fact that they can be performed with the explicit performa­

tive prefix. "Threaten" is an illocutionary, but not

clearly a performative verb. "I hereby threaten you," is

1

2

not clearly a grammatical sentence. In my idiolect it is

ungrammatical.

Austin (1962, p. 99) suggests that illocutionary

acts are acts performed "in" saying something. While this

may separate them from perlocutionary acts he felt (pp. 104-

105) it did not provide by itself a sufficient condition as

joking may be done "in" saying something.

He considers whether it might be the case that there

must be conventional means for the performance of the act if

it is an illocutionary act. Austin (1962, p. 118) concedes

that one might protest by hurling a tomato.

Austin (1962, p. 114) suggested that illocutionary

acts are acts which are constituted essentially by conven­

tion. I intend to offer an analysis of illocutionary acts

which challenges this characterization of illocutionary

acts. In this respect I follow Strawson (1971). Schiffer

(1972, p. 91) made the point quite firmly. "Perhaps there

are some speech acts—e.g., an umpire putting a runner out

by uttering 'Out!'—which are conventional in the sense

intended by Austin, but these are very special cases and of

peripheral interest only." In this dissertation I hope to

make clear why I concur with Schiffer.

Although illocutionary acts are generally easy to

identify, providing an analysis is a rather more difficult

matter. The task can be made even more difficult by failing

to realize that a number of illocutionary verbs pick out two

3

distinct types of cases. Only one of them is of interest

here. The two types of cases may ultimately be related,

either historically, metaphorically, or in some other way.

I do not intend to pursue that connection.

The two different kinds of cases can be illustrated

by the following examples. Suppose that Jones asks Smith

what time it is. Smith hears Jones' question, and delib­

erately responds that it is two o'clock. Smith has

answered Jones' question. Suppose that Jones asks the same

question, but Smith does not know what time it is or does

not hear the question. However, Jones overhears a conver­

sation in which someone tells someone else what time it is.

This person has also answered Jones' question, answered it

in the same sense that he has produced the information which

Jones was seeking. Compare this with the case where Jones

and Smith are shipwrecked on a desert island, and Jones asks

if any animals live on the island. As they search the

island they come across what is obviously a freshly made

animal track. This track answers the question in the sense

that it provides the information Jones was seeking. It

answers the question in the same way a clap of thunder

answers the question of whether there will be a storm.

The second case given above where a question about

the time is answered in virtue of overhearing a conversation

is like the two cases of a natural phenomenon answering the

question. The status that it has as an illocutionary act is

4

not in virtue of being an answer to Jones's question. The

chiming of a clock or the length of the sun's shadow would

have served just as well, and clearly these are not illocu-

tionary acts. (I became aware of the need for distinguishing

these two kinds of cases after discussion with J. L. Cowan

[1975], although it should be noted that he disagrees with

my view that the first two cases are importantly different.)

A list of illocutionary verbs which pick out these

different kinds of cases would include "warn," "threaten,"

and perhaps "question." All of these cases seem to have the

property that in one type of case a natural phenomenon could

be that in virtue of which it is correct to the term to

describe the situation. It is only the kind of case given

initially which would play a central role in a study of

communication. Correspondingly my discussion is intended

to be about these types of cases, and when I speak of illo­

cutionary acts it is these types of cases to which I intend

to refer.

There is a further point to notice in this regard.

The aim here is to handle as much material as possible in a

systematic way. Many of the illocutionary verbs, for

example, "state," do not have exact boundaries in ordinary

language. In such cases I assume it is part of the job of

a theory to draw the boundaries.

It is a sad fact of our history that at one time

both persons who were mentally retarded as well as those who

5

were emotionally disturbed went under the appellation, "mad."

A theory which attempted to distinguish these cases, and

restx-ict the term to one of them, is not to be faulted for

failing to draw the boundary where those competent users of

the language who understood the criteria for the term's

application drew it. A theory of illocutionary acts is not

stepping out of line to demand a certain license to legis­

late at the boundaries or to distinguish cases which are

importantly different. The test of such a theory is how

accurately and systematically it can account for the bulk of

the data. This dissertation will proceed on the assumptions

that there are these two distinct kinds of cases, and that a

theory is entitled to draw (or redraw) boundaries. The study

presupposes some familiarity with the work of Grice (1957,

1969) , Strawson (1971), and Schiffer (1972). I will not for

the most part repeat their replies to general objections to

their programs. They have done this adequately themselves,

and I hope only to add to the foundations they have lain,

and to defend the additions I make.

Strawson (1971) suggested that a speaker performed

an illocutionary act if and only if he meant something.

The notion of what it is for a speaker to mean something he

adopted from Grice (1957). In short, that definition was

that a speaker meant something if and only if he intended

the au.dience to believe or do something in virtue of recog­

nizing the speaker intended them either to believe or do

something. Strawson (1971) was not here denying Austin's

(1962) distinction between meaning and force, but attempting

to analyze what was central to an illocutionary act. He

felt that in certain highly conventional cases the act might

go on in the absence of the speaker's intentions, that form

might take over. But he felt these cases were essentially

deviant or non-standard (Strawson, 1971, p. 36).

Grice's definition ran into a series of counter

examples designed to show either that the analysis was too

weak or too strong, and the definition became increasingly

more complex as Grice (1969) responded to these counter

examples.

Schiffer (1972) attempted to overcome many of the

difficulties of Grice's definition of speaker meaning with a

definition that is essentially faithful to Grice. Schiffer

follows Strawson, and claims that a speaker performed an

illocutionary act if and only if he meant something, with

Schiffer's own definition of speaker meaning replacing

Grice's. He then proceeds to give analyses of various

individual acts. However, counter-examples are easily found

to Schiffer's definition because the intended effect, what

he claims the speaker intends in performing the illocu­

tionary act, is too strong. In Chapter I, I suggest a

weakening of the intended effect which will avoid these

problems, and witn this revised analysis of speaker meaning

it is correct that a speaker performed an illocutionary act

7 %

if and only if he meant something. I then discuss the

features that are required for the identification of what

particular act was performed on a given occasion.

In Chapter II, I turn to a class of illocutionary

acts which appears to stand in the way of an analysis in

terms of a speaker intending to do something. It is the

class that consists of the so-called conventional illocu­

tionary acts, a collection which includes such diverse cases

as sentencing, bidding, and an umpire calling someone out.

It is important to point out, first, that this group is not

homogeneous. There are two very different kinds of cases.

However, even within each of these groups the extent to

which the illocutionary act can go on in the absence of the

speakers' intentions has been exaggerated. And those very

cases where the role of intentions is significantly dimin­

ished are rather rigidly circumscribed by rules which do

duty for the speaker's intentions. The second chapter con­

tinues with analyses of various illocutionary acts, a

taxonomy of illocutionary acts, and finally a discussion of

Moore's paradox as a test of theories.

In Chapter III, I delve more deeply into the con­

nection between the meaning of an utterance and the illocu­

tionary act that it is used to perform. Following Austin

(1962) a number of writers have distinguished the meaning

of an utterance from its source, from the illocutionary act

that would be associated with that utterance.

8

Two writers in particular have made important con­

tributions to that discussion. Cohen (1971) has denied that

there is any need to talk of the force of the utterance,

that talk of meaning is sufficient to handle all the data.

Cohen's exact thesis is obscured by the fact that "utterance"

is ambiguous along both act-object and type-token dimensions.

The resulting clarifications after the distinctions are made

will show that, at best, Cohen's arguments can show that the

concept of illocutionary force can be abandoned in favor of

the concept of speaker meaning.

Frye (1973) points out that even within a specifica­

tion of what a speaker meant, different aspects of what he

meant are involved in specifications of meaning and of

force. The view presented in the first chapter identifies

these various aspects to which Frye refers.

But what of the question of whether utterance object

types (expressions) have a force? Do they have a force, and

if so, how? I develop a conditional argument for the

existence of expression force. It is generally acknowledged

that expressions have meaning, and one contemporary view

of expression meaning is that it is conventionalized speaker

meaning. An expression, x, means that P because, roughly,

there is a pattern of mutual expectations that a speaker

would utter x in order to mean that P. If this is a correct

account of expression meaning, then there is no reason for

denying that expressions have a force. Certain of them

9

satisfy the requirement of there being certain mutual

expectations about what illocutionary act a speaker would be

performing by uttering that expression. By parity of

reasoning this should be sufficient to support the claim

that the utterance has that force.

That expressions do have a force is not implausible

in the case of explicit performatives. However, what about

other, non-performative, sentences? Does some force attach

to them? If so, then as they can ordinarily be used to

perform any of a number of different illocutionary acts a

choice must be made. Either each of the forces belongs to

the expression or some one general force which is not in­

compatible with any specific force the utterance has on an

occasion is the force of the utterance. I opt for the

latter choice, and attempt to characterize such a force.

In the final chapter I begin with a question which

is a very natural extension of the discussion of Chapter

III. Suppose that we had an adequate grammar of the

language. What information about illocutionary acts would

be represented in a grammar? For example, what information

would be information about the meaning of a particular

illocutionary verb? Searle (1969) makes it a condition on

promising that the action the speaker promises to perform be

a future act. I will argue that this fact would be re­

flected in the semantic representation of the performative

verb, "promise," that a grammar would provide. I will

10

assxime for discussion purposes the form a correct grammar

would take would be that described in the Standard Theory

(Chomsky, 1965), and further that Katz (1972) has correctly

described the semantic component of that grammar. I then

argue that from a linguistic description of a given non-

performative sentence, and from the semantic representations

of various illocutionary verbs, it is possible to determine

which of the illocutionary acts named by those verbs the

sentence might be used to perform,,

The utilization of a formalism raises questions

about what the intended interpretation of the formalism is.

Since I will use machinery from Katz semantic theory to

represent various aspects of the meaning of sentences it is

important to see what the use of such machinery commits me

to, and whether it is inconsistent with any of the theory of

illocutionary acts developed in this study.

The fact that a particular sentence was uttered in

appropriate circumstances will not, of course, guarantee

that a particular illocutionary act was performed. To know

if it was would involve knowing the intentions of the

speaker on that occasion. A grammar obviously cannot tell

us this, but there is a very intimate connection between

the illocutionary act actually performed on an occasion,

and the information contained in the grammar for those cases

v/here the speaker did not mean something different than

his words meant. Quickly, for explicit performatives, it is

11

this. The meaning of an expression on the view presented

here is determined by the communication intentions which for

certain reasons it is mutually expected that someone who

utters that sentence might have. To utter a sentence

literally is to utter it having just those intentions. If

the speaker's intentions are the ones that it is expected

that he might have in virtue of that utterance then the

speaker means just what the sentence means. What is is for

a speaker to perform an illocutionary act is to mean some­

thing. The particular illocutionary act is determinable

from various features of what he meant, the features he

would be expected to have intended in virtue of the meaning

of the expression. The grammar tells what intentions to

expect him to have. If he has those intentions then he

means what the words mean. What he means determines the

force of his utterance on that occasion.

Sometimes the illocutionary act performed involves

the speaker having intentions it would not be expected he

would have just from the meaning of the expressions he

utters. Spelling out how in such cases he intends his

intentions to be recognized, as the expression is not

enough, would require developing a whole theory of communi­

cation. Rather than embarking on such a task, I will con­

centrate on one very small group of cases of a speaker

meaning more than his words meaning. These are the cases

of indirect speech acts. Not all of these cases will be

12

handled in the same way. Some of them can be handled via

the mechanisms of particularized conversational implicatures.

Others resist this treatment because the connection between

what is said and what is meant seems too intimate. I will

offer one suggestion on how certain expressions can become

so closely connected with certain illocutionary acts without

it being the case that the expression has that force because

of its meaning, or because that is the conventional force

of that expression.

The other issue touched in this chapter is the issue

of whether performative utterances are either true or false.

I argue that the most commonly given reasons for denying it

are inadequate, and that they are capable of assessment as

either true or false. I turn now to the beginnings of this

project.

Grice's Account of Meaning and Illocutionary Acts

A significant part of the concern with the use of

language has been centered on the concept of the illocu­

tionary act. Austin (1962) was never able to provide an

analysis of what emerged for him as the central concept in

the theory of the use of language; rather he satisfied

himself with point out some of the difference and similari­

ties between illocutionary acts and other major uses of

language, and with classifying various kinds of illocution­

ary acts.

13

Attempts to characterize illocutionary acts received

a decided impetus with the publication of Strawson's (1971)

"Intention and Convention in Speech Acts." Strawson sug­

gested that a speaker, S, performed an illocutionary act if

and only if he meant something, with the operant analysis of

meaning being borrowed from Grice (1957) . Grice distin--

guishes the sense of "meaning" he is after, which he labels

"non-natural," from other senses of "meaning." Grice (1957,

pp. 377-378) asks the reader to consider these sentences.

Those spots mean (meant) measles. Those spots didn't mean anything to me, but to the doctor they meant measles. The recent budget means that we shall have a bad year. Those three rings of the bell (of the bus) mean that the "bus is full." That remark, "Smith couldn't get on without his trouble and strife," meant tnat Smith found his wife indispensable.

He then offers a number of features which distinguish the

first three sentences from the last two.

It cannot be said that the spots mean measles, but

he doesn't have the measles. However, it may be that the

three rings of the bell did mean that the bus is full with­

out it being the case that the bus is full. One cannot

argue in the first cases, but can in the last two from the

meaning of the sentence, "x meant that P," to a conclusion

about what was meant by x. A similar point holds for

arguments from, "x meant that P," to the effect that some­

one or somebody meant something by it. A further test which

14

separates the cases is whether a restatement can be found in

which the verb "mean" is followed by a phrase in inverted

commas. Finally, an approximate restatement of each of the

first three sentences can be found beginning with the phrase,

"the fact that . . . ," while in the last two cases this

would not provide a restatement of the meaning of the first

sentence.

Grice also considers sentences which can be

exemplified by the form "S means to do so-and-so (by x),"

where S is a human agent, as natural senses of "mean," and

so outside the scope of his analysis. In these cases what

is being said is that S intended to do so-and-so.

It does seem initially plausible to maintain that if

a speaker performed an illocutionary act he meant something.

The conditional in the other direction is not so apparent,

but certainly is a plausible enough hypothesis to merit

investigation.

A Grice-like definition is a natural starting point.

The overt character of S's intentions does serve to dif­

ferentiate illocutionary from other means of producing the

same intended effect. If a speaker orders someone to do

something rather than, say, uses subliminal means in trying

to get the person to do it, he intends in the former case

that the hearer know what the speaker is doing. This open­

ness is a feature of illocutionary performances. Grice's

analysis, depending on whether the 1957 or 1969 analysis is

15

used, accomplishes this openness with either a reflexive

intention or with suitably nested intentions.

"A meant something by x" is (roughly) equivalent to "A intended the utterance of x to produce some effect in an audience by means of recognition of this intention"; and we may add that to ask what A meant is to ask for a specification of the intended effect . . . (Grice, 1957, p. 385).

"U meant something by uttering x" is true iff, for some audience A, U uttered x intending (1) A to produce a particular response r; (2) A to think (recognize) that U intends (1); (3) A to fulfill (1) on the basis of his fulfillment of (2) (Grice, 1969, p. 151).

This dissertation will attempt to develop an account

of illocutionary acts within this framework. It will

attempt to see how far a theory which uses some such account

of speaker meaning as the core of an analysis can be pushed

in accounting for the data, and at what points, if any,

concessions must be made. The advantage of a theory which

uses as few primitives as possible is theoretical simplicity.

Every theory, at least every theory of which I know, be it

in physics, linguistics, psychology, or philosophy at some

point is faced with data for which it cannot account. In a

theory which uses very few primitives the danger is running

into that data more quickly than one might have hoped.

The test of such a theory is its ability to with­

stand counter-examples, and to give the most systematic

account of the data. A number of points should be made with

respect to counter-examples to the theory. A distinction

was made in the opening pages about different kinds of cases

16

being picked but by the illocutionary verbs. It was stated

that only the one kind of case is the one for which the

analysis was intended. I will assume that these cases can

be distinguished pre-theoretically. Counter-examples must

be directed to these cases.

The next point to notice in this regard was the

point noted earlier by Schiffer (1972). Some cases are more

important within a theory of language than some others.

Questioning, answering, and telling are more central and

widespread uses of language than are christening or re­

doubling. I do not know how to argue with anyone who does

not see the sense in which the former cases are more central

cases than the latter. For this reason I cannot in any

detail argue the point. If this is a serious shortcoming,

it is one which I must acknowledge. It seems clear, though,

that they are more widespread, that they occur in more

varied circumstances, and are more central to communication.

Communication would be changed more if there were no

questioning or answering than it would be if there were no

christening. For this reason, that some cases are more

central than others to communication, it is more serious a

failing if the theory gets these cases wrong than if it gets

the cases which are less central.

The final point about counter-examples is that

sometimes there are general clashes of intuition about

whether a certain case fits a certain description or not.

17

It is important not to claim that intuitions clash at a

point before all argument has been exhausted. To do other­

wise is to insure that no question will ever be resolved.

But at the point where intuitions really do clash, where it

cannot be resolved whether is such and such a case someone

really was conceding or warning the best theory can be

appealed to in order to decide the cases. This is one of my

philosophical articles of faith.

Further, I intend to legislate somewhat, as will be

obvious later, with "state" and "say." I wish to use "say"

in such a way that what the speaker said is closely tied

with the actual utterance produced. If a speaker uttered,

"He did it," meaning Jones broke the vase then he did not

say that Jones broke the vase, where the reference of in-

dexical items is filled in. He said, "He did it," or he

said that someone did something. The point of this legis­

lation is to allow some distinction I wish to make in

Chapter IV. I think the distinctions are legitimate ones

which no term unequivocally matches, and they will be of

some use. Except for the analysis of "stating" I do not

think they effect other points that I make about illocu­

tionary acts.

This chapter initiates the attempt to develop a

thorough account of illocutionary acts within this general

framework. It begins with one of the most promising

attempts.

18

Schiffer

Efforts to utilize some definition of speaker

meaning as central to an analysis of illocutionary acts have

not met with much success. It will be argued later that

Searle (1969), despite his proclaimed goals, never provides

more than the conditions for several particular act types,

and then only when the speaker uses an explicit performative

sentence. Strawson (1971, p. 35) admits that the list of

conditions he provides for a speaker having performed an

illocutionary act is probably neither necessary nor suf­

ficient.

Schiffer (1972) has produced what is certainly the

most thorough and sophisticated of attempts to carry out

Strawson1s (1971) program. Although his effort is the high-

water mark .of such attempts, the range of counter-examples

to Schiffer's analyses suggest that significant revision is

in order. In this chapter I will offer a number of counter­

examples to Schiffer's analyses of various illocutionary

acts, and then turn to some of the modifications that might

be made in order to obtain a more adequate characterization.

It should become apparent that some fundamental changes will

be required.

Schiffer is able to overcome problems which beseiged

other accounts of speaker meaning because of modifications

he effects in Grice's analysis—the central task of his

work. Schiffer (1972, pp. 18-26) developed successively

19

more complex counter-examples along the lines of those

developed by Strawson as Grice's analysis of speaker meaning

became successively more complex. Then he provided a

definition which was able to avoid these counter-examples.

The counter-examples were directed against the formulations

of speaker meaning in the 1969 analysis. In Schiffer's

cases the speaker had all of the intentions that the defini­

tion required, but had one further intention which he did

not intend the audience to recognize that he had. Schiffer's

analysis used the concept of mutual knowledge*to insure all

the intentions were in the open, and hence not subject to

this kind of counter-example. In his analysis the primary

intention of the speaker was required to be mutually known*

(or at least was so intended) by the speaker and the

audience.

According to Schiffer (1972, p. 33) something is

mutually known* by a group of people if each knows it,

knows the others know it, knows the others know he knows

the others know it and so on. He believes that cases of

mutual knowledge* are not uncommon, and provides a simple

example. Suppose that two people are sitting at a table

with a candle between them. Each knows there is a candle

there. Further, they each would be expected to know that

the other knows that he know there is a candle there. They

mutually know* there is a candle on the table.

20

In his own analysis Schiffer divides cases of

speaker meaning into two classes: the speaker, S, either

meant that P (meant the audience, A, was to believe some­

thing) or meant that A was to ijj (that A was to do something) .

(I) is the definition for the imperative class where S meant

A was to ip. The definition for the assertive class is

analogous.

II)

S meant that A was to by (or in) uttering x iff S uttered x intending thereby to realize a certain state of affairs E which is (intended by S to be) such that the obtainment of E is sufficient for S and A mutually knowing* that E obtains and that E is conclusive (very good or good) evidence that S uttered x with the primary intention

(1) that there be some r such that S's utterance of x causes a to ty/r and intending

(2) satisfaction of (1) to be achieved, at least in part, by virtue of A's belief that x is related in a certain way R to (the act-type) ip-ing

(3) to achieve E (Schiffer, 1972, p. 63).

Likewise, illocutionary acts are divided into two

classes depending on whether S meant that P or meant that

someone was to \p. Now, although it is sufficient for S to

have meant something for him to have performed some illocu­

tionary act, knowing what was meant is not always sufficient

for determining which kind of illocutionary act was per­

formed. For Schiffer, identifying kinds of illocutionary

acts requires being able to identify either what was meant,

21

what types of reasons A was intended to have for believing

that P or for t^-ing, or in some case knowing both what was

meant and what kinds of reasons A was intended to have.

Hence, if S performed an assertive illocutionary act, that

act is identifiable either by the form of beliefs that

characterizes P, the form of beliefs which characterizes

the reasons which were intended, in part, to cause A's

belief, or by both. The description of the imperative class

is the same although Schiffer believes the class identi­

fiable just by what was meant is empty.

Something should be said about the sense of "intend"

which Schiffer and Grice are using in their definitions.

Grice (1957, p. 387) says the following about intentions:

Explicitly formulated linguistic (or quasi-linguistic) intentions are no doubt comparatively rare. In their absence we would seem to rely on very much the same kind of criteria as we do in the case of non-linguistic intentions where there is a general usage. An utterer is held to intend to convey what is normally conveyed (or normally intended to be conveyed), and we require a good reason for accepting that a particular use diverges from the general usage (e.g., he never knew or had forgotten the general usage). Similarly in non-linguistic cases. We are presumed to intend the normal consequences of our actions.

That S "intended" so-and-so does not just mean that the

action was deliberately planned. A variant of an example

that Grice provided in reference to the above would be

relevant here. A man may plan to take a note home with him

that evening, but later catch himself throwing it in the

trash. He might say about his action either that he did not

22

intend to do that or admit that when he caught himself he

was intending to throw it away. The latter of these

descriptions, where to intend is to do more than just to

have formulated previously a plan is the "sense" of

"intend" that is relevant here.

Given the general correctness of Schiffer's (1972)

analysis of speaker meaning and of illocutionary acts it is

reasonable to suppose that illocutionary acts would be

identifiable either by what was meant or by what reasons A

was to have for believing or doing. Some examples should

help substantiate the point. Consider an imperative case

like ordering someone, as opposed to advising someone, to

leave the room. The intended effect of the utterance (A's

i^-ing) might be the same in each of these cases. Just the

fact that there is a certain relationship of authority

between S and A is not sufficient as it is not impossible to

advise someone of less status than oneself to do something.

A possible account of the difference if the reason (motiva­

tion) S intends to provide A for leaving the room. If S

orders A then he intends, roughly, that the institutional

authority he has be part of A's reason for complying. (This

is only an example, and not intended as a final reasons

clause.) If S advises A then S intends, perhaps, that the

fact that S believes a certain action is best for A be part

of A's reason or that S intends A believe S believes the

action is best be part of the reason.

23

But need there be these sorts of reasons that S

intends A to have? Let me make two remarks here in this

regard-

In the absence of any clear counter-examples the

burden of proof is on the objector to specify how ordering

and advising might be better distinguished. It is not the

presence of the performative prefix as a speaker can use the

prefix and not, say, be ordering. It is not, in the case of

ordering, just the authority relationship between the per­

sons. Nor will the tone of voice separate the cases. A

speaker can advise in an insistent tone. "She is not good

enough for you Jones, and so, damn it, give her up!" To

insist that these features are really the features which

identify the acts after being shown cases where they do not

is akin to insisting that what is sufficient to characterize

water is that it comes out of a tap or has fish swimming in

it or falls from the skies. When water is found that does

not fit these descriptions or items are found that fit the

descriptions which are not water it is said that sometimes

these things are sufficient. The point is that none of

these are what make water, water. Likewise, it is not the

presence of the performative prefix or the authority rela­

tionship between the speaker and the hearer or the tone of

voice that makes ordering, ordering.

The way to show that there are not such reasons

other than providing an alternative account which better

24

handles the data is to produce clear counter-examples to the

cases. At this point we are still operating with Schiffer's

intended effect, that A actually believe something or form

the intention to do some particular thing. The claim, is

that it is reasonable to suppose that the reasons for be­

lieving or doing separate some illocutionary acts, since the

intended effects might be the same. As I intend to modify

the intended effect, and the way in which the reasons relate

to believing or intention I will postpone temporarily a dis­

cussion of counter-examples. After the reasons clause has

been formulated I will take up the. issue of whether such

reasons actually do individuate acts. I will continue in

Chapter II, after these initial arguments, to further

motivate reasons clauses as necessary in other illocutionary

acts.

Including a particular reaons clause within the

analysis of illocutionary acts will be shown to distinguish

the cases along the correct lines. There may be other

features which individuate illocutionary acts, but if the

use of language is partly a matter of getting people to do

or believe certain things, it is reasonable to suppose that

some uses would divide along lines distinguishing different

things to be done, and different reasons for believing or

doing them.

Reasonableness is not, however, a guarantor of

truth. I would like to turn first to the imperative class

25

of illocutionary acts, arguing that it is not necessary as

Schiffer's (1972) analysis requires that S intend to cause

A to do something. Since examples like the ones I will con­

sider here have arisen in other contexts I will not spend

much time discussing them. I will use them mainly as

vehicles for introducing several potential replies which

merit attention, and then as test cases against which to

examine alternative hypotheses.

The counter-examples are the result of a widely held

restriction on intending to do something (Grice, 1957;

Schiffer, 1972). One cannot, with one exception to be dis­

cussed later, intend to cause something to happen as a

result of performing some action unless one believes that

the outcome is a real possibility as an outcome of that

action. This restriction, besides having independent moti­

vation which the sources cited above provide, is one way of

avoiding the consequence that a person could say anything,

and mean, thereby, anything. Otherwise, for example, S

could say to A, "Fats Domino is in town," and without any

stage setting mean that the Dodgers will win the pennant, as

it is possible someone might understand him as meaning this.

The two cases I will discuss are ordering and ad­

vising. It is part of Schiffer's analysis of each that S

meant A was to ip.

In uttering x S was advising A to \p if and only if by uttering x S meant A was to (should) ip and S

26

intended it to be mutually known* between him and A that he intended part of A's reason for ijj-ing be that it was in A's interest to ip.

In uttering x S was ordering A to ip if and only if by uttering x S meant that A was to ip and S in­tended it to be mutually known* between him and that he intended part of A's reason for ip-ing (Schiffer, 1972, pp. 102, 103).

The parenthetical "should" in the definition need not con­

cern us. If it allows S not to intend to cause A to ip, but

only to believe some action is required, then it is an

assertive illocutionary act, and the comments to be made

about that case will apply. If it requires A to ip, the

criticism applies.

I will begin by proposing a counter-example to

Schiffer's analysis of advising; it is, perhaps, the more

controversial of the two counter-examples. Suppose S knows

that a friend, A, will not give up drinking, and so cannot

in any clear sense intend an utterance of his to cause A

to give up drinking. It does not seem wrong to say that S

can nonetheless advise him to give up drinking. S's beliefs

about the likelihood of what he says affecting A does not

determine whether what S does either is or fails to be

advising.

Likewise S can order A to do something S does not

intend for A to do. (I use "intend A to ip" throughout as a

shorthand for Schiffer's "intend to cause A to There

is a not uncommon ploy in the army of ordering a trouble­

maker to do something he is likely to refuse to do so that

27

he is subject to the consequences of failing to obey an

order. It is no defense at the court martial to object,

"They knew I wouldn't comply, so they couldn't have ordered

me." S can order with the primary intention of carrying out

what he thinks his duty or because he was himself ordered to

order a third party. Belief or even interest in the hearer's

response is not important. Indeed, there is no oddity, as

there should be if Schiffer's analysis is correct, in saying

that a person disregarded our advice or disregarded the

order as we knew he would.

0 In case it is felt that "order" is being used in a

technical sense because of the rigidly circumscribed pro­

cedures within the army for what an order is, other examples

could be found. S might order a belligerent employee to do

something convinced that he won't in order to show the

other employees that he is not afraid of the man.

Schiffer has considered criticisms of his defini­

tion of speaker meaning which maintain such intentions are

not necessary for S to have meant something. Although his

points are intended to be about speaker meaning they could

be redirected toward the counter-examples of his analyses

of various illocutionary acts. In addition, several of his

replies instantiate forms of philosophical argument which

may be of somewhat questionable validity. For these

reasons I think they merit attention.

28

The first line of reply I wish to discuss arises

out of Schiffer's response to counter-suggestible cases.

He notes that although in some deviant cases S may not have

the intentions the analysis says he has, in the standard or

normal cases S must intend for A to i|i. The non-normal cases

are parasitic on the cases where the appropriate intentions

are present; there would be none of the deviant uses unless

there were normal ones (Schiffer, 1972, p. 71).

Apparently the reader is supposed to conclude that

there are, hence, reasons why an analysis need handle only

the normal cases. Unless some such inference is intended

the observation about normal cases having the requisite

features is beside the point. Schiffer gives no reason why

the observation about normal cases justifies the inference,

thus meeting the counter-examples to his analysis. Ad­

mittedly reference to normal cases is commonplace in con­

temporary philosophy, but that it is commonplace does not

dispel suspicion that it serves only to conceal the fact

that it is not a response. Although it is unclear, then,

exactly how the point about normal cases is to be taken,

there are three criticisms I would like to direct at it.

1. Even allowing that normal cases of ordering must be

as Schiffer describes, it does not follow that S's

having a primary intention to cause A to i|) should be

part of the analysis of ordering. If intending A to

4> is necessary, it would seem, analogously, that in

29

the standard or normal case actual i|j-ing by A is

required. Unless there were some cases where A \p-ed

we would probably not have the practice of ordering.

Certainly if some cases of having the requisite

intentions are necessary for the continuation of the

practice, so are some cases of actual compliance.

But no one would suggest that this standard outcome,

which is necessary for either the starting or the

continuation of the practice, is part of the anal­

ysis of ordering. Consider a more or less parallel

case from epistemology. In perception it may be

standardly or normally the case that we see ordinary

physical objects. We cannot on this ground accept

an analysis of perception which said that perceptions

of after-images or rainbows did not need to be

accounted for as normally physical objects are what

are seen.

Need there even be the normal cases? People might

come to develop a sadistic turn to their person­

alities. They order people to do things generally

hoping they won't comply. It is not clear why this

would not be ordering. It may even be (although I

don't think this is so) that there might have had to

be normal cases at one time for the concept to

emerge. It is not clear that such a requirement is

30

other than a contingent one (see the following

paragraph in this regard).

3. Finally, what constitutes a standard or normal case?

To say it is the one where certain beliefs or in­

tentions are present without some argument that they

are an essential part of the definition in question

is circular. The standard case is not the most

commonly occurring case, or if it is this may be

entirely for accidental reasons. One answer is that

the conditions for the standard or normal case are

those conditions which are required in the situation

in which the concept is learned. There are at least

four difficulties with this reply. First, it is not

clear that there is only one type of situation in

which we learn to use the concept in question, or

that conditions in different situations are the same.

Second, language learning situations are most likely

those where the requisite intentions are absent. A

parent could teach a paralyzed child imperatival

language without having an intention that the child

do any of the things the examples used would make it

appear he was commanded to do. Third, even sup­

posing some conditions are necessary, how are the

"conceptual" conditions of the situation to be

separated from those which are only causally neces­

sary, for example that the child have certain

31

neurological functions. Fourth, in any event an

argument is needed as to why conditions necessary

for learning how to use a concept are crucially

related to the analysis of that concept.

Schiffer's (1972, pp. 68-69) other potential

response to the counter-examples is along the lines of his

response to the lingerie example.

Customer: "Where is lingerie?"

Clerk: "Lingerie is on the fifth floor."

It is not clear in such a case there need be any

intention on the clerk's part to get the customer to believe

that lingerie is on the fifth floor, yet it seems clear that

he meant that it was on the fifth floor. In the discussion

which accompanies this example Schiffer claims that this is

analogous to the case where the President slips at a news

conference, and says something he did not intend to say. He

says that the President, in this case, had a momentary in­

tention to get someone to believe something. So, in the

lingerie case it is suggested that the clerk had a momentary

intention to induce a belief in the customer. If this line

of reasoning can be applied to cases of ordering or advising

where the speaker does not care whether a person complies,

then in these cases the speaker had a momentary intention to

get someone to do something. For example, when an officer

32

routinely orders his company to attention at morning in­

spection he has a momentary intention.

Two objections can be made to this appeal to momen­

tary intentions. First, in those cases where S knows A

won't comply at the time he produces the utterance the

presence of momentary intentions would mean that at the same

time the speaker intended to get the person to do something

(momentarily), and also did not intend to get him to do

something as surely his beliefs about the possible efficacy

of his utterance don't disappear as he begins the utterance

and then reappear on its completion. Such conflicts re­

garding what we intend may be possible with illocutionary

acts, but it certainly divorces them from the rest of our

intentional actions. I don't pull the trigger intending

thereby under the same description both to kill someone and

not to kill him in the same sense of "intending to kill

him. "

Secondly, there is no reason to believe that other

cases have the features which lead to the description of the

verbal slip as a case where a momentary intention was

present. Admitting, which seems possible, that Schiffer

has correctly described the verbal slip case, the question

is whether there is any reason to suppose this explanation

could be extended to the lingerie example or even further

to the ordering example. There is not in either of these

latter situations someone saying something which he

33

previously did not intend or wish to say. The clerk was not

planning before and after his utterance not to reveal the

location of the lingerie. We need some reason to suppose

these cases are in some way like the verbal slip case. The

factors which incline us toward Schiffer's explanation of

the one are. missing from the other. This is not to say that

the action was not an intentional one, that S did not intend

to say something about the location of the lingerie. It is

just that the description of the case does not reveal that

he had an intention, momentary or otherwise, to get someone

to believe something. The case differs from cases where it

might be reasonable to suppose this. The examples with

illocutionary acts share this resistance to the description.

Thus far the arguments have turned on what Schiffer

sees as a feature of intentions, namely, a person cannot

intend to cause something by performing some action unless

he believes that the outcome is a real possibility as a

result of that action. Otherwise a person could utter any

sentence and without any stage setting mean, thereby, any­

thing. The restriction on intentions rules out these kinds

of cases without invoking, as with, say, Searle (1969),

conventional features of the utterance type.

There is a question as to whether such restrictions

disallow too much. Schiffer waivers on this point when he

considers a case which seems to question the restriction.

A man is on the top floor of a burning building and jumps.

34

It may be that he considered it extremely unlikely that he

would survive the jump, yet it seems possible to describe

his action by saying that he jumped intending to save his

life. The dilemma appears serious. If this description of

the jump is accepted, the door has been opened to numerous

counter-examples to this type of definition of speaker

meaning where to mean something by an utterance is essen­

tially a matter of intending in a certain way to achieve a

certain effect. To disallow it is to rule out what seems

like a perfectly adequate description of the situation.

Opting for the "fire" account of intentions would

require weakening the notion of intending to make it some­

thing like wanting or hoping. It should be clear by now,

however, that S could mean for A to i|) without hoping or

wanting A to ip. Hence, versions of all the old counter­

examples in addition to new ones would appear.

Short of saying that the man in the building had

no such intention how might the situation be described to

better explicate what is going on? Suppose there had been

no fire, but the man jumped out of the window realizing that

he probably wouldn't survive. It would not matter if he had

calculated the odds and they were exactly what they were in

the case where there was a fire. With the fire the jumping

was necessary or offered a better chance of survival than

staying in the building. So, perhaps the best description

of the situation is that he intended to perform the action

35

that was necessary for survival or offered a better chance

of survival.

I argued with respect to the imperative illocutionary

acts a speaker could perform the act without having meant,

on Schiffer's analysis of speaker meaning, that someone was

to do something. Turning to the assertive cases I will

argue very briefly that when S performs an illocutionary act

it is not necessary that he intend to produce in anyone the

belief that P. Again, there are three different kinds of

cases. Schiffer (1972, pp. 99-102) offers examples of P,

r(t), and P and r(t)—identifiable acts.

In uttering x S was objecting to A's claim that q (and objecting to P) if and only if S meant by uttering x that the fact that P constitutes a good reason for thinking A's claim that q is false.

In uttering x S was reporting that P if and only f by uttering x S meant that P and S intended it to be mutual knowledge* between him and A that he intended part of A's reason for believing that P to be that S uttered x intending A to think that S believed (he knew) that P.

In uttering x S was estimating that the value of so-and-so was such-and-such if and only if by uttering x S meant that the value of so-and-so was such-and-such and S intended it to be mutual knowledge* between him and A that he intended part of A's reason for believing the value of so-and-so was approximately such-and-such to be that S uttered x intending A to think, that, as a result of having examined so-and-so and of having applied to his findings certain general knowledge . . . S believed (he knew) that the value of so-and-so was such-and-such.

Unfortunately the complexity of Schiffer's defini­

tions here obscures the fact that he is allowing different

36

things as substituends for P. I will change the notation in

hopes of clarifying the definitions. Suppose someone says

that it is a good day for an outing, and S objects that it

is raining. Now when S objects that P (objects that it is

raining) Schiffer does not say S meant that P (meant that it

was raining) where what is meant is identical with what is

objected. Rather, what Schiffer says S meant is that P is a

reason to think another claim is false. It is the proposi­

tion that expresses this more complex form of beliefs that

was meant and by which the act is identified. Now, in the

reporting case where S reports that P (reports that it is

raining) Schiffer says S meant P (meant that it was raining).

Here the P that was meant is the same P that was reported.

For clarity I will refer to propositions like it is

raining, where this is what is reported or objected by "P"

and refer to the more complex form of beliefs that include

P, for example it is raining and that is a good reason to

think some other claim false, by "PP." Then, if S

answered, affirmed, described, or objected that P, Schiffer

does not say that S meant that P, and by this P we can, in

part, identify the act performed. Rather, he says S meant

that FP. It is an open question whether S also meant P in

these cases although it seems that if he meant FP he also

meant P.

Clarifying the terminology does not, however, save

the program. It is not necessary either that S meant some

37

P or some FP in performing an assertive illocutionary act.

The argument to show that a speaker can, for example, report

that P without having meant that P will essentially be the

same argument used in the imperative cases. I will proceed

only in broad outline as the details should be clear.

Knowing A's devotion to some politician S may be

certain that A will not under any circumstances believe that

he is a crook, but still feels it is his duty to report it

to A. He may not even care if he is believed. He would not

have the requisite intentions, but it does seem he has

reported.

Or, someone may predict that P because he wants his

prediction on the record in order to be able to tell someone

"I told you so." Predicting to be able to say "I told you

so," seems almost to require that the speaker expect not to

be believed. Any number of predictions are made in the face

of overwhelming scepticism. They do not cease to be predic­

tions on that score.

Henry Byerly (1975) has pointed out, in written

comments, that what such cases tend to show is the irrele­

vance of the "use" of an expression or its intended use to

the meaning of that expression. I take it that it is not

because something can be used in such and such a way that

it is a report. Rather because it is a report, and has

certain characteristics there are a wide variety of points

or purposes a speaker might have when using it. See the

38

section, "A Taxonomy of Illocutionary Acts" for more on the

point or purpose of an utterance.

Those acts where it is claimed that FP is meant

rather than just P are more difficult. (Since for Schiffer

in every illocutionary act S either meant that P or FP from

here on I will use [F]P for both. Context should make clear

which is needed.) If S objected that P what must be shown

is that S did not mean (did not intend to create the belief)

that P is a reason to think some other claim is false.

The statement of (F)P is ambiguous. S can intend A

think P is a reason to think q false in one sense even

though he is not now maintaining that P is true and q is

false. This would be true if all S were doing was informing

A of the logical relationship of P and q or the empirical

relationship between the events mentioned in P and q. The

police are discussing a case, and agree that if the suspect

is not an expert on safes he could not have been -$;he one who

broke into the bank vault. But clearly whoever makes this

point has not objected that the suspect is not an expert on

safes. In this sense that P is a reason to think Q false is

a precondition of successfully objecting that P rather than

an analysis of what it is to object that P. What Schiffer

must mean is that P is (probably) true, and hence q is false.

But meaning this is not necessary in order for some­

one to object. S need not intend A believe P, and believe

then that q is false in order to object. The police believe

39

that S was near the scene of a crime. S objects that he was

in Washington reading a paper at the APA meeting that day.

The suspect objects even though he is certain the police

believe him guilty, and so will not believe he was in

Washington reading a paper.

An attack can also be mounted on the role the

program gives the reason A is intended to have for his

belief. If S estimated it is not necessary that he intended

A's reason for believing the value of so-and-so was such

and such be that (A believed) S uttered x intending A to

think that S as a result of having examined, believed the

value was such-and-such. I do not wish to challenge the

view that if someone estimated that an examination of the

available data is in some way involved. This is possibly

what separates estimating from guessing. The objection is

to the status the reason has in Schiffer's analysis. The

requirement placed on belief also infects the reasons

clause. S and A have a bet where each of them is to

estimate the number of beans in a jar. Suppose A always

enters these kinds of contests, always wins them, and that

S is very bad at it. Suppose each of them knows these facts

about the other. Now A would not take S's estimate as a

reason for believing the number of beans in the jar was

such-and-such, and S would know this. It would follow that

S could not estimate the number of beans in the jar to A

because he could not intend A's reason for believing be that

40

S uttered x intending A to believe that so many beans are in

the jar because S using general knowledge and having looked

at the available evidence determined there were approximately

that many. But S can estimate the number of beans in the

jar to A. Similar points can be made about other acts that

are identifiable entirely, or in part, by the reasons, the

(F)r, A is to have.

Modifications of the Analysis

What is required is some modification of the in­

tended effect to avoid these kinds of counter instances.

Grice's (1969) final intended effect in his analysis of

speaker meaning is not that A believe that P, but that A

believe that S believe that P. Harman (1974) has suggested

that for an account of speaker meaning this is more

promising than Schiffer's emendation. However, when this

analysis of speaker meaning is used as part of an analysis

of illocutionary acts it is clear that this change is not

sufficient to avoid the examples that felled Schiffer's

account. Holmes believes Moriarty is guilty of a crime, and

Moriarty knows Holmes will not believe his claims of

innocence. So, Holmes believes Moriarty believes himself to

be guilty, and Moriarty knows this. Moriarty could still

object or state or protest that he is innocent even though

he does not intend Holmes to believe that he, Moriarty,

believes himself to be innocent. It appears that an

41

individual can perform an illocutionary act for anyone of a

wide variety of reasons which do not require any intention

to create either beliefs or intentions in the audience. If

the view that to perform an illocutionary act is to mean

something is correct, then what it is that a speaker does

when he means something must be something that he does even

when he does not intend to induce a belief or intention by

performing an illocutionary act.

The following weakening of the intended effect

satisfies that requirement. What S intends (to cause) is

not A's having a belief or A's forming an intention to do

something, but merely A's entertaining the thought that (F)P

is true or entertain the thought that A should (Harnish,

1975). The intended effect of an illocutionary act qua

illocutionary act is quite minimal. What is sought is the

minimal sense of communicating P. In Chapter IV it will be

suggested that one way of viewing this will be in terms of

some notion of representation. Later in Chapter II it will

further be distinguished from theiillocutionary act of

suggesting something. For now it can be viewed as having a

certain proposition before one's mind or understanding the

content of an utterance.

Then, S could report that Jones is drinking without

intending anyone believe that Jones is drinking. S intends

that A entertain the thought that it is true that Jones is

drinking. In those cases where S intends A actually believe

what he has reported that S so intends does not follow from

the fact that S reported, but from further facts about S.

In such a case reporting would seem closely tied with the

intention to induce a belief because it was the means that

S used to induce a belief. Still there is nothing in the

act of reporting itself that involves the attempt to create

a belief.

Intending that A entertain the thought that (F)P is

not, of course, sufficient for reporting. First, there is

a reasons clause in the definition of the act that must be

satisfied as it does seem that for many acts knowing the

character of the reasons A is intended to have is necessary

for identifying the act that was performed. And, the change

in the intended effect will not be without effects in the

reasons clause of the analysis. If the reasons clause in

the definition of speaker meaning is not changed, the

analysis would say that S intends A to have the reason

characterized by some specific (F)r for entertaining the

thought that P is true. Suppose that the reason A is to

have when S reports that P is something to the effect that

A believe S has evidence for P. Then S intends part of A's

reason for entertaining the thought that P be that A be­

lieves S has evidence for P. But clearly this is not what

is to cause A to entertain the thought. The connection

between hearing x and entertaining the thought that P is not

mediated by reasons in the way the connection between

43

hearing x and believing that P is. People do not have

reasons for entertaining thoughts in the same way they need

to have reasons for believing something. A may entertain

the thought even though he fails to recognize the form of

the reason he is intended to have. Yet it has been and will

be further argued that act-type identification for a great

number of illocutionary acts requires knowing the character

of the reason.

The following is a first attempt at a reasons clause

with the weakened intended effect. S intends it be mutually

known* that he intended that if A actually believe that (F)P

then (F)r be part of his reason for believing (F)P. Even

though S may not intend A believe that (F)P, if he performs

an (F)r-identifiable act, then he intends that there is some

reason which fits the above description. Reasons remain

where they belong; reasons are essentially reasons for

believing or reasons for doing.

Before proceeding there is one objection to this

way of stating the reasons which should be considered. If

the conditional is read as a material conditional, then

whenever S does not intend to create a belief the condi­

tional would be true. In this case the reasons clause for

every illocutionary act would be satisfied. Whenever S did

not intend to create a belief there would be no way of

determining which illocutionary act S performed. If it

held, this objection would indeed be serious, but it fails

because the conditional is within the scope of S's inten­

tion. That it is false and S intends to induce a belief

does not effect the truth of the claim that S intends that

if A believes that P, then such and such be part of his

reason.

It would be appropriate here to see if there might

be problems with the view that reasons separate the cases.

J. L. Cowan (1975) has suggested that examples like those

used against Schiffer can be used to defeat this analysis.

Suppose S is ordering someone to do something only because

he was told to or in other circumstances where he does not

really want his audience to comply. I will admit that S

did order even though he did not intend the audience to

comply with the order, but will not wishing them to comply

mean he did not have the intention about any reasons they

might have?

Suppose S were to consider what if A did really go

ahead and do it. What would S know about A's performing the

action in this case? S would know that if A actually did

ip, then A would have as a reason for ijj-ing that S had

ordered A. S would expect that a certain reason is avail­

able for A if A actually does ty. If A knows what an order

is, then A will know that he has a certain reason to in

virtue of being ordered. If he supposes that S knows what

an order is A will suppose that S knows A knows this. Like­

wise, S would know A would know this. I think it could be

45

shown to be mutually known*. Suppose we are now talking

about a particular order. In this one case how is it that

it is mutually known* that if A ip, then he has a certain

reason to i|>. It is mutually known* in this particular case

because S ordered. But S knew by saying this that he would

bring about the state of affairs that this is mutually

known*. It is not just that it is reasonable to expect

that this would happen. S knew it.

The question at this point is whether if S knows

that something will occur as the outcome of his action

whether he can, in any sense, be said to intend that out­

come. This is why the proviso about it being more than just

reasonable to expect is important. In these cases S may not

intend because even though it is reasonable to expect him to

know it he might not have. But, S did know that it would be

mutually known* that such and such. S knew that it was his

utterance that made it or would make it mutually known*.

The issue is not one that is restricted to

linguistic contexts. It has arisen in other contexts, for

example, with the question of abortion. Suppose that a

doctor must crush the skull of the fetus in order to save

the life of the mother. What the doctor intends to do is to

save the mother's life. Some wish to say that he did not

intend to kill the fetus he only intended to save the

mother's life. The destruction of the fetus is an expected,

but not intended, consequence of the action. But, it does

46

seem reasonable to suppose that the doctor intentionally

crushed the skull. Crushing the skull is the action he per­

formed, not some result he expected to result from his

saving the mother's life. He would, if asked if he in­

tended to crush the skull, likely say yes. Under this

description it seems not unreasonable to maintain that in

crushing the skull, knowing that this would kill the fetus,

that if he intended to crush the skull he intended to kill

the fetus.

It might seem that it is not unreasonable to main­

tain on similar grounds, that S did intend the known con­

sequences of his ordering. In the cases where S did not

intend actual compliance, S still intends that should A

comply he would have a certain reason for complying.

Other cases might be proposed, though, to show that

a speaker does not intend every known consequence of his

action. S is an engineer on a train coming to a divide in

the track. On one of the tracks are five people who will be

unable to get out of the way of the train. On the other

track there is one person who will not be able to get out of

the way of the train. The brakes are gone, and unable to

stop the engineer turns onto the track where there is only

one person, knowing that the train will run over and kill

that person. However, it might be felt that it is incorrect

to say that the engineer intended to kill the person. He

knew that it would happen, but he did not intend to kill

him. The objection is that although it may sometimes be

correct to say that the person intended to do what he knew

would be the outcome of doing something this is not always

so. Even should it be correct to say that if the doctor

intended to crush the skull knowing this would kill the

fetus then he intended to kill the fetus it is not obvious

that even though he intended to turn down one track rather

than the other knowing this would kill the person on the

track he did intend to kill the person. What reason is

there to suppose in the example given with illocutionary

acts that they are like the former rather than the latter

case?

I have three points to make in this respect. I do

not know how adequately they meet the challenge, but if they

do not get the ball back into the other court they at least

get it in the vicinity of the net. First, consider the

following two cases. S wishes to join a terrorist gang, and

to prove his loyalty he must throw a hand grenade into one

of two houses which have been pointed out to him. In one

house are five people. In the other is one (who has

miraculously just survived being run over by a train, but

that is no matter). The people are innocent civilians, so

S throws the grenade into the house where there is only one

person knowing that it will kill the person. It does not strike

me as implausible to say that he intended to kill the person,

he intended to kill one rather than five. I suspect he could

48

be convicted of first degree murder which in most juris­

dictions requires the intention to kill.

Suppose for the other case that S is rolling down a

hill unable to stop with a live grenade in his hand which he

cannot get rid of. However, by shifting his weight S can

roll one way or the other. If he shifts his weight one way

he will roll into a house with five people in it. If he

shifts his weight the other he will roll into a house with

only one person in it. There is no doubt at all in his mind

that the persons in the house he rolls into will be killed.

He rolls into the house with the one person. Even though he

intended to roll one way rather than the other it is not

correct, perhaps, to say that he intended to kill the

person in the house, but he did know that by rolling into

the house the person would be killed. The roller would not

be guilty of first degree murder.

This offers one possible way of separating the two

cases. The train case is like the rolling case. The

temptation to refrain from calling them cases where someone

was intentionally killed may have a moral basis. The person

could not help but to kill somebody, no matter what he did.

This may not be correct, but it would separate the two

grenade cases.

The second point is that these cases are complicated

further by various things which a speaker might mean when he

uses "intend." Sometimes it is used equivalently to

49

"hoping" or "trying" or "expecting" or even "wanting."

Other times a speaker means by someone's "intention" his

conscious plan. The variation in uses was pointed out to

me by Henry Byerly (1975). It might be that in some cases

where it is being claimed that a person did not intend to do

such and such what is being said is that in one of these

particular ways he did not intend. In different circum­

stances the various ways in which it is relevant or in which

we wish to assess actions will make one use of "intend" the

obvious one which people would be concerned with, and hence

there is an obvious answer which seems forced on us. In the

case where people were killed by someone our concern is

whether the person was trying to kill them or whether he

could help it. He could not, so we say he did not intend to.

This would not show that there is not another use of

"intend" in which he did, but it might be felt that in the

circumstances it is not relevant. It may be that the

circumstances even dictate which of the uses is permissible,

so we say he intended to although he did not really want to.

In other circumstances asking whether a person intended to

is only to ask whether he wanted to.

A third response is possible to the train case. It

involves an appeal to the need for explaining what occurred.

He could not help but to turn down one track or the other.

Surely he did something intentional in the circumstances,

say push a lever right rather than left. This would send

50

him down the left track rather than the right track.

Suppose someone were to ask why did he push the lever one

way rather than another. The reasonable answer seems to be

because he knew by doing that he would kill one person

rather than five. The engineer decided if he had to kill

someone to kill as few as possible. But it may be that

there is a sense in which to decide to do something, and to

perform some action which will bring about what one has

decided to do is to intend to do what one has decided to do.

When the intention is put in the form of a disjunction, and

we realize there are a number of things that might be meant

by intending to do something then it is not as unreasonable

to say he intended to kill one person rather than five.

That is why he went down the left-hand track. The appeal to

intentions is justified because it explains why he performed

some other action, namely pulling a certain lever. Or, if

it is said that the intention to go down one track rather

than the other explains his pulling the lever, that he

intended to kill one person rather than five would explain

why he intended to go down one track rather than another.

At a minimum an alternative explanation of why he pushed the

lever one way rather than another is needed. Notice in

this respect that he would not say that he wanted to kill

the person. If it was suggested that he did he might say,

"I didn't want to kill him, but I couldn't help it." How­

ever, when asked a different question such as why he went

51

down a certain track he might very well say, "If I had to

kill someone, I wanted to kill one rather than five."

These answers may not ultimately stand, but I do

think they merit some response. While responses are brooded

over the thesis that S intended the known consequences of

his ordering can take a few more breaths.

One other case of ordering without the requisite

intentions has been proposed, but my intuitions on whether

S ordered in this case are just not clear. Suppose that S

wished to show someone that a group did not understand

English. S says to them, "Attention!" They do nothing. S

knew that they would not understand him, and so could not

have intended that they have any sort of a reason for

complying. I am just not sure whether S ordered in this

case. But, should it turn out that it is, the responses of

Schiffer (1972, pp. 72-80) noted earlier provide a number of

proposals for revising the definition of speaker meaning

without seriously changing its form which would handle such

cases.

Then, the schema for identifying assertive illocu-

tionary acts where Schiffer's analysis is the model would

be more or less as follows.

In uttering x S was 0-ing that P if and only if S

meant that (F)P, and S intended it be mutually known* that

(he intended) by uttering x that if A believe that (F)P

then part of his reason be (F)r.

52

The intended effect in the analysis of speaker

meaning will be that A entertain the thought that (F)P is

true. This does not completely determine the form of the

analysis of "S meant that P."

A Schiffer-like analysis with its recourse to mutual

knowledge* has a number of advantages, among them that it

gets all of the speaker's intentions in the open. It has

the disadvantage of requiring that people have an infinite

amount of knowledge; it is not obviously true that they do.

Recourse to mutual knowledge* is not, of course,

the only way of getting the intentions into the open.

Grice's (1957) original formulation used a reflexive in­

tention, and it is not obvious with the weakened effect it

will not survive the criticisms directed at it and at other

formulations. The statement of the reflexive intention

would be roughly that S intended to produce the relevant

response, in this case A's entertaining the thought that P

is true, in virtue of A's recognizing that very intention to

produce the response. That is, S intended A to do some­

thing via, in part, recognizing that S had the intention

that he do that thing. Despite appearances, there is

nothing paradoxical about having such an intention. Certain

other mental states are reflexive in just this sense.

Suppose that S is a tyrannical despot with a

grandiose view of his own importance. Then, S might want A

to and have as his reason for i^-ing just the fact that S

53

wants A to . There may be questions about why such a want

should arise, but that does not mean that such a want

arising is not possible or that S's having such a want is

unintelligible. There may be other reasons for the origin

of S's wanting A to ip. S may, for instance, find the

prospect of A's ijj-ing quite amusing. This has no effect on

what type of reason S intends A have for ip-ing.

Another possible criticism of the reflexive inten­

tion involves the nature of intentions. Intentions are not

the sort of thing that can be gotten at directly as can S's

words or gestures. Usually we know what a person intended

because of certain other things about him or about the

situation or because of what we have learned. But if

intentions, at least the intentions of others, are not

always immediately available to us how could A recognize S

intended A to ip by recognizing S's intention? The answer is

clearly in the same way that S recognizes any other inten­

tions. That S's intention has this reflexive character does

not mean that it is not determined in the same way that any

other intention is. There is no more difficulty here than

for any other intention where S intends A do something via

recognizing some mental state of S. S may intend that A ip

by realizing S wants him to. Wants share this feature of

not being gotten at directly. But, it is not impossible to

determine in this case what S intends.

54

In the assertive cases of speaker meaning what S

does when he means that P would be something like the

following: S meant that P by uttering x if and only if S

uttered x intending for A to entertain the thought that

P/(F)r is true partly in virtue of recognizing that S so

intended by uttering x. (The reason, (F)r, is to be under­

stood as above.) If Schiffer's notion of a state of affairs

E is made use of in the reflexive formulation we get the

following.

S meant that P by uttering x if and only if S

uttered x intending thereby to realize E which is (intended

to be) sufficient for A to recognize that by uttering x S

intended for A to entertain the thought that P/(F)r is true

partly in virtue of recognizing that S so intended. The

last two cases are not parallel as they now stand because

in the latter what E would allow A to recognize is that S

intended A entertain some thought, while in the former, S

is intended to entertain the thought via recognizing that

he is intended to entertain the thought. That is one

appears to say the object of the reflexive intention is that

A entertain the thought that P, and this is what S intends.

The other says there is an intention to recognize the re­

flexive intention. Opting for just one intention, the

reflexive intention, is simpler, and seems to lead to no

real problems that the other escapes.

55

The reflexive intention is the communication in­

tention. Uttering something with such an intention, in the

wide sense of uttering something, is both necessary and

sufficient for meaning. If further inferences are intended

beyond the reflexive intention, then this intended effect is

not a part of the communication intention, and so not a part

of what is meant. Likewise, intentions that S recognize

what thought he was intended to have other than by recog­

nizing he was so intended are not sufficient for meaning.

This is not to say that in meaning something S may

not intend A go through some process of reasoning. However,

at the point where the inferences stop and A gets to the

thought that P, S must intend that he entertain this

thought. This suggestion is due to Harnish (1975). Other­

wise he did not mean it. He may have implied it or even

intended without implying that A figure it out, but he did

not mean it.

Viewing the communication intention in this way,

then if S said, "If it snows Jones will freeze, and it's

going to snow," in addition to meaning what his words meant

S may have meant that Jones will freeze. (There may be a

certain type difficulty, Byerly [1975] has pointed out, in

saying that what a speaker meant is what the words meant for

it is not clear they mean something in the same way. I will

deal with expression meaning and the relation between

speaker meaning and expression meaning in Chapter III. For

56

now what I mean by such phrases is that a specification of

what the speaker meant would not differ from a specification

of what the words meant.) In this case S intended for A to

entertain the thought that Jones will freeze, and at the

point he had this thought he was to recognize that S in­

tended him to have this thought via seeing what S intended.

Principles of logic did some of the work of getting him to

the thought, but if he was not intended to recognize that

this is what S was intending that he think, and do it by

seeing that S so intended then S did not mean that Jones

will freeze. Certain not too serious difficulties can be

raised at this point by cases which are audienceless or

where the remark is not directed at any audience. Schiffer

(1972, pp. 72-80) offers a number of ways of dealing with

these cases.

Certainly S will not mean P whenever it is reason­

able to expect someone to infer P from what was said, not

even in every case where A was expected to make the in­

ference. In fact, Grice (1969) kept modifying his analysis

because of cases that had this characteristic. But where

the requisite intentions are present S did mean that P.

Notice in this case not only did S mean that Jones will

freeze, but also would be described as meaning that if it

snows Jones will freeze, and it will snow. It does not seem

implausible to maintain that S meant both what the audience

was to infer, and what the inference was to be based on. It

57

would just be incorrect to maintain that S did not mean both

of them. And both satisfy the requirements for the third

definition. The analysis seems to get the facts of the case

correct.

But in describing this case correctly the analysis

is in danger of running aground on cases of irony or sar­

casm. Suppose S says, "Jones is a great philosopher," but

everyone knows that S thinks Jones is a hack philosopher

who should be making a living shining shoes. If S was being

sarcastic he did not mean what his words meant, what the

expression, "Jones is a great philosopher," meant. S meant

that Jones is a poor philosopher.

S may have intended his audience to reason in the

following manner. S said that Jones is a great philosopher,

but he does not believe that, and he knows A knows he does

not believe this. Either he did not mean Jones was a great

philosopher or he was up to something because one does not

usually say what everyone knows he believes to be false.

(He might have been up to something like hoping word would

get back to Jones about what he said as Jones was voting on

whether S should be tenured.) But there appears to be

nothing S can be up to. So, S was being sarcastic, and

meant that Jones is not a good philosopher. On the analysis

of speaker meaning S could be said to have meant that Jones

was not a good philosopher. Arguably at the point where he

entertains the thought he does it via recognizing he was

58

intended to entertain the thought. However, it seems that

the intermediate step in the inference where A entertains

the thought that Jones is a great philosopher also satisfies

the requirement for meaning. A must entertain the thought,

and it appears, entertain in because S intends him to in

order that he may make further inferences. Then, S both

meant Jones is a great philospher and that Jones is not a

good philosopher. S did not, however, mean the former.

The problem is to escape the consequence that S

meant that Jones is a great philosopher, but allow that in

the previous case that S did mean the elements upon which

the inference was based, that if it snov/s Jones will freeze,

and that it will snow. Getting between the two cases does

not appear an easy matter. Schiffer (1972) and Grice (1957)

have an easy account of the difference. S did not intend A

believe that Jones was great, and so did not mean it. But,

as we have seen, the belief requirement is too strong.

However, S does not satisfy the requirements of the

definition of meaning that P as far as meaning that Jones

is a great philosopher is concerned. This is because S does

not intend that if S believes that P, then part of his

reason be of a form characterizable by some (F)r.

With respect to this case the following should be

noted. Some cases of sarcasm will turn on the speaker's

intonation. It is likely in these cases that determining

what the speaker meant does not require any inference on the

59

part of the audience. If a sentence of constative gram­

matical form is uttered with a rising intonation contour the

speaker has asked a question, and has not made a statement.

It is very doubtful that in these cases the speaker intends

or that the audience actually does first entertain the

thought associated with the assertion of (F)P, and then by

some inference entertain the thought associated with re­

questing an answer about P. Likewise, cases of irony or

cases of sarcasm which are keyed by intonation alone are,

perhaps best thought of as involving an inference to a

particular thought.

This, in conjunction with the following fact,

suggests another way of avoiding the conclusion that S meant

that Jones was a great philosopher. Not every case of

saying something known not to be believed will start an

inference of the relevant sort. If S seems completely sin­

cere in his utterance, then A's belief about S not believing

what he has said will most likely generate only puzzlement.

Perhaps, then, in every case of sarcasm there is some

feature of S's production of the utterance, his intonation,

his manner, or something to this effect that triggers the

inference.

Features of the production of the utterance do seem

to have the power to cue what is meant. The reader can con­

vince himself of this by saying something he does not be­

lieve in a way that would lead an audience which knows he

60

does not believe it to believe he is being sarcastic, and

means the opposite, and then in a way that would lead an

audience with similar beliefs only to confusion about what

was meant. This is not to say the speaker can identify the

various features or that there is only one way of producing

the utterance which will lead to the inference. The point

is that however it is done, speakers know how to say what

they say in the correct way.

The difficulty of meaning that Jones is a great

philosopher can be avoided because there is no intermediate

step of entertaining the thought that Jones is a great

philosopher in virtue of recognizing S intended this thought

be entertained. S's uttering x in a particular manner, and

A's recognizing this manner, will lead to a consideration of

S's belief about P. S would not have meant that Jones is a

great philosopher because he would not have the relevant

intention with respect to this thought. In the other case

where A is intended to entertain the thought that Jones will

freeze, S's intonation or manner does not play the same role

with respect to A's inference. He infers what S means from

something else which he is to consider in the appropriate

way.

Such an answer would have independent motivation in

that it would explain why only some of a speaker's utter­

ances which are knoyn not to be believed by the speaker will

lead an audience to believe he is being sarcastic, and that

61

speakers have some control over whether an utterance will

generate the line of reasoning in the audience. Further, it

accounts for the fact that sometimes we believe a speaker is

being sarcastic even when we know nothing about his beliefs.

Whether there are always such features of the pro­

duction of utterances intended sarcastically is a matter for

empirical enquiry. The enquiry is not unmotivated. Since

for other reasons S does not mean that Jones is a great

philosopher, the reasons clause is not satisfied, even

should this speculation about sarcasm cues turn out to be

incorrect cases of sarcasm will not present a problem for

the analysis.

Attention may now be directed at another question.

Schiffer's (1972) analysis was designed to escape problems

where a speaker apparently had the intentions the definition

required, but could not be said to have meant something.

Will the revised definition with the reflexive intention

avoid these problems in virtue of the changes that have been

made in it? Since the arguments (Grice, 1969; Schiffer,

1972) designed to show that Grice's analysis is not suffi­

cient are of essentially the same form one of them should be

sufficient as a test. They were directed against suc-

ceedingly higher level nested intentions. The counter­

examples differ from each other only in the level of the

intention which A is not to recognize. Hence, a simple

rather than one of the more recherche examples can be used.

62

A variant of a counter-example first suggested by

Schiffer (Grice, 1969) can be adapted to test the new

analysis with the reflexive intention and the weakened

effect. S knows that A is notoriously avaricious, and A

knows that S knows this. Further S knows A knows that S

knows. S wants to get rid of A, and so contemptuously

throws a twenty-dollar bill out of the window. He intends

A to reason that S threw the bill out of the window in­

tending A to infer that he is to leave.

It is difficult to see exactly what must be done to

make this a counter-example to the analysis given previously

(see p. 54). S's intentions would have to be roughly as

follows. A is to realize that since S believes he is

greedy, S believes he would follows. And since his per­

formance was so contemptuous he must intend A realize S

intends him to go after the money. To do this A would have

to leave, so, S must mean that A is to leave. But for S to

have meant something by throwing the money out of the window

he must have intended A to entertain the thought that he was

to leave in virtue of recognizing the intention that he was

to entertain the thought. If it is not out of the recog­

nition of the intention that S intends A to entertain the

thought, S has not meant anything. Suppose that S intended

A entertain the thought via recognizing that S so intended,

that is, at the point where S infers the final thought he

does this in part by seeing this was what S intended.

63

First, it is not clear just how, or if, S could intend this,

but suppose this is how A were to realize he was to leave.

Did S mean something? More specifically, did he mean that

S was to leave? My intuitions are just not able to decide

the case one way or the other. When the counter-example is

reformulated to fit the definition it is not clear that S

did not mean something by throwing the bill out of the

window. The case is just too borderline for a theory to

turn on, and more clear cut cases must be found.

There are two competing accounts, then, of speaker

meaning which can play a part in the analysis of illocu-

tionary acts. Both get S's intention into the open. One

does it via the reflexive character of the communication

intention. The third and fourth examples are examples of

this (see p. 54). The other does it through the use of the

concept of mutual knowledge*. The fifth is a revised

Schiffer-type analysis of speaker meaning with the modified

effect for assertive cases, as follows.

S meant that P by uttering x if and only if S

uttered x intending thereby to realize a certain state of

affairs E which is (intended to be) such that the obtainment

of E is sufficient for S and A mutually knowing* E obtains

and that E is conclusive (good) evidence S uttered x with

the primary intention

1. that A entertain the thought that P is true;

64

2. that there be some r such that S intends that if the

utterance of x causes A to believe that P then part

of his reason be r(t);

3. (may read as in Schiffer's definition).

Schiffer included a clause specifying the nature of the

relationship between the reasons and the intended effect

because of a belief that only certain kinds of reasons are

relevant to meaning. If this turns out to be true, this

definition as well as the two formulations with the re­

flexive intention can be modified accordingly.

Then on either account S will have performed an

illocutionary act if and only if he meant something. And,

what illocutionary act S performed will be identifiable by

the form of beliefs that characterizes what was meant and by

the reason A was to have. Possibly, however, there are

other features by which illocutionary acts are individuated.

One possible aspect of differentiation should be examined.

Other Features of Act-Type Identification

Searle (1975) has noted that some performative verbs

are verbs of manner; they identify a manner of performing

certain actions. It is possible that some illocutionary

acts, likewise, are individuated by the manner in which they

are performed, more specifically by the manner in which S

produces the expression which he is using to perform the

65

act. As we shall see, part of the problem is in specifying

what manner here is.

Suppose that S demanded that A leave the room rather

than, say, ordered or requested that A leave the room. What

makes it an act of demanding rather than one of the other

acts? The intended effect appears to be the same in each

of the cases, that A entertain the thought of his leaving

the room. This cannot serve to separate the cases. It may

be required to know the manner in which the act was per­

formed. This is possibly how an argument for a manner

aspect to identification would go.

A number of questions remain if the investigation

into the existence of acts identifiable by manner is to be

carried out. How widespread is the phenomenon? Other

examples should be sought. If additional cases are not

forthcoming this is some reason to suspect that manner is

not generally a means of differentiation of acts, and some­

thing else might be at work. A fact about manner that must

be account for is that the presence of a certain manner

precludes the possibility of a certain form of reasons

occupying the position in the reasons clause, but does not

preclude the presence of any particular intended effect.

Why is this? Without an explanation it is more reasonable

to argue that the (F)r is really doing the act-type identi­

fication, and that in a case like demanding that someone do

something we just have not looked closely enough at the

66

reasons clause. The general question that this point raises

is whether another basis of act-type identification can be

found in the manner cases. I will take up each of these

issues, but there is a prior worry about manner that merits

attention. Exactly what is it that is being talked about

in this talk of manner? It is not clear that manner here

picks out some physical feature of the speaker's performance.

It is not clear that S cannot beg in the same tone of voice

in which he requests something or demand in a calm calculated

manner. But manner does appear important in those cases in

which there is no performative verb such as "demand" present.

It might be that in explicit performatives the performative

verb takes over the work of manner (Harnish, 1975), for

example, by announcing the manner. I will offer an alterna­

tive explanation of this fact directly. Let me first return

to the other considerations.

Other examples of manner acts are not easy to find.

Pleading and begging might both be manner acts as apparently

the (F)P and the (F)r could vary from case to case while the

act performed remains constant. There seems to be assertive

as well as imperative acts of demanding, begging, and

pleading. These are not, however, obviously assertive

cases. S can beg A to believe that S is loyal, but cannot

beg A that S is loyal. These are probably best thought of

as imperative cases where A is to entertain the thought that

he is to do something, and it is something over which S

67

assumes A has some measure of control as otherwise the cases

do not make sense. The class of manner cases, then, appears

to be quite small, and to contain only imperatival acts.

There is, as has been said, the possibility that the

manner cases have not been thoroughly enough investigated,

and that a closer look at the reasons clause will reveal

the basis of differentiation. That this might be so is

suggested by the apparently close connection between the

various manners of performance and particular reasons which

are intended to be operative in A. S cannot beg A to ijj

intending to provide by the act type as A's reason for

ijj-ing S's authority over A, but see the qualifications below

in this respect. If that is to be the reason, then S did

not really beg. There are not, though, these sorts of

relations between the thoughts that A is to entertain and

the various reasons that are to be operative. Some of the

combined (F)P's and (F)r's might be unusual, but the

oddities would be contingent ones involving the implausi-

bility or the strangeness of achieving a certain effect in

a certain way.

The manner-reason clashes are not just clashes of

this sort. It might be claimed that they are, that the

general point or purpose of a certain manner blocks the

use of a reason which normally is not used to achieve this

point. This just does not seem to be the case here. The

point or general purpose of begging, if such talk is

68

legitimate, is to get someone to do something, and this

does not clash with any particular form of reasons. Certain

manner acts cannot be performed, however, if S intends

certain forms of reasons to be present because of the per­

formance of the act type of such and such.

If manner is interpreted as particular physical

features of S's performance, there is an argument that there

is no manner basis of act-type identification. As noted, S

cannot beg A to ip intending that A have as a reason for

ijj-ing if A does iJj S's authority over A. But S might very

well order A to 1(1, and here the intended reason would be the

authority S has over A, with the physical manner usually

associated with begging, demeaning oneself, appealing for

sympathy, or whatever it turns out to be. Suppose S is a

meek officer who is being intimidated by an enlisted man.

S is frightened, and in an unsure voice cowering behind his

desk says, "I order you to leave my office, now." S has

ordered. S can have A court-martialed for failing to obey

an order should A not comply. The description that best

describes the case is that A was ordered although not in

an authoritative way, and not that A was begged or pleaded

with. It is not an argument to claim that A can defend

himself by saying that he did not think it was an order.

He can claim this no matter how S ordered, sometimes with

justification, sometimes not. Thus, it is not a physical

manner that makes a certain form of reasons inappropriate

69

since that physical manner can be present with that forms of

reasons, but the so-called manner acts cannot be used

appealing to that form of reasons.

It seems that the best explanation is that begging,

pleading, demanding, and other similar acts are not indi­

viduated on the basis of manner, but by the reasons clause

in the analysis of the individual acts. I am not clear on

what the reasons clause would look like for these acts. It

may be that if S begs A to then S intends that if A

actually tjj, then part of A's reason be that has had demeaned

himself. This may not handle panhandling which one might

also wish to include as begging. What about the persistent

panhandler who intimidates his targets into giving him money

through his persistent manner? It may not be that the

giving of money because of the persistent nature of the

begging is a reason generated from the begging. It is not

the fact that S begged that made A give him the money to get

rid of him, but S's annoying manner of asking for money. It

may also be that in this case we would not wish to call what

he did begging. Demanding seems to involve the claim of

some entitlement, but of various sorts, physical superiority,

some right, and so forth.

These are not certainly, final accounts, but if the

reason is something to this effect, the conflict can be

explained. S would attempted via his act to provide A with

a reason, and any other reason would, if the act is

70

(F)r-identifiable, mark off a different act. S might intend

to provide A with more than one reason to do something,

perhaps even conflicting reasons, but not via the perform­

ance of a single (F)r-identifiable act type.

An answer can now be given to the claim discussed

earlier that only when the performative verb was used could

the manner be absent, but otherwise it would be present.

The reason physical manner is sometimes present in acts of

begging is that this manner reveals what S intends A's

reason to be. The performative verb does this without going

through the instrumentality of manner of performance. But

it is not true that one of these two must be present. Any­

thing which is sufficient to reveal the kind of motivation

S intends to provide will show that the act is, for example,

a case of begging. S might just say, "please."

A number of revisions and qualifications are neces­

sary before what has been said about the existence of a

manner dimension above is acceptable. Someone might con­

test what has been said as follows. Certainly someone

might say, "I beg you to do it out of love for your

children." Here S intends to provide A, via one act, with

two different reasons for doing something. If the above

account of begging is correct he intends both the fact that

he has appealed for sympathy or demeaned himself and A's

love for A's children both be reasons for acting. Does not

the answer above rule this out since it rules out intending

71

to provide one reason for doing something by showing that

some other reason is what the act provided? But if one

(F)r-identifiable act can present by itself two reasons for

doing something, recourse to the reasons clause has not

shown why certain illocutionary acts are incompatible with

producing certain reasons.

There is no problem. It is the act type, or in

virtue of the act type, that certain reasons are provided,

and not the specific illocutionary act. Speakers can

provide more than one motivation to do something in the

production of a sentence. In (F)r-identifiable illocu­

tionary acts one of those reasons will identify the act

type. S can order someone to do something or be shot. A

now has a reason to do it, namely to save his life. But it

was not in virtue of being an order that he has had this

added motivation. (Although in certain groups, a disci­

plined revolutionary army, such might be built into the

concept of an order within that group.)

Another counter-example has been offered to show

that no reason is involved with begging itself. Suppose

that S says, "I beg you to leave the country solely out of

love for your children." Has not S here intended A's

motivation be only A's love for his children? Hence he did

not intend the begging itself provide a reason for or be

part of A's for leaving the country. S had asked A to do it

72

just out of love for A's children. This is due to J. L.

Cowan (1975).

Quite obviously though, S did intend the begging to

be part of A's reason for leaving the country. If the

utterance was not supposed to effect A's action why did S

say what he did? Why did not S walk up to A and kick him in

the stomach, stay in bed all day, go to a movie, or just

stand around with a blank look on his face hoping A would

leave the country solely out of love for his children? S

said something because he thought saying something might

help to get A to do something. To maintain otherwise re­

quires an explanation of why S would beg or ask or plead

with A to ip rather than sharpening a pencil, saying, "It's

a warm day," or doing nothing at all.

Another possible response requires separating the

act of begging from the act of begging someone to do some­

thing for a certain reason. Begging provides him with a

reason not just for doing something, but for doing some­

thing in a certain way. This introduces the new problem

of individuating acts, but it is not a problem on which the

project rides or falls.

It is crucial to keep two points distinct. The

first is the claim that to perform an illocutionary act is

to provide someone with a reason for doing something. The

second is the claim that a particular illocutionary act

provides such and such a reason. The analyses I will give

73

of particular illocutionary acts may be incorrect. By

itself this would say nothing to the correctness of the

first point. Further, it may even be that certain acts

allows for the possibility of more than one reason. As we

shall see in Chapter II, reporting appears to be such a

case.

The best description of the facts is that the

candidates for manner-identifiable acts are really reason-

identifiable acts. Some reason has been given earlier in

the chapter why illocutionary acts should separate along the

various things meant or the various motivations to be pro­

vided the hearer. That is because it appears in many

illocutionary acts what is being done is getting people to

believe or do something, and providing them with reasons

for believing or doing those things. Manner lacks such an

independent motivation. The argument for excluding manner

is, then, a simplicity argument. It was suggested that

where manner does play a role it does it by cueing certain

reasons.

If someone wishes to argue that manner needs to be

included in any account of act-type identification he must

do one of two things. He might provide a theoretically

motivated account of the necessity for manner. Or, he

might seek counter-examples to the claim that manner

operates in the way this theory says it does. That is, he

might find cases which require reference to manner. Should

74

such examples be found it would not be difficult to add a

manner parameter to the theory. There would be the re­

sulting loss of simplicity by the addition of one more

primitive. The change would not be complicated.

A futher possibility along these lines is that

manner acts are really a sub-class of reasons-identifiable

acts, where the manner S adopts is itself part of the reason

S intends to provide A. Whether this is so or whether

manner is just a means to produce some independently

specifiable reason, until the examples are forthcoming to

show otherwise the theory will not include manner as a

separate characteristic which individuates illocutionary

acts.

Briefly, the theory sketched so far is that S has

performed an illocutionary act if and only if he meant

something. The relevant notion of "he meant something"

should by now be clear, Schiffer's (1972, p. 2) notion of

S-meaning, of what it is for someone to mean something by

or in producing or doing x. To mean something in this way

is just to intend in a certain way for someone to entertain

some particular thought. The choice has been left open

between two accounts of speaker meaning to serve in the

analysis. Since it is not necessary to intend to create a

belief or get someone to do something in the performance of

an illocutionary act, an analysis cannot require this. The

modifications of Grice (1957) and Schiffer (1972) offered

here have the advantage of not requiring this, thus avoiding

a large class of counter-examples. They do it, however,

without deserting Strawson's (1971) suggestion that to

perform an illocutionary act is to mean something. The re­

visions also allow room for the observation that there are a

wide variety of reasons and purposes a speaker might have

for performing a given illocutionary act, a rather pedes­

trian observation, but one that outpaces a number of high-

powered theories.

Such an analysis is not without significant prob­

lems. For a large class of illocutionary acts the speaker's

intentions or what the speaker meant seem irrelevant to

whether the act was performed. The naturalism of this

account, in the sense of relying only on certain psycho­

logical concepts avoiding popular notions like constitutive

rules, runs up against illocutionary acts which are said to

be essentially conventional or essentially institutional.

Will such illocutionary acts require abandonment or serious

revision of this account? In the next chapter attention

will be directed to this issue.

CHAPTER II

INSTITUTIONS, CONVENTIONS, AND ILLOCUTIONARY ACTS

In order to see how adequately this analysis will

handle conventional illocutionary acts it must be determined

how conventional acts relate to other illocutionary acts,

and exactly what it is about them that produces the tempta­

tion to separate them from other illocutionary acts.

Strawson (1971) suggests that illocutionary acts form a

continuum, and that at one end are acts like marrying,

bidding, calling out, and pronouncing a sentence which are

essentially conventional. He notes that if play is strict

a person has redoubled in bridge even though his utterance

in unintentional, and he did not mean to redouble. How can

an analysis which asserts that to perform an illocutionary

act is to mean something handle such cases? Are there

other accounts which provide a better account of such cases?

There are other accounts which seem particularly

well suited to these kinds of cases. Searle (1971, p. 42)

claims that to perform an illocutionary act is to engage in

a rule-governed form of behavior in much the same way that

to play a game of chess is to engage in a rule-governed

form of behavior. This suggestion is carried out in some

detail in Speech Acts (Searle, 1969). Because it can

76

77

account for many of these kinds of cases, and because of its

well-regarded position in the philosophy of language,

Searle's account does merit discussion.

In his work, however, Searle (1969, p. 54) does not

provide an analysis of the concept of an illocutionary act

in these or in any other terms. His remarks suggest that

he intends otherwise as he says, "The ground has now been

lain for a full dress analysis of illocutionary acts," but

the analysis is never forthcoming. Rather, he contents

himself with providing a list of necessary and sufficient

conditions for promising, and from this secures a list of

rules for the use of the force indicating device for

promising. He suggests how various other acts might be

analyzed in the same way.

Searle's theory, despite its apparent affinity for

cases like those mentioned by Strawson (1971), distorts the

nature of the facts upon which it is based. Further, it is

of very limited scope, and it is not clear how it can be

extended at all.

An examination of his list of conditions for

promising should reveal that the analysis he is attempting

is far more limited than a general analysis of promising.

It is at best an analysis of promising when the speaker

promises with an explicit performative sentence. Searle's

(1969, pp. 60-61) conditions 8 and 9 require

8. S intends (i—1) to produce in H the knowledge (K) that the utterance of T is to count as placing S under an obligation to A. S intends to produce K by means of recognition of i-1, and he intends i-1 to be recognized in virtue of (by means of) H's knowledge of the meaning of T.

9. The semantical rules of the dialect spoken by S and H are such that T is correctly and sincerely uttered if and only if conditions 1-8 obtain.

Even granting his requirement that T is not ambiguous this

clearly will not suffice as an analysis when T is not an

explicit performative sentence. Suppose S promises by

saying, "I'll be there." Clearly this sentence is not

uttered if and only if S intends to produce the knowledge

that he is undertaking an obligation to do A for even

holding the expression's meaning constant the sentence is

also correctly uttered if S is predicting or guessing he

will be there.

This type of problem affects his account of the

rules which are supposedly the underlying basis of language.

These are rules for the use, he says, of any force indi­

cating device, Pr, for promising. These are the rules that

Searle (1969, p. 63) provides.

Rule 1. Pr is to be uttered only in the context of a sentence (or longer stretch of discourse) T, the utterance of which predicates some future act A of the speaker S. I call this the propositional content rule. It is derived from the propositional content conditions 2 and 3.

Rule 2. Pr is to be uttered only if the hearer H would prefer S's doing A to his not doing A, and S believes H would prefer S's not doing A.

Rule 3. Pr is to be uttered only if it not obvious to both S and H that S will do A in the normal course of events. I call 2 and 3 the prepara­tory rules, and they are derived from the preparatory conditions 4 and 5.

Rule 4. Pr is to be uttered only if S intends to do A. I call this the sincerity rule, and it is derived from the sincerity condition 6.

Rule 5. The utterance of Pr counts as the under­taking of an obligation to do A. I call this the essential rule.

It appears that these rules apply only to a force

indicating device like the performative prefix, "I promise."

Someone might promise by nodding, but certainly nodding is

not governed by any such set of rules unless this analysis

is qualified by prefacing it with a phrase such as "When

promising by using Pr," or something equivalent. But in

that case the rules are not any help in understanding

promising since we would not be able to proceed by starting

with an account of the rules for the use of devices since

the promise-making function of those devices would already

have to be identified. Further, the preface is needed even

for a force indicating device like the performative prefix

since even "I promise" is sometimes not used to promise.

The analysis is not saved by adding a qualification to the

effect that Pr is correctly uttered or literally uttered

only if Rules 1 through 5 obtain. The non-performative uses

80

of the prefix are not in any sense incorrect. They just are

not uses via which S promises. It is hard to see how the

"correctly" or "literally" could be spelled out on Searle's

ground except to the effect of being uttered to make a

promise. All this achieves is to shift the work of the

analysis to one of these terms.

Searle (1969, p. 68) tries to handle one dimension

of the difficulty by saying that sometimes there is no

explicit force indicator, in which case the context and the

utterance make it clear that the essential condition is

satisfied. The essential condition must be Searle's condi­

tion 7 (p. 60) that "S intends the utterance of T will place

him under an obligation to do A." It is hard to see how

this is to say any more than that sometimes various elements

of context will reveal our intentions when T underdetermines

them. This will not help with any problem that we have with

his rules unless an account of how intending that a certain

condition be met will save a biconditional that fails in

both directions. Nor does it save his list of conditions

since even the use of a performative formula underdetermines

the illocutionary consequence, when S, for example, makes a

joke and is not promising.

The difficulty is with the form of the rules. They

fail on the one hand because they need additional qualifica­

tions even to get all promises with the performative prefix,

and on the other because at best they get only the promises

81

with the performative prefix. Searle (1971, p. 44) says

there are a variety of force-indicating devices in English,

among them word order, intonation contour, and mood in

addition to the performative prefixes. However, supposing

that the rules are rules for the use of these devices, then

there are rules which require, for example, a speaker to use

such and such stress only when the hearer would prefer a

certain action being done or requiring that a certain mood

is used only when the speaker wants a certain action per­

formed. These are more dubious that the original case with

the performative prefix, and certainly do not provide any

way out of the difficulties.

Searle has worked out in detail only one case, that

of promising. He suggests what some of the conditions

might be for other illocutionary acts, but he has given us

no idea of how the analysis might help to give us an

analysis of the concept of an illocutionary act. It is not

clear that generalizing on these conditions would work or

even how one would generalize on them. Searle's promise is

unfulfilled.

Even in their more limited endeavor the rules do

not reveal much. Perhaps his rules can be reformulated to

avoid these problems. But the attempt to salvage Searle's

pattern of analysis need not be made. Although the features

of illocutionary acts like promising or sentencing suggest

some analogy with rule bound activities like games, there is

82

no need to seek an analysis of such acts in terms of the

systems of rules which may define these games. An analysis

like the one in the first chapter which relies on psycho­

logical notions is adequate for handling the conventional

cases as well as the more "naturalistic" cases like stating

or objecting. At this point I will abandon the discussion

of Searle, and turn to the conventional cases.

Institutional Acts

Lists of conventional acts like Strawson's (1971)

collapse two kinds of cases which prima facie should be kept

distinct. There are acts such as sentencing and bidding

which do seem intimately involved with social practices or

social institutions. Other examples are calling out,

betting, and christening. These are different from cases

such as greeting which we are tempted to label as conven­

tional, but where the surrounding social practice is not

clearly identifiable. What is it other than the act? To

have a name for this former type of case I will refer to

them as institutional cases.

I shall argue first that certain ways in which these

acts might be thought to involve institutions is incorrect,

and then attempt to point out what marks them as institu­

tional. If I am correct, then much of the motivation for

viewing language as analogous to games is a result of mis­

reading various facts about these institutional cases.

83

First, it should be obvious that it is not just

making contact with some institutions which, makes an act

institutional. Although the existence of institutions would

be necessary for the existence of particular acts like

asking what the bid was or stating that someone was

sentenced to thirty years, neither of these is institutional

in the way in which bidding or the sentencing itself would

be. What needs to be done is to determine exactly what it

is about these acts that separates them from the more

naturalistic acts.

The institutional status of these acts is a result

of the special role they have within the social practice.

Sentencing (and throughout when I speak of sentencing I am

referring to the judge's act of pronouncing sentence) in­

volves the determination that someone is to pay a certain

penalty for some alleged offense. This is certainly not

sufficient as an account of sentencing since the determina­

tion of the penalty might have been made by some natural

means in which case it would not have been described as an

act of sentencing. There could be a society in which the

age of the first person whose shadow falls on the criminal

after the criminal is caught will determine how long the

criminal is to serve in prison. This could occur without

anyone ever announcing that this was the sentence. In this

way the practice of establishing a certain penalty for a

person, which is a part of the larger institution of

84

punishment, could go on without the act of sentencing. A

crucial feature of sentencing is that whoever is empowered

to sentence intends his utterance to perform the role of

determining the penalty, of being what people appeal to if

wish to know the penalty. The authority intends his

utterance itself, because he has certain powers, to be the

people's reason for believing the determination has been

made. This is not to say that his determination is not

based on various facts. However, his act is itself the

determining feature, and is intended to be such.

If this is correct it would mean that one of the

characterizations given to such acts is in error. It is the

view that for certain cases, the institutional and con­

ventional ones, just to utter x in circumstances C is to

perform the illocutionary act. But for sentencing, and as

I will argue later, for a number of other cases, the

appropriate S saying the appropriate thing, "I sentence you

to eight years," in the requisite situations is not the

performance of the illocutionary act without the appro­

priate intentions on S's part.

Suppose that a good friend and golfing partner of

the judge comes before the bench on a charge of making an

illegal turn. The defendant is always complaining to the

judge that the courts are soft on criminals, and instead it

should be making examples of them. The judge, valuing a

good joke more than judicial propriety, says, "Courts have

85

long been too soft on criminals, and I intend to make an

example of you. I hereby sentence you to forty years."

Intending to make a joke to his friend in circumstances

where it is obvious it is true that he is joking the judge

did not sentence him to forty years. Such an action may be

inappropriate, and subject to criticism on a number of

grounds, but it is not an act of sentencing.

It cannot be maintained that despite his intentions

he did sentence any more than it can be maintained that a

person told someone something even if he was making a joke.

At a meeting of the Young Democrats someone stands up and

says, "John Ehrlichman and Bob Haldeman are the finest

pbulic servants it has been my pleasure to know." A person

who reports to another that this person told the meeting

that Haldeman and Ehrlichman were the finest public servants

he had known would be incorrect. He did not tell anyone

that. He did not intend to be stating something; he was

joking. Likewise, the judge did not sentence, and to insist

that he did just because of the words he used is to mis-

describe the situation. The person who wishes to deny that

intentions are necessary must explain why in cases where all

that is missing is the person's intention to 0, such as if

he was joking, this is sufficient to withhold in some cases

the claim that he did perform the act. At the Young

Democrat's meeting this might have been all that was dif­

ferent from a case of telling P that S did not intend to

86

tell anyone P, but only to make a joke. It might differ

from a case of telling only in that what was necessary to

reveal his intentions was present. If intentions are not

necessary in this case, why did he not tell? The argument

for the presence of intentions with the institutional cases

will be of this form. In some cases where S did not perform

the illocutionary act the only difference is that his

intentions were absent. If intentions are not necessary

why should this mitigate against the performance of the act.

The following (Cowan, 19 75) has been suggested as a

counter-example to the claim that a judge must intend to be

sentencing. Suppose that, as described above, the judge had

just said to his friend, A, "I sentence you to forty years,"

whereupon the judge dies of a heart attack. He did not

intend to have sentenced. However, suppose that the court

officers were not aware that it was a joke, and send A away

to spend the next forty years breaking rocks and ironing

shirts. Was not A sentenced even though the judge had not

intended to sentence him? A is, after all, now languishing

in the state prison.

It is not clear that A is serving a prison term

because he was sentenced. Suppose that a judge sentenced

Don Mudd to be hanged. The hard-of-hearing court recorder

thought he said John Rudd, and dutifully recorded the name.

Shortly thereafter, Mrs. Rudd is a widow. Now Rudd was not

hanged because the judge sentenced him, but because court

87

officials believed he had. Mudd was sentenced. Likewise,

in the other case it seems correct to say that A is in

prison not because he was sentenced, but because some people

thought he had been.

Doubtless, judges should not make jokes in such

situations. The institution is important and serious, and

joking demeans it. That does not alter the role that the

intentions of the speakers play in the institution.

Another of the cases mentioned by Strawson (1971),

an umpire calling out, has features like those ascribed to

sentencing. Again, note that it is a determination, a

determination of which way play is to continue, but there

is certainly no necessity that it be the umpire's call that

provide that determination. If all players were honest, and

no plays were close there would be no need for anyone

calling anyone out. The game would continue in pretty much

the same way, and there would be no reason to think it was

not the same game. But, players are not all honest, and

even if they were sometimes plays are close enough that a

determination is needed as the player may not be in a

position to judge the situation.

In this case there is a certain dimension for

assessing the umpire's calls. He may be right or wrong

about the facts. But for the game to go on some authori­

tative determination must be made. The call by the umpire

is, itself, a reason to continue in one way rather than

88

another. It replaces certain other facts as reasons for

proceeding in a given manner because what is needed at some

point is a reason which is not open to dispute in a way that

the question of whether the runner actually did beat the

ball to the base is. The umpire either said, "out" or

"safe" or made the appropriate hand signal. That is a

reason for continuing one way rather than another.

That contestants can argue with the umpire shows

that his determination is supposed to be based on the facts.

But the correct description of the act is not exhausted in

saying that it is a determination of the facts since this

could be done in other ways which would not be described as

illocutionary acts. As said above, it might have been the

case that there were never any disputes at all, but that

there are and that an authoritative determination of the

facts is made is not the basis on which calling someone out

is said to be an illocutionary act. The umpire's role may

someday be eliminated in favor of a closed circuit replay

and a computer's evaluation. The printout might simply

read, "Player did touch base prior to ball's arrival." It

is certainly open to dispute whether the computer's pro­

ducing a printout was the performance of an illocutionary

act. But there was an authoritative determination of which

way play was to proceed, for example, the player was to

remain on base. What makes the umpire's call what it is are

89

the rules which empower the umpire to make certain decisions

and his intentions to make those decisions by his calls.

Still, the role of the intentions in these game

environments is not clear. I will try to construct a case

to show that the intentions are not necessary, and then turn

to a discussion of that case.

A group playing bridge is discussing a famous

incident from a bridge tournament. Stanislaus was known to

have a terrible temper, and once during a match when his

partner followed an opening bid of one club with a bid of

two clubs Stanislaus leaped across the table, and strangled

him. The bidding opens with one club, and wishing to make a

joke S responds, "Two clubs!" If it is maintained that he

did bid here even though everyone knew he was joking

stronger cases can be constructed. It is S's turn to bid,

but before he does someone asks him what the bid was last

hand. S responds by saying, "Four clubs!" He did not here

bid, it might be maintained, even though all the circum­

stances were appropriate. He was playing a game, it was

his turn to bid, and so on. The point is that even in cases

like this the intentions of the agent are part of the re­

quired conditions for the performance of the illocutionary

act.

It may be insisted that if he said, "Four clubs,"

then he bid four clubs. He can be held responsible for

bidding four clubs regardless of his intentions. There are

90

a number of different things that might be said here.

First/ it might be admitted that S bid, but what belongs to

bidding as an illocutionary act, that is, as a speech act,

be restricted. It may be that the illocutionary, speech

act, part of bidding may be exhausted by deliberately

uttering a sentence which means S makes a bid. Hence what

S intends to do is utter a sentence that means such and

such. This does not exhaust what it is to bid, but these

extra features of bidding have nothing to do with bidding

being an illocutionary act.

That S must deliberately utter a sentence which

means, roughly, that he bids is suggested by the following

considerations. Suppose that it was discovered as S made

his bid that a wire ran from S's head to a control box in

the hand of one of the opposing players, and every time that

player pushed a button S made a bid. S would protest after­

ward that he had not intended to make any such bid. The

opponent would tell him to shut up, and quit trying to take

back his bids. It is decided that the wire runs to the part

of the brain that controls part of S's speech functions, and

that the other person is controlling S's speech. It is

decided that S did not make the bids, and should not be held

responsible for them. The correct description is not that

they have decided to let S take his bid back. They decided

he did not make those bids.

91

How seriously is the program damaged, though, if

someone insists that even here S really did bid regardless

of his intentions? I think the answer is that the program

is not much affected. The cases where this might be the

case are extremely limited in scope. They do not include

all institutional acts as shown in the case of sentencing.

They seem restricted to a few game or game-like cases such

as bidding or resigning in chess. It is of the nature of

games that an unequivocal determination of what constitutes

a move is necessary, and not a feature of the nature of

illocutionary acts. These cases are very limited, and play

no central role at all in communication. They have the

character they have only in one context, that of playing

that particular game. Further, they could keep their

essential character as moves, and be performed in a manner

that would not be described as performing an illocutionary

act. Given their restricted role, and their distance from

the center of communication, if cases such as bidding or

resigning in chess can be performed in the absence of the

intentions this would not derail the project.

And it is not clear to what extent the act can be

performed in the absence of S's intentions or what is going

on in these cases. In the case where S mentioned what an

earlier bid was it may very well be that one of the other

players would have taken him to have been bidding just

because clearly that was not his intention. It may even be

92

that there is some (although I am not sure) equivocation in

the phrase "S bid such and such." Suppose S mumbled some­

thing that was not a bid, but that everyone misunderstood

him, and thought he said "I bid three hearts." S is not

paying attention or is afraid to say he did not say that,

and before he can anyway, everyone passes. We may say here

that his bid was three hearts. That is what he will try to

make or surpass. We might very well say, if officials

decide it is his fault for not stopping the action, that his

bid is three hearts. Yet it is also clear that he did not

bid anything. He just mumbled something about the lunch

that he was going to have or cleared his throat. The

equivocation in "S bid such and such" might be between this

latter sense in which S did not bid, and the sense of S's

bid being recognized as such and such. S is held responsible

for having bid such and such, and so did bid such and such,

but S did not really bid anything. Unless there is some

equivocation this sentence should be a contradiction. It

is not clearly one.

Four options are available then. It might be denied

that S bid if he did not intend to in any sense. This is

too strong. The illocutionary act part of bidding might be

identified as only part of what is going on. "Bidding"

might be viewed as equivocal. The argument in the paragraph

above is an argument for this. Anyone who would deny this

must claim that the description above is contradictory or

93

provide an alternative explanation. Finally, one might

yield here, but point to the very special kind of case, the

limited role in communication, and maintain that admitting

such does not harm the program. Either of the last two

strike me as acceptable. Until an alternative account is

offered as to why the statement above is not contradictory

there is no reason not to choose the third.

Even though S's saying x in circumstances C is not

simply what it is to perform the act, it seems that if S did

utter x in C the assumption is that he did perform the act.

The utterance is more closely tied to the performance of the

act in these cases than it is in cases like reporting or

predicting. If S said it the presumption is very strong

that he meant it, that he performed the act. Why in these

cases is there a closer connection between saying the right

thing in the right circumstances and performing the act than

there is with non-institutional acts? This also is to be

explained in terms of the special function that these acts

have. Sentencing and calling out are acts which occur at a

particular point within the practice or institution, and

arise because a reason is needed for people to continue in a

particular way. A decision is needed to provide people with

a reason. The pronouncement will be a reason for the con­

cerned parties, and it is intended to serve as such a

reason. If the circumstances are correct, for example, a

person has been found guilty of a crime and is brought

94

before the judge, then when the judge says, "Ten days or

ten dollars," nothing more is needed in most cases as

evidence of his intentions. This is because such pro­

nouncements were designed to play a certain role in fairly

rigidly circumscribed situations, and given the utterance

and the situation it will be assumed that it played the

role. Most likely this is just the role the speaker

intended it to play.

Other institutional acts share these character­

istics of being designed to do a particular job in

particular circumstances. Hence, when they occur in these

circumstances the assumption is that the speaker is in­

tending to use them in this way. Non-institutional acts

are not restricted to such specialized roles which is why

it seems speakers intentions matter more in these cases

than in institutional cases. As far as the illocutionary

speech act part of the act is concerned, the intentions are

necessary.

Institutional acts are illocutionary acts which play

a certain role in some practice or institution. But since

the speaker's intentions are necessary they are not really

outside of an account like the one presented in the first

chapter. It is just that the intended effect or the reason

A is to have must make reference to some institution. They

differ from naturalistic illocutionary acts in that the

form of beliefs, F, which characterizes P or r will make

95

reference to the institution. The non-institutional acts

touch the institution through P, and not through F. This

is just to say that the individual act, and not the act

type is related to the institution as the P is what is

variable through particular acts•

For example, in warning that P, the F would be

characterized to the effect that P represents a danger. If

S warns A that A will be sentenced to forty years in jail,

the act of warning does make contact with institutional

notions, but through the P, and not through the F.

This account of illocutionary acts reveals what I

consider a central error in a rather widespread account of

illocutionary acts. The account maintains that language

or language use is rule bound in much the same way that a

game is rule bound. Searle (1971, p. 42) takes the analogy

quite seriously. The illocutionary acts which are mustered

in support of this thesis, and which lend it an air of

plausibility, are those acts like bidding and sentencing

and calling out. It seems quite likely, however, that

bidding seems like a move in a game not because of any

feature it has in virtue of being an illocutionary act, but

because in addition to being an illocutionary act it is a

move in a game. But because it has certain characteristics

in virtue of being a move in a game, it certainly does not

follow that other illocutionary acts which are not moves in

some game share these features, or that the features are

96

basic ones for an analysis of illocutionary acts. To argue

from these institutional acts to the fact that all illocu­

tionary acts are to be understood in this way is akin to

arguing from the fact that certain human actions such as

sliding into second base require explication in terms of

the rules of baseball to the conclusion that all human

actions must be understood as governed by rules. Focusing

on the institutional features of these acts has shifted

attention from the features that they share with other

illocutionary speech acts, the fact that the speaker has

certain intentions in performing the act.

It might be argued that there is one further feature

of the institutional acts which seems to separate them from

the naturalistic cases. Is not the intended effect some­

thing stronger than just the audience's entertaining a

particular thought? Does not S intend as a part of the

performance of the act to provide the audience with a reason

for actually believing or doing something? Does not the

judge intend that A actually believe that it has been

determined that the criminal is to spend forty years in

jail, and not just to entertain the thought?

Whether or not this is so depends in part on

exactly what the intended effect is. That the person

actually pay the penalty seems too strong. It may be clear

that for some reason or another the person will not actually

spend the time in jail, and so what S intends is not that A

97

believe or even entertain the thought that someone will

actually spend the time in jail. It may be widely believed

that he will escape or be freed by powerful political allies

What the audience is to believe or consider is not that

some person actually will spend a certain amount of time in

jail or pay some other penalty. The content of the intended

effect is that it has been determined by S that the person's

penalty will be such and such. It may further be expected

that the penalty will actually be exacted from A, but this

is expected because of beliefs about the circumstances in

addition to any beliefs about the sentencing having occurred

But now, must the judge intend the relevant audience

actually believe that the determination has been made that

someone is to pay a certain penalty? Or is the intended

effect only that the audience entertain this thought? Con­

siderations analogous to those in the previous chapter will

show that even here the requirement of intending that A

believe something is too strong. Suppose that the law

allows that S is the proper person to pass sentence in the

case, but S is a paranoid who believes that someone is out

to destroy his career. He is sure that these people who

seek to destroy him have convinced his fellows that he is

an imposter who is not legally empowered to pass sentence

in this case. He is sure that these people will convince

the review board or higher court, if it comes to that, that

he is not properly empowered to pass sentence. He is

98

convinced that when he passes sentence, it will not be

believed that the sentence has been determined. Being a

creature of duty, a Scorpio, he feels that he must nonethe­

less pass sentence. When the criminal is brought before him

he makes the pronouncement that the person is to spend so

much time in jail. He has sentenced even though he is con­

vinced that no one will actually believe the determination

was really made any more than they would believe the deter­

mination was really made if a stranger had walked into the

court room, sat in the judge's chair, and announced a

sentence.

It is not necessary to resort to speakers who are

deranged in some way. S may have very good grounds for

believing that no one will believe that the sentence was

pronounced. This will not prevent what he does from being

sentencing. Similar points can be made about various other

institutional acts. Anything stronger than the modified

intended effect seems too strong.

Conventional Acts

The institutional acts are not the only acts which

at first blush appear not to be amenable to the speaker

meaning analysis. Other acts, which will be referred to as

conventional illocutionary acts, are not acts which play a

special role in some institution, but they share features

99

with this class not shared with those in the naturalistic

class.

Among those acts best thought of as conventional are

greeting and bidding someone farewell. Here too the tempta­

tion to claim that just to say x, for instance, "Hi," in

particular circumstances is to greet is mistaken. Greeting

shares with other illocutionary acts the feature that the

speaker intends the audience to realize he is performing

that act. Suppose S and A pass on the street. S is deep in

thought, and does not wish to interrupt his thinking to

address A. S is going over a perplexing dialogue w ich

occurred earlier in the day. Occasionally he repeats some

of the phrases aloud. He utters, "Hi," not intending to

greet, but merely repeating something from the earlier

dialogue whose significance eludes him. Clearly S did not

greet A even though the circumstances were such, and the

utterance such as might normally be present if S did greet

A. S recognized A. S and A were encountering each other

for the first time in a certain period of time. The dif­

ference between this case, where S did not greet, and a

case of greeting is that S did not intend the utterance be

taken in a certain way. It is just the absence of inten­

tions that separates the cases. The difference is not in

the circumstances, that earlier in the day S had a per­

plexing dialogue with someone else which is is now going

over. This might have been the case^ and S still greeted,

100

The difference is just in how S intended the utterance to

be taken.

Against suggestions that there might be some

possible circularity in providing an account of greeting in

these terms, I think there is no difficulty here. The

analysis of greeting, and see below for more detail in this

respect, will require the intention that A entertain a

certain thought, for example, that A's presence is being

acknowledged by S. (This is not sufficient for there will

be other requirements, but it will provide some idea of

what the analysis would look like.) The analysis of

greeting will not say that one greets only if one intends

to be greeting. The analysis will be as above. But there

is nothing, then, wrong or misleading about saying that one

greets only if one intends to, at any rate, nothing circ­

ular. Some jurisdictions require the specific intent to

kill for first degree nonfelony murders. But we understand

someone who says that in order to have murdered one must

have intended to have murdered.

Greeting difffers from both the institutional and

the naturalistic acts discussed so far. It differs from

institutional acts in that it is not designed to play a

certain role in furthering some institution or practice,

Again, greeting may occur in some such way that a descripr-

tion of a particular utterance token may require reference

101

to some institution. The act type, however, does not

depend on the institution.

It shares with the institutional cases being such

that given its utterance in certain circumstances, the

presumption is that the act was performed. And, there are

certain circumstances where the act takes place. A speaker

greets on encountering a person, not that person's de­

parture .

The question as to how the conventional cases differ

from the naturalistic cases as well as how they differ from

the institutional ones is still before us. There are a

number of differences, but they seem to be basically the

result of one feature, that greeting lacks content. It

does not have both force and content as other acts do. A

speaker just greets, but a speaker does not just promise of

state or object. He promises to do something or states or

objects that something is the case. This appears to be

reflected (Byerly, 197 5) linguistically. Illocutionary

Verbs paired with the conventional illocutionary acts do

not take a "that" clause (or in some cases "for1') , In the

fact that there are only certain ways to perform the

illocutionary act lies part of the reason why there are

only certain ways to perform the act, and why the convene

tional act is tied more closely to certain circumstances.

If the act has a variable content as does promising or

stating the speaker can mention that content in the

102

appropriate circumstances, and perform the act. He can say',

"I'll be there," and have promised to be there. There are a

number of ways a person can promise for each specific act

type of promising. There is no variable content to greeting,

and consequently this cannot be referred to as a means of

revealing our intention to greet. Hence it is more likely

that there would be some small number of formulas which

would arise for the performance of the act, and given that

the speaker utters one of them the more likely that he is

greeting.

This opens the door, although not very far, to the

inclusion of certain other acts in this category. Other

acts, even though they have some content, do not have it in

the same way that other acts do. These are acts like

thanking and apologizing. That they differ from ordinary

naturalistic cases is revealed by the fact that the act is

not performable by referring to the content, A speaker can

perform the act by referring to the relevant mental state,

the one associated with the act, roughly regret for

apologizing and gratitude in thanking. Hence S can thank

by saying that he appreciates something or apologize by

acknowledging regret. Of course, having the mental state

and acknowledging it is not sufficient to perform the act.

But as there is only one or a small range of mental states

associated with the act rather than an indefinite number the

parallel would hold. The ways in which the act can be

103

performed are limited, and the act comes to be associated

with that small number of ways. All in all it seems that

thanking and apologizing do share a number of features with

greeting, and might be thought to shade toward the conven­

tional end of the spectrum

Possibly what is happening with these cases is some

ritualization of the performance which has already occurred

with greeting. As the act becomes more and more a matter

of doing the expected, the psychological state or the con­

tent diminishes in importance. It is just something one

does. The way in which it is done would also then become

restricted to a limited number of means which become asso­

ciated with that act until the ritual becomes the dominant

factor. It is just a matter of doing that thing. Greeting

is closer to fitting this description than are the other

cases, but they might be thought of as moving in that

direction

In summary, then, within the broad range of acts

described in the literature as "conventional" there are two

different types of cases. Neither escapes analysis in terms

of speaker meaning. The members of each are species of M-

intending (Grice, 1969) to create a certain effect, namely

A's entertaining the thought that P is true, In some of

the institutional cases discussed above it is the fact that

at some point an unambiguous and final reason for continuing

in one way or getting on with the practice that separates

104

them from the other acts. The official's decision would be

such a reason. Other institutional cases function in dif­

ferent ways. Bidding itself is an integral part of certain

games, and not just something that allows the game to con­

tinue at the point where a decision is needed about what

has happened or what is to be done. Conventional acts are

acts which move toward a dimension where acts are performed

in rather fixed ways or where content increasingly

diminishes.

Austin (1962, p. 103) said that illocutionary acts

were conventional in that their force could be made explicit

by means of the performative formula. Clearly this is in­

dependent of the way in which acts like sentencing or

bidding are conventional as opposed to naturalistic illocu­

tionary acts. In the former sense the acts are conventional

in that there is a way of doing something which more or less

uniquely determines the act. There is a conventional way

or a conventional means of accomplishing that end. This

must be distinguished from saying the illocutionary act type

is itself conventional. There may, for example, be con­

ventional ways of getting someone's attention, but it

certainly does not follow that getting someone's attention

is a conventional act.

Not all acts are conventional in this latter sense,

and even cases usually labeled conventional^-promising,

greeting, sentencing, ordering, and so on-—are not univocal

105

cases. Part of what I have done is sort out the differences

at this second tier of conventionality. This clears the way

for attempting a classification of all of these acts.

A Taxonomy of Illocutionary Acts

Certain illocutionary acts among those described

above share properties with some which they do not share

with others. It is appropriate to ask if these common

properties suggest that illocutionary acts can be classified

into kinds. Austin (1962) first attempted to answer the

questions about how many and what kinds of illocutionary

acts there are. Searle (1975) claims that Austin is pro­

viding his readers not with a list of illocutionary acts,

but with a list of performative verbs. Austin C1962,

p. 148) does not keep the task of classifying illocutionary

acts and classifying performative verbs distinct, but it is

clear that he was attempting more than just a classification

of verbs. "We said long ago that we needed a list of

explicit performative verbs; but in the light of the more

general theory we now see that what we need is a list of

illocutionary forces of utterances." And in a later

passage Austin (1962, p. 149) says he rejects talk of

performative and constative "in favor of more general

families of related and overlapping speech acts, which are

just what we have now to attempt to classify," In any event

106

it can at least be said that Austin's classification is to

be determined by the various forces of the utterances.

Unfortunately Austin provides no consistent basis

for his classification. He admits that his classification

is far from satisfactory, but claims it at least provides

a vehicle for examining the true-false and fact-value

distinctions. In his discussion Austin divides utterances

into five classes.

Verdictives are those which involve giving a

verdict or finding, either a final one like a judge or an

umpire or a temporary one like an estimate or appraisal.

Exercitives are the exercising of powers or rights.

Appointing, ordering, warning and advising are exercitives.

Commissives commit the speaker to something. Promises are

commissives, but so are declarations which commit the

speaker to a particular view or position. Behabitives have

to do with the speaker's attitude or with forms of social

behavior. Apologizing and commending are examples,

Expositives demonstrate the way in which what the speaker

says fits into the conversation. In listing members of this

class Austin does not talk in terms of speech acts, but in

terms of performative prefixes. Hence he says that

expositives include, "I reply" and "I assume" among its

members.

107

In introducing his taxonomy Searle (1975, p. 15)

makes short shrift of Austin's classification, Searle sums

up his objections to Austin as follows:

In sum, there are (at least) six related dif­ficulties with Austin's taxonomy; in ascending order of importance: there is a persistent con­fusion between verbs and acts, not all verbs are illocutionary verbs, there is too much overlap of the categories, many of the verbs listed in the category don't satisfy the definition given for the category and, most important, there is no consistent principle of classification.

Searle is correct that the last is Austin's biggest failing.

He attempts to overcome this in presenting his own

taxonomy. Since a taxonomy presupposes criteria by which

illocutionary acts can be distinguished, he begins by

listing various of the criteria used in distinguishing

acts. Several of these are of particular importance in

his classification. They are as follows.

There are differences in the purpose or point of

various acts. Searle describes the point of an order as the

attempt to get someone to do something, A consequence of

the illocutionary point is the direction of fit between the

words and the world, This provides his second basis of

classification. In an order we attempt to get the world to

match our words; a statement is a matching of our words to

the world. A further basis of differentiation is the

psychological state expressed. In stating a belief is

expressed, in ordering a desire or want, in apologizing a

regret, and in promising an intention. The state expressed

108

is determined by the sincerity condition (Searle 1969) for

the act.

Searle never defines what it is that a speaker does

when he expresses a certain psychological state. He notes

that a speaker may not have the psychological state he is

said to express in the performance of the act. S does

nonetheless express it. Searle's (197 5, p. 6) argument that

the state is expressed is that it is "linguistically

unacceptable (though not self contradictory) to conjoin the

explicit performative verb with the denial of the expressed

psychological state."

With these data as guides he develops his taxonomy.

The basis for the taxonomy or at least for the classifica-~

tion of an act within the taxonomy is to be found in

Searle's (1969) essential condition for the act. It is the

basis because the essential condition reveals the point or

purpose of the act. I will briefly sketch the alternative

to Austin's taxonomy that Searle offers, and then turn to

criticism of it.

The first class of illocutionary acts he discusses

are the representatives. Acts like stating or insisting or

hypothesizing that P are said to commit the speaker,

although in varying degrees, to the truth of what is

expressed. The psychological state expressed is a belief.

The point of directives, the next class, is to get

the hearer to do something. Included are cases like

109

requesting and ordering and begging. The expressed psycho­

logical state is a want or wish or desire.

Commissives are those acts whose point is to commit

S to some future course of action, the strength of the

commitment varying depending on the act performed. The

expressed psychological state is an intention, and the

illocutionary point determines a world to words direction

of fit. Promising is the only example he provides, but

probably vows and threats would be included.

The whole point of members of the fourth class,

expressives, is to express the psychological state which

is specified in the sincerity condition of the act.

Apologizing and thanking are expressives. Searle claims

that in these cases there is no direction of fit between

the words and the world, but rather the expressed proposi­

tion is a presupposition for the performance of the act.

The point of declarations is to bring about

correspondence between words and the world just through

their successful performance„ If the performance is

successful the change has been brought about. Examples of

declaratives are nominating, marrying (declaring someone

married), resigning, and calling out. It is this class

according to Searle where contact is made with extra-

linguistic institutions which are necessary for the

performance of the act,

110

Searle's taxonomy is an impressive systematization

of facts about illocutionary acts. It is, unlike Austin's,

motivated. It determines the classification of an act off

of the point or purpose of the act which can be found by

checking the essential condition for the act. Schiffer

(1972) finds that there are two basic things speakers do

with language. They try to get others to believe things or

they try to get them to do things. Wittgenstein (1953,

p. 23) says that there are countless uses to which language

can be put. In between stands Searle who finds, as has

been seen, five.

Although his account is more revealing of the

variety of the facts than is Schiffer's and more theoreti­

cally satisfying than Wittgenstein's, the taxonomy that

Searle provides is not without its difficulties. For one,

it puts into different classes acts which are not clearly

of different kinds. At least, his classification does not

show what it is about these acts which provides the

temptation to put them into one category or claim that, at

a minimum, they have a number of things in common. Consider

his class of declarations. He includes as an example of a

sentence which can be used to perform a declaration, "I

declare that your employment is hereby terminated," This

act would, on his.story, have no connection with the act

performed by saying, 111 declare my intention to marry your

daughter," The illocutionary point as he identifies it

Ill

would be different; the latter would be a commissive or a

representative. The expressed psychological state is

different as declaratives have no sincerity conditions, and

hence no associated psychological state. The direction of

fit is not the same. On Searle's basis of classification

there is nothing for them to have in common. They share

only the performative verb which is used in the performance

of the respective acts. This seems wrong. There are

differences; one for example requires the existence of

extra-linguistic institutions for the performance of the

act type. But it does not seem that there is nothing that

they have in common.

Further, his classification of some acts seems to

conflict with points he made earlier, and on which there is

no reason to believe he has changed his mind. Declarations

are said somehow, if they are successfully performed, to

create a certain state of affairs. However, earlier Searle

(1969, pp. 171-182) argues that just to say, "I promise to

pay Jones five dollars," creates an obligation to pay Jones

five dollars. He considers obligation an institutional

concept. But promising is a commissive so there is no way

for the taxonomy to reflect this fact about promising.

The second criticism of his taxonomy is the

following. Classes of illocutionary acts are said to divide

depending on the kind of psychological state expressed.

Representatives have belief as the associated state,

directives have desires or wants, commissives have inten­

tions, and expressives have various pro and con attitudes.

Searle never says what it is to express a psychological

state, but argues that they are expressed by appealing to

Moore's paradox. It is odd for S to assert that P, and to

add a disclaimer to the effect that he does not believe

that P. That is, it is odd to say, "It is raining, but I

don't believe it." But if this is the criterion of

expressing a psychological state many of Searle's other

cases fail the test. The psychological states which are

supposed to help identify the class the act belongs to fail

the test of being expressed by that act. There is nothing

odd about saying, "I order you to go to the front lines,

but I don't want you to go." In the first chapter cases

where not only was it not odd, but it was likely that

something to that effect would be said. Even on first

blush, without any stage setting, this case is not odd.

Even cases which might be odd, like promising to do some­

thing and announcing an intention not to do it, are not

clearly odd in the same way that the case of asserting while

proclaiming non-belief is. Harnish (197 5) has suggested

that the oddity in the promising case might just be the

self-defeating character of the utterance.

It does not seem that Searle has identified any

clear relationship between classes of illocutionary acts

and psychological states. He admits that the speaker need

113

not have the state, only that he must express it, but his

criterion of expressing the state shows that the illocu-

tionary acts in a class in part individuated by that state

do not require expressing the associated psychological

state. Further, the class he says has no sincerity condi­

tion, the condition from which the psychological state is

determined, do seem to pass his test. A speaker who says,

"I declare you man and wife, but I don't intend to," has

run aground of just the oddity that Moore spotted.

Searle's taxonomy ultimately rests on the illocu-

tionary point of the act. Requests and orders both have

the same point; they are both attempts to get the hearer to

do something. But what does it mean to say that requests

have this as their point? Certainly the point or purpose

of each individual act is not to get the hearer to do

something. At least this may not be the speaker's point.

The difficulties of the first chapter in identifying

standard cases would apply to an answer that the standard

or normal cases of requests or orders have this as their

point. It is not clear that a case which does not have

this point is in any way non-standard. It is not clear,

either, that the point of all expressives is to express

the state specified in the sincerity condition. It is

arguable that today the point of welcoming someone has

become a matter of carrying out a certain requirement of

114

etiquette rather than expressing the speaker's pleasure that

someone is present.

Searle tells us that the point is determined by the

essential condition, or at least there is some intimate

relationship between the essential condition and the

illocutionary point. The only example that is provided by

Searle (1969, p, 60) of an essential condition says that the

speaker intends to do a certain thing by producing the

utterance. But none of the things which he describes as

the point are things that the speaker must be doing or

intending to do in an expressive or in a directive, at

least not in any straightforward way of intending.

One attempt to save his notion of the point is to

say that the point of an act type is the point that the act

would sometimes have to have in order for the practice, say,

of ordering or requesting or welcoming to continue. But

why believe that a point which acts of a particular type

needed in order for them to come into existence has

anything to do with the point of that act on each occasion

of its use?

Is not it more reasonable to suppose that points

and purposes are something that speakers have in the

utterance of illocutionary act tokens, and not something

that belongs to the act type? A more fruitful approach

might be to seek the characteristics of certain illocu­

tionary acts in virtue of which speakers can use them for

115

a particular kind of purpose. The available data seem to

suggest that there are a variety of points or purposes that

a speaker might have in the performance of an illocutionary

act which do not correspond to the variety of illocutionary

acts, and hence point or purpose is not an adequate basis

for a taxonomy.

It is not an adequate basis for a taxonomy because

Searle is forced to describe acts as having points which

they do not in any clear sense have. Further, the kinds of

psychological states supposedly expressed by members of

certain classes of illocutionary acts fail his test for

being expressed by that act. Finally, his account separates

acts which appear to belong in the same class. At least

they have something in common, but his taxonomy cannot

explain what that is.

If the account given in the first chapter is

correct, then in performing an illocutionary act S attempts

to get A to entertain a thought. The thought to be enter­

tained can be characterized either as P or as A's ij;-ing. As

far as P is concerned, A can either believe it or fail to

believe it, S may not intend that A believe it, but some­

times an illocutionary act, at least an assertive one, is

intended as a vehicle for getting a person to believe

something. Among the most efficacious means of getting

someone to believe something is giving him a reason for

believing it. Illocutionary acts can fill this need, and

116

the analysis of the first chapter reflects this fact. S

need not intend A believe that P in performing an assertive

illocutionary act, but if he does intend for A to believe

something, he intends by performing the assertive act to

make a reason of a certain kind available to A. Certain

acts share the intended effect, but differ in the reason,

and the reverse holds. It turns out, if what has been said

so far is correct, that illocutionary acts can be

identified by determining the thought to be entertained or

the reasons A is to have for believing that thought.

A natural way to parse assertive illocutionary acts,

then, would be on the basis of how the act is identified.

Within the assertive class there are (F)P, (F)r, and (F)P

and (F) r*- identifiable kinds of illocutionary acts.

In performing an imperative illocutionary act S

intends that A entertain the thought that A ip. S would not

always intend that A actually ip, but sometimes he will.

(A minor digression is needed. When A orders A to 1(1, and

wishes A to \p, he is trying to get A to \p via forming the

intention to ip- S does not intend that his utterance will

produce an impairment in A's nervous system which will

cause A to perform the act, at least not just in virtue of

his act being, say, an order. So when it is said that S

intends A ip, this is to be understood to mean A's tjj-ing

via forming the intention to ip, ijj-ing intentionally.) But

if S intends A to do something a natural way to do it is to

117

give A a reason for ip-ing. Often, but not always, the

reasons will come in a stick or carrot variety.

Both beliefs and intentions to do certain things

are such that they can be changed by reasons. This seems

to be a good part of the motivation for Grice (1957, 1969)

and Schiffer (1972) making beliefs and intentions to do ip

the intended effect in their analyses of speaker meaning.

Schiffer (197 2, p. 57) says reasons are grounds for (which

warrant) the belief. Further (p. 21) reasons are,

ultimately, themselves beliefs A has. The operation is one

belief, P, changing belief Q because A believes the truth of

P warrants Q.

Arguments were given earlier for believing that

imperative illocutionary acts were identifiable in just the

same ways that the assertive illoctuionary acts were.

Dividing imperative illocutionary acts in the way in which

assertive illocutionary acts were, in terms of how they are

individuated, there are six possible classes of illocutionary

acts, three in the assertive class and three in the impera­

tive class, Further, it has been shown that illocutionary

acts can be classified as naturalistic, institutional or

conventional. These are not features of individuations

along the same lines as the (F)P's and (F)r's. Rather, each

act will be identifiable by two of these features, for

example, an act will be said to be naturalistic and (F)P

identifiable. Some of the (F)r's will involve things like

118

S having a belief or having evidence; there will be others

that will be to the effect that S has invoked extra-

linguistic rules, and this is to provide A with a reason.

I turn next to an attempt at analyzing a number of

illocutionary acts. From this a chart can be constructed

placing the various illocutionary acts within classes

according to the taxonomy just suggested. First, however,

the acts must be analyzed to identify the (F)P's and

(F)r's involved.

Analyses of Some Illocutionary Acts

The point of these analyses is two-fold. First,

they will show how an analysis of conventional and

institutional acts can be accomplished within the speaker

meaning framework. Secondly, it will show how various

acts fit into the basic (F)P and (F)r classification. The

discussion should also reveal to some degree how I believe

this kind of semantic analysis must proceed.

Schiffer's (1972, p. 103) analysis of advising is

a good start at an analysis of that act, but is mistaken

in certain ways. He claims that the intended effect is that

A \p twith the modifications herein that A entertain the

thought that A \p) , and that his reason for ip-ing be that it

is in A's best interest to ip. The reason seems too

restrictive. S may realize that it is not in A's interest,

but wish A to iji because it is in accord with moral

119

considerations or even rules of etiquette which in the

circumstances are relevant considerations. The reason is

not that it is in A1s interest, but that in the circum­

stances there are certain reasons why ijj-ing is the best (or

a good) thing to do. Why it is the best will vary.

Ordering differs from advising just in the reason

which S is providing A. If S orders A to ijj, then S intends

it be recognized that he intends part of A's reason for

l^-ing be that in virtue of some relation of authority over

A, S expects A to if». This is not, of course, sufficient.

There are various ways in which S can expect his authority

to be the reason A does what S tells him to do which are

not ordering. S may be recognized as an authority in a

particular field, and tell A to run an experiment in a

certain way, and expect the fact that S is an authority to

be the reason A is to have. Here it would not be true that

S ordered A to run the experiment in a certain way. The

authority must be over A in the sense that A is recognized

as subservient to S at times when each is acting in a

certain capacity. His authority must be in the sense that

he has authority "over" A.

The requisite authority relationship is exemplified

in the relationships where S's authority empowers him to

impose certain sanctions on A if S refuses to carry out the

order. It does not seem, however, that S must intend the

fact that he is empowered to impose a sanction on A be part

120

of A's reason for complying. That such sanctions are

available may not even be necessary for the existence of

orders. People may be sufficiently motivated by con­

siderations of duty to those with power over them that

considerations of possible punishment never arise.

A special difficulty that arises here is that it is

important to keep ordering from collapsing with a number of

other invocations of authority. If it is claimed that S

orders if and only if S means that someone is to do some­

thing, and his reason be a certain institutional authority

between S and A, then might not, for example, excommunicating

or sentencing or annuling meet the requirements of the

schema.

I do not think they do. This is because A's tp—±ng

is supposed to be something that can be brought about by A's

forming the intention to ip (Schiffer 1972) . In the case of

sentencing or annuling or demoting or excommunicating A's

^"-ing is not best described in this way. The point of these

acts is not to create in the hearer the intention to \i>.

These would be cases of S meaning that P, and not meaning

that A was to ip. In the first section of Chapter IV an

attempt is made to give a formal assertive and the impera­

tival intended effect.

Should there be other cases that might meet the

ordering schema Schiffer (197 2, pp. 102-103) provides one

possible way of expanding the reasons clause to rule out

121

such cases. If it becomes necessary it should not be too

difficult to add. I will not, however, use his specifica­

tions of the relationships involved.

I introduce the following abbreviation. If S

R-intended that A ip because of (F)r by uttering x, then S

intended it to be mutually known* (or intended it be

recognized partly because he intended) that S intended in

uttering x that if A ip then part of A's reason for ijj-ing

be (F)r.

Then, S ordered A to ip by uttering x if and only if

S meant by uttering x that A was to ip, and R-intended in

uttering x that A ip because S had a position of authority

over A in virtue of which A's ip-ing is required.

It is too strong to say that A's reason is to be

that he believes it his duty. S might be an employer who

orders an employee out of his office. It is not clearly

correct to say that A has or is to believe he has a duty to

leave the office. It may be that someone might wish to

deny that these are really orders, and to separate orders

from commands along these lines. Perhaps, but I see

nothing that is gained by this.

It might be objected to this analysis that simply

to say that S is entitled to expect compliance by A or that

A's compliance is in some way required leaves too much of

the illocutionary act unanalyzed. How is it that S is

entitled? It is not sufficient to answer this demand by

122

explaining that A is required to x because if A fails to ,

S may impose some sanction since the same sort of question

could be raised about the basis of S's entitlement to

impose a sanction. However, no answer to the demand for an

account of the notion of entitlement is needed since the

demand is really a request for an explication of certain

institutions, and not of illocutionary acts. The correct

account of these institutions within which certain illocu­

tionary acts arise is another matter. I suspect that Hume

was correct in holding that such institutions are

ultimately to be understood in terms of the mutual expecta­

tions about the behavior of the people involved.

Another imperatival act, begging, was partially

handled in the previous chapter. Likewise, it is (F)r-

identifiable. It differs from advising in that there is

not even a pretence that i/j-ing is the best thing to do. S

is not appealing to any authority he has over A even though

he may have that authority. What S may intend as part of

the reason for A is that S has appealed to A for sympathy.

S begged A to 4> in uttering x if and only if S

meant that A was to tp, and by uttering x S R-intended A ip

because S had appealed to A for sympathy (or had demeaned

himself).

Explaining is an assertive illocutionary act, and

is (F)P-identifiable. There is, I think, a distinction to

be drawn between explaining why and explaining how (to).

123

There may be some more general characterization of

I explaining which will capture both cases, but the following

is intended with respect to the former only.

S explained P to A in uttering x if and only if S

meant in uttering x that Q is the reason why (or for) P.

The rationale should be obvious. To say that in

uttering, "The fall killed Jones," S explained Jones's death

is to say the fall is the reason (cause) why Jones died or

the reason for Jones's death. The events or facts referred

to by Q can be determined either by x or by something

contextually specified. A number of different kinds of

things can be reasons depending on the context, the

information the hearer already has, and so on. This

definition lets "reason" run, as it usually does, over

all of them.

It may seem that explaining is also (F)r-identifiable

on something like the following grounds. If S is not

attempting to explain, but only conjecturing on the possi­

bility of Q as the reason for P he also intends that A

entertain the thought that Q is the reason for P. The

difference might be that if S is actually attempting to

explain P, he intends that if A believes (F)P then part of

A's reason for believing be because S believes or has

evidence for Q as the reason why P occurred. This is

plausible because in some cases of conjecture S makes no

pretense that he has any reason or evidence or knowledge

124

that P is explained by Q. If A believes the conjecture at

all, the only reason the conjecturing itself provides is'

that it is a plausible (intended to be one) account.

The reasons for rejecting this are in part based on

the theory of speaker meaning given in this paper. The

analysis of speaker meaning required a conditional connec­

tion between certain reasons and the beliefs of A. In cases

of conjecture if the plausibility is to be part of the

reason A is to have for believing rather than, say, evidence

which the objector feels identifies explaining, then if A

is to believe that (F)P the plausibility of (F)P is to be

part of A's reason for believing. This is doubtful. The

plausibility is intended, perhaps, as a reason for thinking

that Q might explain P, but not for believing that Q

explains P, Perhaps, if S was conjecturing that Q was the

reason for P, what S meant was that it is plausible or

possible that Q is the reason for P. If S intends A to

believe something just on the basis of his conjecture, it

is that it is a plausible account of something else. The

reasons that S might intend A have in conjecturing might

be quite diverse. If conjecturing itself is identified by

(F)P, that P is plausible as an account of something, then

explaining can remain (F)P-identifiable.

There is an alternative and more general account of

conjecturing that would handle all of these data. It may

be that S conjectures that P when S has no reason, or a

125

reason in the circumstances which is not sufficient to

support his claim that P is true. This would explain why,

no matter what intentions S had with respect to the audience

unless he had some reason to support his claim he did

nothing but conjecture. Having support or sufficient

support for one's claims may stand to conjecturing as

having the requisite authority does to, say, excommunicating

If conjecturing is identified by whether in making some

claim one has evidence, then where S proposed something as

a plausible account this is not what makes what he did

conjecturing. It was proposing it without adequate support.

Lurking in the shadows of conjecture was a poten­

tially troublesome problem for the analysis. If all that S

intends is that A entertain a particular thought, and not

that A believe something, then what separates illocutionary

acts like stating or asserting from merely conjecturing

that P or putting forward P as something to consider?

Some of the assertive illocutionary acts can be separated

from conjecturing by the reasons clause in their definitions

but all cannot because of the existence of some (F)P-

identifiable assertive acts. The analysis of conjecturing

given above will prevent the collapse of all of these acts

into conjecturing. To conjecture is not to intend for

someone to entertain the thought that P, but for them to

entertain the thought that P when one lacks grounds or lacks

adequate grounds.

126

S denied that P in uttering x if and only if S meant

that P was not true.

Denying is another (F)P-identifiable act. The

analysis of speaker meaning will insure that some form of

reasons is present in these acts. They will just not figure

into the identification. They might even vary from case to

case.

Some assertive illocutionary acts may be not just

CF)P-identifiable, but merely P-identifiable. Stating is

a candidate for this class, but there is a special diffi­

culty in providing an analysis. In performing a number of

other illocutionary acts S is also capable of being

characterized as stating something. If S describes P as

being red, then S stated that it was red. Hence, in some

cases of stating certain reason are present because of

the illocutionary act which S performed, but not because S

stated. The reason is that some performances are the

performance of more than one act, and certain varieties of

reasons enter the performance and are present there through

one and not the other. This is not the real difficulty.

The real problem arises in this way. First, some

way is needed of separating cases like stating or asserting

from merely saying, I will deal with this in some detail

in later chapters. Briefly, S did not state that Jones is

priggish just because he said, "Jones is priggish." He

might have been joking or he may not have understood what

127

it meant. It would be incorrect in these cases to say S

stated that Jones was priggish. Part of the difference is

that if S said it he did not mean it just as a result of

saying it, but if he stated it then he meant it. Suppose,

however, which seems quite possible, that stating or

asserting is only (F)p-identifiable, in fact merely P-

identifiable. By itself this is no problem. It means that

if S explained that the fall was the reason for Jones's

death he stated that the fall was the reason for Jones's

death. This is plausible. But it seems we must also

accept that if S was greeting, and the analysis of greeting

is roughly that S acknowledges A's presence, then S was i

stating that he was acknowledging A's presence. It is not

clear that if S greeted A that he stated this. In fact it

seems more clear that he did not. It appears that to state

is not just to mean that P by uttering x unless some other

account of this case can be found. Notice, though, in other

cases it is permissible to describe S as having stated what

is specified in the intended effect. If S denies that P it

is not a distortion of what S did to say that he stated that

P is not true, which, of course, is not the same as stating

P and negating it.

The problem can be handled, I think, in one of two

ways. Conventional cases such as greeting do not have a

content. Hence they are not so much (F)P as just F-

identifiable. I suggest that where the content, P, is null

128

it cannot be asserted or stated or told because there is

nothing to assert or state or tell. Further, greeting is

(F)r-identifiable as A is to have as his reason for enter­

taining the thoughts acknowledges his presence that there is

a convention to do this.

Reporting is for Schiffer (1972, p. 101) a reasons-

identifiable act. The reason A is intended to have is that

on the basis of S's observations S believed that P. Schiffer

is appealing to the view that reportings, unlike tellings

or statings, involve observational evidence for P. Before

attempting to determine the extent to which this is true a

certain ambiguity in "reporting" should be noted. Some

cases of reporting, rigidly institutional ones, are essen­

tially cases of filing a report that P. Here S reports

that something is the case partly because there is a

requirement to make the report. In these cases a person

can report even though he has no evidence for the truth of

the report, observational or otherwise, and he is aware

that everyone around him knows this. What follows is not

intended to apply to the cases of reporting that are

essentially matters of filing a report.

If there is no intended recognition of the presence

of observational evidence does the act fail to be a report?

Suppose that S is a scientist convinced of a theory of his.

He has been waiting six months for the occurrence of an

event which will either confirm or disconfirm his theory.

129

Just before the occurrence of the event S is distracted, and

does not observe whether or not it takes place. His

associates know that he was distracted, and did not observe

anything. S is so convinced of his theory he is sure the

event occurred even though he did not observe it. Clearly S

can state to his associates that it occurred. It does not

seem, though, that he can in this case report to them that

it occurred. S can report what he does not have evidence

for, but he at least must intend that his audience think

he has evidence. The following is a first attempt.

S reported that P in uttering x if and only if S

meant that P, and S R-intended by uttering x that A believe

that P because S had evidence for P.

Two qualifications are necessary. The first

concerns the relevant notion of evidence. I am using

"evidence" so that one can claim as evidence for there being

a tree in front of him the fact that he now sees that there

is a tree in front of him.

The other qualification is the more important one.

It is motivated by facts like the following. There is

something peculiar about saying to a friend, "I report that

there is a letter in your mailbox," It is not just using

the performative prefix that produces the oddity because

we would not describe the speaker as reporting to us that

there was a letter in the mailbox. The exception is if it

was S's job or duty to inform A that there was something in

130

the mailbox. Hence, soldiers on the berm surrounding their

defensive position report that there is movement in their

sector, but a husband does not report to a caller that his

wife is in the shower even though he has observational

evidence. He informs or tells or states that she is. A

person is, however, said to report strange goings on next

door. The strangeness is important. S is not said to

report routine occurrences unless there is some reason to

report them. Strangeness is, prima facie, a reason to

report.

The reasons clause of the analysis needs to be

modified to include not just some observational or eviden­

tial requirement, but reference should be made to the fact

that there is a requirement to bring it to A's attention

because of a duty or because it is of special interest to

A, Just to have a label for these kinds of features of duty

or interest they will said to be D-related to A.

S reported that P in uttering x if and only if S

meant in uttering x that P, P is D-related to A, and S

R--intended by uttering x that A believe that P because S

had evidence for P.

It has been suggested by Cowan (1975) that the

evidence requirement can be defeated by an example like the

following. Two people conspire together to make a false

report about a third party. S tells A that even though it

is not true, tonight he will report that Jones was seen

131

talking to enemy agents about our country's latest

philosophical theories. That night S says, "I report that

Jones was talking to enemy agents." Did not S report that

Jones was talking to enemy agents? First, a distinction was

made between cases of filing a report and reporting. If

this is a counter-example to the analysis it must not be

aimed at cases that are essentially filing a report. Assume

that it is not. Then, were other people around or was S

talking to A when A was alone? If they had conspired

earlier to do this, and said it when S was alone with A, it

is not clear that S reported to A that Jones was talking

to enemy agents. He might be saying that he was reporting,

but unless some filing a report sense is intended, he was

not reporting this to A, He was not reporting in the

relevant sense. Suppose other people are around. In this

case we might be willing to say that S did report something.

But the relevant question is who the intended audience of

his remark was. If other people were around, but S said it

to A so low he knew none of them could hear, that is, he

intended his remark only for A whom he had conspired with,

it is not clearly a report. It is when the remark is

intended for the consumption of others who are not in the

know that it is a report. They are the intended audience,

the ones who are to entertain the relevant thoughts.

One further justification can be given for the

requirement of observational evidence. It separates reports

132

from guesses in the most systematic way. To guess is just

not to have the same kind of support that one has, or

claims to have, when reporting.

Predictions share with reports the connection in

some way with evidence. The evidence is part of what

separates them from prophesies as well as from guesses

about the future. On the other hand, the fact that they

are about the future rather than about the present or the

past distinguishes them from reports. It is not clear

whether reports may be about the future, but it is clear

that predictions must be. Some purported reports about

the future are really disguised reports that a person now

has a reason to believe that something will happen in a

certain way in the future. I do not wish to claim that this

is always so, or if it is it is a conceptual rather than a

contingent truth. The requirement that predictions must be

about the future will separate them from certain other acts.

One qualification is necessary in this regard. The

future requirement will be, perhaps, epistemological future.

That is, what is in the future is future only with respect

to our present knowledge. This would account for the

existence of claims like, "I don't know where Hazel went for

her vacation, but from the travel folders she was looking at

X predict she went to Cape Cod." This observation is due to

Cowan (1975),

133

There is one further element of predictions. They

involve the notion of calculating on the basis of the

available evidence. S in some way goes beyond the evidence

by calculation. Again, this is part of what it is to

predict rather than prophesy that something will happen.

S predicted that P in uttering x if and only if S

meant in uttering x that P will occur, and S R-intended by

uttering x that A believe P because on the basis of

observation and calculation S believed that P will occur.

Predicting is both (F)P and (F)r-identifiable.

Promising is a good test case for any theory. It is

a central illocutionary act, and makes contact with certain

institutions as well as sharing features with naturalistic

acts. The initial content appears to be that S will do

something. More is needed as not every case of a speaker

saying that he will do something is a promise. S, if he

promises has obligated himself to ijj, and he has this

obligation because he has promised. It is because of this

that A is entitled to expect S to i|i. However, it appears

to be an assertive rather than an imperatival act. To use

Searle's terminology, the point is not to cause oneself to

do it, but to create a belief in them that you intend to

do it.

S promised that P (P may be that S will ip) in

uttering x if and only if S meant in uttering x that P, and

S R-intended by uttering x that A believe that P (believe

134

that S will ip) because S has undertaken an obligation to ̂

by uttering x.

Two qualifications are necessary with regard to

the correctness of this schema. First, this allows, as it

stands, acts like vowing to also be promises. This is

incorrect. What needs to be done is to place certain

qualifications on the obligation to rule out this. This

too might be done by following the proposal that was made

for ordering noted earlier by Schiffer (1972, pp. 102-103).

The other qualification is that it may be that the creation

of the obligation is more central to promising than this

analysis makes it appear. Does not S intend primarily to

create an obligation? If this is so it might be necessary

to modify the analysis to read that S meant that S had

undertaken an obligation. I do not know if this is what is

best for capturing the facts of promising or not. So, I

will let the original stand as a tentative account, or more

tentative than the other tentative accounts.

As in the case of ordering I am leaving certain

institutional notions untouched. Specifying what it is to

undertake an obligation is something that can be done

independently of what it is to promise.

Promising resembles the institutional illocutionary

acts described above. Promising creates a reason for A to

believe something and for S to do something. But it is

making contact with institutions of which it is not an

135

essential part, namely obligations. This is to assume that

obligations are institutional notions. Promising is

essentially a naturalistic act which uses an antecedently

existing institution as a reason by itself being able to

bring into existence one of these institutional facts as a

reason. Swearing to get revenge may at one time, when

vengeance was a more well-defined social notion, have been

a similar case.

Other analyses would be made in this way. The

identifying conditions are found by determining how the acts

differ from other acts. Warning that P differs from stating

that P, for example, in that it is intended to be recognized

that P constitutes a danger of some sort for A. Figure 1

is a chart of various illocutionary acts showing where they

fit into the classification proposed in the last section.

Moore's Paradox as a Test of Theories

It was noted that Searle appealed to Moore's

paradox to demonstrate that certain psychological states are

expressed in the performance of certain illocutionary acts.

Ee views it as lending support to his theory that he has an

account of Moore's paradox. The speaker is expressing the

psychological state specified in the sincerity condition for

that act, but is denying that state. Hence, the oddity of

saying, "It i's raining, but I don't believe it."

136

(F)P (F)r (F) P & (F) r

Assertive Cases

Naturalistic Explain Deny Object Warn State Conjecture

Report Estimate Predict Prophesy

Institutional Promise Call out Sentence

Bid Christen

Conventional Greet Bid Farewell Introduce

Imperative Cases

Naturalistic Tell Advise Beg

Institutional Order

Figure 1. A Chart of Illocutionary Acts

137

However, it is not clear that Searle has accounted

for Moore's paradox. As pointed out earlier, if the oddity

arises from denying the psychological state expressed in

the performance of the act, then any illocutionary act in

which the psychological state specified in the sincerity

condition is denied should be odd, and odd in the same way.

Further, if something about the performance of the

illocutionary act is the basis of the oddity, the oddity

should be present in various performances of the act. But

it has been pointed out by Harnish (1975) that the oddity

disappears with the presence of the explicit performative.

"I hereby state that it is raining, but I don't believe it,"

is not odd in the way of Moore's paradox. Even what oddity

there is in "Leave the room, but I don't want you to!"

vanishes with the addition of the performative prefix. Both

of the former are genuine cases of stating, and both of the

latter are genuine cases of ordering. Both are illocu­

tionary acts, and it is in the performance of the illocu­

tionary act that Searle's solution is to be found. It

appears that this account of Moore's paradox is wrong, and

cannot be used as support for Searle's theory.

The fact about the non-oddity of the explicit

performative sentence means that one other possible account

will not work, Armstrong (1971) like Grice (1969) modifies

the intended effect to A's believing that S believes that P,

It would contain a natural explanation of Moore's paradox.

138

S would have attempted to get someone to believe that he

believes something, but claimed that he did not believe it.

However, this runs aground of the non-oddity of the

explicit performative. No one, then, has an adequate

speech act account of Moore's paradox. The theory developed

in this dissertation is no exception. It is not clear,

however, that the oddity is to be explained in terms of

speech act information, rather than, say, principles of

conversation. It is not apparent that an explanation

appealing to speech act data can explain the performative-

non-performative discrepancy.

In summary, illocutionary acts that are in some

broad sense conventional do not provide overwhelming

difficulty for an analysis in terms of speaker meaning.

The features of the conventionality of certain acts, and

the ways in which they can be performed raise questions

about the meaning of certain expressions and the illocu­

tionary acts that can be performed with them. The next

chapter will detail some of these relationships, and

consider other accounts of the various other accounts of

the relationships. The first to be discussed argues that

the concept of force can be abandoned entirely in favor of

talk of meaning.

CHAPTER III

UTTERANCE MEANING AND FORCE

A number of questions about the concept of illocu-

tionary force have been raised which we are now in a some­

what better position to answer. The issues that need to be

examined are raised by questions such as the following. Is

there a distinct feature of force? Or is it to be abandoned

in favor of some other feature such as meaning? Is force

something that sentences and expressions themselves can be

described as having or is it only in the use of sentences

and expressions that talk of force arises? How does meaning

relate to force? As they stand these questions, themselves,

may not be clear, and part of the task here is specifying

just what is being asked.

Austin (1962) suggested that utterances have not

only a meaning, but also a force. Part of the difficulty in

evaluating Austin's claim is that it is just not clear

whether he is talking about the utterance object (the

expression as an object for study by a grammar) or the

utterance act, "Utterance" is ambiguous in just this way,

and it is reasonable to suppose that different answers to

various of the questions about force, and about the

meaning-force relationship will be given depending on which

13 9

140

of the senses of "utterance" the remarks are intended to

be about.

A good amount of discussion has taken place on these

issues, and approaches are varied. Searle (1969, p. 18)

says that the study of meaning and the study of speech acts

are in reality the same enterprise. Searle (1969, p, 21)

contends that there is a series of analytic connections

between the concepts of speaker meaning, expression meaning,

and illocutionary force. Cohen (1971) has challenged the

whole notion of force, arguing that it can be abandoned in

favor of the concept of meaning. Part of the problem with

Cohen's account is that it shares with Austin's the lack of

clarity as to the relevant sense of "utterance," The

result, as we shall see, is that his arguments at best show

that we can abandon talk of force in favor of talk of

speaker meaning, Frye (.1973, p, 291) presents a case that

meaning and force, even speaker meaning and force, must be

kept distinct, and she characterizes the difference in her

observation that "what a speaker means by what he utters is

not identical with what he means to do, in the way of an

illocutionary act."

Some of these notions will be untangled along the

way of clarifying the issues involved, I will discuss the

points made by both Cohen and Frye in their articles on

these issues. I will then turn to the relationship between

speaker meaning and force, arguing that the account presented

141

in the first chapter is an adequate way of understanding the

differences that Frye characterizes, and that features of

the analysis given in the first chapter correspond to those

differences. I will then turn to the question of whether,

and in what sense, utterance objects can be said to have a

force.

Cohen

Cohen (.1971) enters the force-meaning dialogue by

challenging the existence of illocutionary force. He argues

that the concept is otiose, that talk of force can be

abandoned in favor of talk of the meaning of the utterance.

He has given arguments which are designed to show both that

utterances do not have a force, and that those arguments

presented in defense of force are inadequate to establish

its existence.

He tackles the former of these tasks by suggesting

that in Austin's sense of meaning there is no way to

identify the force because there is nothing to it except

meaning. Cohen (1971, p. 583) asks in the case of the

explicit performative, "X warn you that . . "But what

locutionary act do we then perform? Where is the meaning

of our utterance as distinct from its illocutionary force?"

The query is as to just where the meaning is to be found if

it is something different than the force of the utterance.

Cohen (1971, p. 584) continues by noting that since the

142

locutionary act is defined for Austin in terms of the

phonetic and phatic acts of uttering certain noises that

belong to a certain language that the addition of the

performative prefix must make a difference to the locu­

tionary act performed. The meaning must be different. If

the performative prefix makes a difference to meaning why

not say that it is meaning and not force that the prefix

makes explicit? Further, if every illocutionary act is

also a locutionary act, then utterances like "I. nominate"

must have a meaning. It is implausible that the meaning the

specific locutionary act disappears with the addition of the

subordinate clause.

In seeking to determine what could have led Austin

so far astray, Cohen examines the reasons that might lead

someone to postulate that utterances might have a force.

Several of his arguments are directed at what can be

regarded as dubious grounds for the existence of illocu­

tionary force, and he can be yielded to on these points.

In other cases, however, his attempted abandonment of force

depends on blurring some important distinctions. I will

concentrate on a number of his conflations. This should

reveal the distinctions which he scrupulously avoids.

He suggests that people might be led to believe in

force because the same utterance can have a number of

different forces, that is, can be either a promise or a

warning or a prediction. In the absence of a device to

143

make the force explicit the force is determined by reference

to contextual considerations. But, notes Cohen (1971, p.

588), exactly the same is true of meaning. The expression

"They're all gold," can mean either they are a certain color

or they all consist of a certain metal. That reference to

context determines that some utterance is a command rather

than a request is, insists, Cohen, a feature of the meaning

of that utterance and not of anything called force.

It is not clear that the cases he compares are

similar. That "I'll be there," might have any one of a

number of forces does not turn on any single feature of the

utterance. In the "gold" case the various possible meanings

do turn on one such feature, the meaning of "gold." If the

hearer does not know which meaning of the utterance is the

one which the utterance has here, this can be determined by

figuring out which sense of "gold" is intended. In the case

of the force disambiguation of, "He'll be there," what is

the comparable feature of the utterance on which the various

possible forces turn. What is the feature of the sentence

in which it is, say, a promise and not a threat.

Another argument Cohen considers is that sometimes

reference to contextual considerations is necessary to

determine which force an utterance has. But Cohen insists

it is not correct to argue from the role played by context

to the existence of force because context plays exactly the

same role in cases of meaning. He notes that the meaning

144

of the utterance, "He hit her," depends on contextual

considerations, namely the reference of the personal pro­

nouns. Why argue, then, for a separate feature of force?

However, here too there is an important difference between

this and the cases of context revealing the force of the

utterance, Again, particular features of the expression

can be identified which are the basis of the fact that

meaning is not clear. These are the references of the

personal pronouns, In the force case even after the

reference of the pronouns has been decided the question of

force remains. In the sentence, "He'll be there," after it

has been determined that "he" refers to Jones and "there"

to South Bend, the question of force remains. Likewise, the

force of the utterance can be determined while these other

questions remain in doubt.

It should begin to be apparent at this point that

when Cohen talks of the meaning of the utterance he is

referring to what the speaker means by the use of that

utterance. It is doubtful that someone who does not know

who "he" refers to on a particular occasion fails to under­

stand the meaning of the sentence in which it occurs or

fail to understand what "he" means. Is there reason to

suppose, asks Frye (1973) that his linguistic competence,

the knowledge he has a grammar would characterize, has

deserted him since the last time he heard a sentence with

145

the pronoun in it? What is not understood is the reference

the speaker intended.

So far Cohen's examples to show that the features

that lead to a belief in a separate feature of force are

also features of meaning have not established a parallel

because unlike the force cases the meaning cases involve

separately identifiable features of the expression used.

But it is not sufficient to save Cohen on this score by

finding ambiguity that depends not on some particular

feature of the sentence, but on the grammatical structure

of the whole sentence, A question that would arise

in such a case about the meaning could be resolved here

too in a way that questions about force cannot. The

description that an adequate grammar would assign to the

utterance would end any question about meaning, but it has

not been established that a grammar should or could resolve

the question of force. This is contra Ziff (1971) who

denies that a grammar could even do the former. This is

because speakers of English would recognize "They witnessed

the shooting of the children," as ambiguous, but not "They

witnessed the shooting of the elephants." Even should this

latter sentence be ambiguous other examples like, "They

witnessed the shooting of the bees," would not be. What

Ziff's points bring up is the need for doing two things.

At some point a distinction must be drawn between what is

knowledge of the language and what is knowledge of the

146

world. Within the former category ways of blocking

certain readings for a sentence must be motivated. Selec­

tion restrictions (Katz 1972) can be motivated as a way of

handling the latter chore.

What Cohen's cases do lead us to is to insist on a

distinction when talking about the meaning of an utterance.

Someone who is trying to resolve the meaning of an

utterance might be trying to determine either what the

expression itself timelessly (Grice 1969) means or what the

speaker means in using those words. Resolution of one does

not involve, necessarily, resolution of the other. Very

often a specification of what the speaker means will be

identical with a specification of what the words mean, but

they are conceptually distinct. Determining what the speaker

means depends, Cohen (1971, p. 589) sees, on the intentions

of the speaker. He notes that the meaning of "He hit her,"

is the meaning the speaker intended it to have. He says

that what underdetermination of force comes to is only

underdetermination of the intended meaning by what was said,

Cohen's claim at most can be, then, that force can

be abandoned in favor of speaker meaning, not in favor of

expression meaning. But the differences in the cases he

provided suggest that just to say this is not to give the

whole story. We turn next to someone who tries to give a

little more of that story.

147

Frye

Frye (1973) is unequivocal in her insistence that

speaker meaning and illocutionary force are distinct. The

difference is to be found in the intended effect of a

speaker's issuance of the utterance. What is common accord­

ing to Frye (197 3, p. 292) to every instance of any illocu­

tionary act is that S intends the audience to recognize that

S intends to perform that act. Common to any force is the

intention that the audience recognize that S intends that

force. She says, "I think it is not too misleading to dub

this aspect of the intended effect the 'illocutionary

aspect.'" The difficulty with this characterization of the

difference is that it does not distinguish an illocutionary

aspect from a meaning aspect. It was argued with respect to

any case of a speaker meaning that P that he intends the

audience to realize that he intends to mean that P.

But there does seem to be something correct in her

statement that there are different aspects to the intended

effect which are involved in clarifying meaning and

clarifying force. Her own example will serve. It can be

determined that someone was making a claim in uttering, "the

boy delivered the speech." It can still be open as to just

what he meant, Bid he give the speech or just bring it to

the office? Or it can be determined that the boy brought

the speech to the office without knowing if the speaker was

claiming, assuming, or guessing that this was so.

148

As Frye (1973, p. 293) points out specifications of

certain aspects of the intended effect reveal force, while

specifications of others reveal meaning, A full specifica­

tion of both of these aspects will reveal both the speaker

force and the speaker meaning.

The account developed in the opening chapter does

some of the work of identifying just what these different

aspects of the intended effect are. For a large number of

illocutionary acts what must be identified to show which

act type the speaker was performing is the F (see Chapter I)

which operates on the P. If a fuller specification is being

given of the F, then what is being clarified is force. If

it is the P that is being given a fuller or more detailed

specification, then meaning is being clarified. A fuller

specification of the entire effect will clarify both.

Suppose S warns A that the bank is in financial

trouble by saying, "The bank might collapse," If the

speaker was warning A, and intended this force, then F

will be in part that P, the bank's collapse or possible

collapse, represents a danger or possible danger for A.

S does not warn A if S supposes the event to be beneficial

to A, Marx did not warn his followers about the demise of

the capitalist state; he predicted it. He might have warned

the bourgeoise, P will be some particular meaning of the

phrase, "the bank might collapse," Suppose that F and P

can be characterized in these ways. To characterize the F

149

is to characterize the act type. If part of the intended

effect is that P represents a danger then the force is that

of a warning rather than, say, a promise. This may not be

sufficient to delimit a particular force, as this might also

be characteristic of some other acts, but suppose for the

moment that it is sufficient. That no doubt remains about

the F, about what force the utterance has does not mean

that all questions about meaning are settled. Did he mean

that there was a structural weakness in the building

occupied by the bank or did he mean that the bank was in an

untenable financial position? Or did he even mean that the

channel wall of the river might cave in? Getting clear on

the F will not help here. One must specify the P.

As it stands this will not guarantee that the force

aspect can be separated from the meaning aspect of every

illocutionary act. Sometimes what needs to be done is to

specify the F that characterizes the reason that A is to

have. This would be required to differentiate a guess from

an estimate. As was said earlier estimates involve calcula­

tions based on certain specific and general knowledge of the

speaker, If not, the person did not estimate that P, but

only guessed. However, when S estimates that P it is not

clear that the requirement that it be considered that he

calculated enters through the intended effect rather than

through the reason.

150

So, what Frye dubs the illocutionary aspect is not

just an aspect of the intended effect, that is, an aspect

of the thought to be entertained. It is an aspect of some­

thing A is intended to realize, and is something which

identifies the act type. It is part of that aspect of

speaker meaning which was characterized in the analysis and

in the schema of an illocutionary act as F, In this way the

view that force can be analyzed in terms of speaker meaning

is preserved while not abandoning the insight that there is

something different about identifying the force of an

utterance and identifying its meaning although a full

specification of meaning (of the speaker meaning) will give

us both. When asking for meaning what is usually being

asked for is that part of the thought to be entertained that

has been characterized as P. When asking for force what is

being asked for is something about the way in which P is to

be taken. This might involve the way it modifies P or the

reasons it produces or purports to produce in support of P»

The apparent remaining difficulty for this explica­

tion is in the cases where there is nothing to the per­

formance of the illocutionary act over and above meaning

that P. There may be no such acts, but it will be argued

in the next section that telling someone that something is

the case and telling someone to do something are good

candidates for acts that are identifiable in just this way.

That apparent difficulty is this: Suppose it is known that

151

S meant that Jones is a thief, but that it is still open as

to just what illocutionary act S performed. If to perform

the illocutionary act is just to mean that P how can there

be any doubt about which illocutionary act was performed.

What is not known in this case is whether there is

some modification of the P or some special reason for

believing that P which S intends. If there is none, then

the FP is just P itself, and specifying the force would be

to specify that S just meant that P, and not, say, that P

represents a danger for A or that P is not true. This is

often reflected in our language in statements like, "I was

not blaming you, I was just telling you what happened,"

when replying to someone who does not know how the

utterance is to be taken.

An account of speaker force has been provided which

reveals how it can be that to perform an illocutionary act

just is to mean something, but how it is that specifications

of meaning and specifications of force are sometimes taken

to be two different things. I would like to turn to

another question about force now. Do utterance object

types, that is, such items as sentences themselves have a

force? Or is such talk in some way confused?

Utterance Force

Throughout the remainder of this chapter when I

speak of the meaning or the force of an utterance or an

152

expression I am referring not to the speaker's use of that

expression, but to the expression type as an object in a

language, as an entity to which a grammar would assign a

structural description and a meaning specification. The

question is whether utterances, in this sense, have a

force, and if they do how they have it. Do they have a

force in something like the way in which they have a

meaning? Or is talk of expression force just a roundabout

way of saying that the speaker intended a certain force in

using it.

One consideration in opposition to the view that

they do is the fact that if the utterance has a force this

almost always underdetermines speaker force because of the

wide range of illocutionary acts which a speaker might be

performing with any given utterance, What force could the

utterance have which, even when the speaker does mean what

the utterance means, allows such variation in force? The

consideration is not overwhelming. It is not clear that

this situation with respect to force is any different than

the situation is with respect to meaning. Does not the

speaker's meaning almost always go beyond the utterance

meaning by its inclusion of an intended reference, a time

indication, a particular reading of ambiguous terms, and so

on. No one takes this as evidence that utterances do not

have meaning. So why should the pervasive underdetermina*-

tion show that utterances do not have a force or that there

153

is any major discrepancy between the way that utterances

have a force and the way in which they have a meaning.

There is one general kind of reason which can be

given to suggest that utterances do have some or other force.

The sentence which is used does seem to restrict the range

of possible speaker forces that might be present, barring

rather special circumstances. For example, "I'll be there,"

can quite naturally be used to promise or predict or warn,

but not to question or command without rather special stage

setting. The simplest, though clearly not the only,

explanation of this would be to say that certain forces or

some general force attaches to the utterance because of

features of the utterance itself.

I would like to suggest one rather general reason to

suppose that utterances might have force. It involves

appealing to a rather popular account of the way in which

they have meaning. Briefly, that account is that the basic

concept in a theory of meaning is speaker meaning, and that

utterance meaning is to be understood as conventionalized

speaker meaning. Utterance meaning on this view is a matter

of convention. More will be said directly about the way

this view takes meaning to be conventional. I am thinking

here of views like those of Schiffer (1972), The argument

will be that force can be conventionalized in certain

utterances in the same way that meaning can. Certain

utterances are conventionalized ways of performing certain

154

illocutionary acts in the same way that certain utterances

are conventionalized ways of meaning certain things. If

the existence of such conventions with respect to meaning

is sufficient to say that utterances have a certain meaning,

then they should also be sufficient for saying that the

utterances have a certain force.

I will begin by outlining the type of view of

utterance meaning that is of interest here. I believe that

something like it is probably correct, but I will not spend

time arguing that it is the correct view of utterance mean­

ing, Nor will I spend time with the details of such an

account.

Such an account of utterance meaning takes speaker

meaning to be the logically prior of the two, Schiffer

(197 2, p. 7) gives two reasons why speaker meaning should be

regarded as the basic concept. First, we know that a whole

utterance type means, he claims, only if we know what S

(.in general) would mean by it. The problem with this

argument is that there are any number of utterance types

which we understand in the sense of knowing their meaning,

but do not know what a speaker would mean by their

utterance. Examples might be, "I am dead," and "I am not

now speaking English."

His other argument is that we can mean something by

an utterance x even though x has no meaning, Perhaps this

is not conclusive, but it is a better argument. It is open

155

to reply that we understand something by such an x only

because there is something which does have meaning for which

it is doing duty, but at least the argument shifts the

burden of proof. The fact must be accounted for.

Another reason for adopting the view that speaker

meaning is prior is a sort of theoretical simplicity.

Speaker meaning lends itself to analysis in terms of

psychological notions such as intentions and beliefs, A

minimum in the way of semantical notions or the like will

be necessary in the basis of the theory.

In what way, then, can expression meaning be under­

stood in terms of speaker meaning? Schiffer (.1972) provides

such an account of expression meaning, Schiffer notes that

it is not sufficient for x to mean P that it be an efficient

way for someone to mean that P within a group. This is

because S may be able to mean that P because of some natural

feature of x. For example, x may be "grr," and be used by

someone to mean (non-natural) that he is angry. He may

intend that people believe that he is angry because "grr"

resembles the sound that certain animals make when they are

angry. Then, although "grr" can be used to mean that one is

angry no one would say that "grr" means that one is angry.

Schiffer says there must be a practice or precedent of

meaning P by x, but that this is not sufficient. The

practice or precedent of meaning P by x must figure into

the reason that S expects them to have for understanding

156

that he means that P. That is, there might be a practice,

but everyone present is ignorant of it, and thinks that

something else will allow everyone to see that he means P.

Each may think that he is the only one who is aware of the

practice so it is not enough that each is aware of the

practice or precedent. It must be mutually known* that

there is a practice or precedent. The situation is

complicated by the fact that more than one utterance type

might mean that P or that x might have more than one

meaning. What these facts require adding to the analysis

is that there is some agreement or public acceptance of the

fact that if someone utters x he may mean that P. Schiffer

strengthens each stage of this argument by providing a

counter-example to show that anything less is not suffi­

cient, As we shall see in Chapter IV it is still not

sufficient. Schiffer (1972, p, 130) arrives at the follow­

ing definition of utterance meaning for non-composite whole

utterance types,

x (nc) means "P" in G iff it is mutual knowledge* amongst the members of G that

(1) if almost any member of G utters something M-intending to produce in some other member of G the activated belief that P, then what he utters might be x;

(2) if any member of G utters x M-intending to produce in some other member of G the activated belief that P, he will intend the state of affairs E (which he intends to realize by. uttering x) to include the fact that x is such that there is a precedent in G for uttering x and meaning that P (or an agreement (or stipulation) in G that x may be uttered to mean thereby that P).

157

Schiffer then goes on to show how a convention might

arise and perpetuate itself among a group of people. He

does this by showing it would arise as a solution to a

recurrent co-ordination problem. The following is the

account of a convention that Schiffer (1972, p. 154)

develops.

There prevails in G a convention to do an act (or activity) of type X when (or only when) . , . iff it is mutual knowledge* amongst the members of G that

(1) there is a precedent in G for doing X, or an agreement or stipulation that one will do X, when (or only when) . . . , (2) on the basis (in part) of (1), almost everyone in G expects almost every­one in G to do X when (or only when) . . . , (3) because of (2) almost everyone in G does x when ( o r o n l y w h e n ) . . . .

Schiffer believes that there being a convention in G to

utter x when or only when one means thereby that P is

sufficient for x to mean that P in G. However, he does not

feel that it is necessary condition of x meaning that P that

there be such a convention. There are he believes, though,

a number of senses of "convention" so that if x means that

P in G, x will be a conventional way to mean that P.

The view that utterances themselves have a force

which might be conventionalized in a way such as the above

is more plausible with certain kinds of cases. These are

the cases of explicit performatives, and the cases where a

small number of formulae have developed for the performance

of the act. An example of the latter might be "I am

158

sorry," which is used for apologizing. It might be argued

that this meets the conditions of being a conventional way

of apologizing. S may not actually be apologizing when he

utters "I am sorry," but only adopting a slang way of

describing his own character, but this is no more of a

problem than are cases where a speaker does net mean what

his words mean. That S might not use the utterance with

its conventional force no more shows it does not have such

force than parallel cases would show that utterances do not

have a certain meaning. If a speaker uses an expression

like, "I am sorry," or "I apologize," the natural presump­

tion is that the person did apologize.

Is there a convention, on Schiffer's analysis, to

utter "I apologize," in order to apologize. If there is,

the following must be the case: (1) there exists a precedent

for uttering "I apologize," to apologize or an agreement or

stipulation to utter it when (or only when) apologizing;

(2) on the basis (in part) of (1) almost everyone in G

expects almost everyone in G to utter "I apologize," when

(or only when) they are apologizing; (3) because of (2)

almost everyone in G utters "I apologize," when (or only

when) apologizing.

It is reasonable to suppose there does exist such

a precedent or agreement. And if there is a precedent or

agreement to apologize by uttering the performative then

it would be expected that people would apologize in this

159

way. This is because apologizing, in virtue of being an

illocutionary act, is such that if a speaker apologizes he

wishes someone to know what he is doing. S would wish to

make it clear that apologizing is what he were doing if he

were apologizing. If there is a precedent or agreement

which is mutually known* to utter a certain phrase to

apologize then others would expect S to use this when he

wished to make it known he was apologizing, that is, when

he was apologizing. This is because acting in accord with

a precedent is an obvious way to get people to see what one

is doing, and people expect people to do the obvious natural

way of accomplishing something if they expect others to see

this is the obvious way. If S wishes to make people realize

what he intends to do it is reasonable for him to act in

the way they expect him to act in order to get them to

realize what he was doing. In this case what it is they

expect is for S to utter, "I apologize," in order to

apologize. So it is reasonable to suppose this is what S

would utter, and utter because people expect him to. This

assumes that people do what it is reasonable to do, but this

seems a safe enough assumption in these contexts.

Hence the requirements for there being a convention

to utter, "I apologize," in order to apologize are met. By

the conditional argument set up earlier the force of

apologizing would attach to this utterance. It may be that

the force attaches to the utterance in a more indirect way,

160

The convention may be to mean something by uttering the

phrase. To apologize is just to mean this. So the force

would attach because of the meaning of the phrase. I will

return to this later.

Although, then, there is the same reason to believe

these utterances have a force that there is to believe they

have a meaning, if the hypothesis is correct, the situation

becomes more difficult when we move to other cases. What

force attaches to an utterance like, "I'll be there."

If such utterances have a force, two possibilities

emerge as to how. It may be that such sentences have as

their force every speaker force which S might normally

intend in the literal utterance of the sentence. Hence, the

sentence above has the force of a warning, a promise, a

statement, and so on. Admittedly it may be used to perform

any of these acts, but it does not seem correct to say

that it is a conventional way of performing any of them any

more than saying, "Yes" is a conventional way of meaning

"I've decided to buy the boat." This is what a speaker may

mean when he says, "Yes," but clearly "yes" does not

timelessly mean this. Likewise it seems reasonable to

suppose that the sentence above does not have all of the

forces which a speaker in its normal utterance might invest

it with.

The other possibility is that some generic force

attaches to the expression. For example, the force which

161

attaches to, "I'll be there," might be the force of telling

(someone that something is the case). Then, even if the

speaker force was that of a warning S was also telling A

that S would be there. That S was also telling in these

cases does not seem incorrect. In fact, someone might very

well describe the situation by saying that in telling A he

would be there S was threatening A or promising A that he

would be there. In each of these illocutionary acts S was

telling A something. Other generic forces would have to be

found for those expressions which do not appear to be

tellings. "Tell" unfortunately is ambiguous. In one

sense telling someone to do something is more or less

equivalent to commanding. "He didn't advise me to quit, he

told me to." On the other hand, "When he advised me he told

me to take practical courses," is a meaningful sentence. I

wish to use "tell" in a sufficiently weak sense where if S

told A something or told A to do something what is meant is

that S related such and such to A. I intend nothing stronger

than this.

Cases where it is correct to say that in performing

a certain illocutionary act S also told A something are

numerous. In apologi2ing by saying, "I'm sorry," S told A

he was sorry. If S threatens A by saying, "I'll make your

nose look like a pizza," S told A that he would make A's

nose look like a pizza. Likewise, if S predicted that P,

then S told A that P.

162

Considerations like these lend plausibility to the

view that some non-performative utterances have one or more

of a variety of generic forces that attach to them, forces

which are not incompatible with whatever additional speaker

force might be present. The situation is analogous to the

case where S meant more than his words meant. He meant

what they meant, that he would be there, but he also meant

more, meant that he would be at the party. There are dif­

ferences. We do not describe the utterance as having a

generic meaning which it shares with a number of other

utterance objects. The difference is not crucial.

In both the cases of utterance meaning and of

utterance force the speaker meaning and the speaker force

may not include the utterance meaning or the utterance

force. If a speaker is being sarcastic then a specification

of what he meant will not include a specification of what

the utterance he used meant. S, a revolutionary, says that

Gerald Ford is a leftist, but means that he is a bourgeois

revisionist running dog lackey of the ruling class. Or in a

case of metaphor S may not mean just what his words mean.

There are, likewise, cases of speaker force which do not

include utterance force; these no more show that the

utterance does not have this force than the above cases

show that the utterance does not have the meaning which it

has.

163

If utterances have a certain force, a force which

might individuate a particular illocutionary act such as

telling someone that something is the case, and the account

of act-type identification herein is correct, then this

illocutionary act should be identifiable by (F)P or (F)r or

by both. Further, this (F)P or (F)r cannot be such that

its presence is incompatible with the (F)P or (F)r which

identifies the further illocutionary act such as promising

or warning that S might be performing. This is unless in

some cases the.specific illocutionary act does block the

presence of the act of telling in which case an explanation

is needed as to how the (F)P and (F)r of the generic act

are modified.

I suggest as a first attempt at specifying the

generic force of simple indicatives, that the act is

(F)P-identifiable, and that (F)P is P itself. Suppose that

this generic force is the force of telling (that), in the

sense explained above. Then, S told A that P if and only

if S meant that P. Since the analysis of speaker meaning

itself guarantees that there is some (F)r, then there will

be some reason A is intended to have, but it will need only

to be such to meet the requirements imposed on reasons in

speaker meaning (Schiffer 197 2, p. 57). It will not rule

out any specific characterization of the reason that A is

to have.

164

Suppose that S issues an utterance which has this

generic force of telling, but that the illocutionary act he

performs goes beyond this force. The following case should

suffice for an example. Suppose that S is objection to an

assertion by A that Jones is an honest man when S says that,

"Jones spent four years in jail for robbery." The generic

type of the utterance is telling, and it seems that if S

objected this, then he did tell A that Jones spent four

years in jail. But if S objected S meant that the fact that

Jones spent four years in jail is a reason to think another

(contextually specified) claim is false. Here that claim

is that Jones is an honest man. As noted in the first

chapter there is an ambiguity in "is a reason to think

another claim is false." In the appropriate sense if S

meant that the fact that Jones spent four years in jail is

a reason to think Jones is not an honest man, S meant that

Jones spent four years in jail. Hence S told A that Jones

spent four years in jail.

In this case it is possible to construct from what

S told A and from intentions of S which are not revealed

just by what S said that S was objecting. In other cases

the specific act will be determinable solely from what

S told A.

Suppose that S tells A that it is not the case that

truth is beauty, that is, S says, "Truth is not beauty."

What S told A, the P, has the form of the (F)P that

165

identifies denying. So, just from what was said, and that

S meant what was said, it can be determined that S denied

that truth is beauty.

The general description seems to get the facts

right. For each of the (P)P-identifiable illocutionary acts

S can perform that act just by telling A something. It

can be determined from the (F)P of the act in question just

what is necessary to tell someone in order to perform the

act.

The assumption of generic forces also has another

feature which speaks to its correctness. In cases like

objecting or replying or denying it seems inadequate to

describe what S did as telling something even though it is

not strictly false. This account suggests why just saying

that S told something is inadequate. S did more than just

tell. But it does not follow from this that S did not

also tell.

It would be a mistake to expect each specific act

type to fall under a particular generic act type. There may

be much that would be found in the way of cross genera, but

the work in this area remains to be done.

There remain some supporters of a view that holds

because the conditions appropriate for the use of one

description are not the same conditions appropriate for the

use of another, the one cannot belong in the analysis of

166

the other. The point would have relevance here, and bears

examination.

S and A are at a baseball game. A had gone for a

hot dog, and on returning wishes to know what the last

batter up did. The batter had hit a homerun. S says to A

that the batter got a hit. The rules define a homerun as a

hit where the player is able to touch all four bases without

being tagged out either because he had hit a fair ball out

of the playing field or had hit a fair ball into the playing

field, and had touched all four bases without being tagged

out, and there was no error. S did not say anything false.

His remark is felt to be inappropriate because he did not

say enough. If all A were interested in was computing

batting averages then S's answer that the batter got a hit

would not seem inadequate. Likewise, in the cases above to

say that S told A such and such is not so much to say to

something false about to the situation, but merely to

underdescribe it.

It is just the feature that would lead some to say

that it is wrong to say of someone that the person believed

it when he knew it. But that it is inappropriate to say

this does not show that if S knows that P he does not also

believe that P. Nor would it show that if S warned A that

P he did not tell A that P.

Anyone who wishes to argue from the fact that a

certain description is inappropriate to the fact that it

167

cannot be implied by an appropriate description or a part of

an appropriate description must come to grips with cases

like the following. S knows that Ortcutt is a spy. One

of Ortcutt's operatives comes to S, and asks if S believes

that Ortcutt is a spy. Thinking that this may be one of

Ortcutt's men S answers that it is not the case that he

believes Ortcutt to be a spy. If knowing does not imply

believing, then S has not lied for he knows that Ortcutt is

a spy; he does not believe it. But it seems perfectly

clear that S did lie. This is just how we would describe

the situation.

Note that my case is that it is not the case that S

believes that P rather than S believes that not-P. On the

latter what would be shown if it is agreed that S lied is

not that knowing that P entails believing that P, but that

knowing that P entails not believing that not-P. (This was

pointed out by Byerly [1975],)

The situation with various speech acts is the same.

The problem with saying that S told something is that he did

more, and A may wish or need to know the more. It no more

follows that S did not tell than it follows the batter did

not get a hit when he hit a homerun, The presence of generic

forces which attach to utterance types would seem to explain

a great deal of illocutionary act phenomena. It captures a

wide range of disparate facts.

168

Operating on the (F)P is not the only way that a

generic force becomes more specific. Sometimes it is by

adding to the reasons that A is to have. This would not

affect the fact that S told since all telling requires is

that there be some reason. S says, "I'll be there." S is

promising. It seems that despite the increased complexity

of the reasons clause S still did tell A that he would be

there. If a third party, U, asked A whether S told him

that he would be there A would reply that S had. Similarly,

S could have told A he was sorry when apologizing, and told

A that P would occur when predicting.

It is, as has been suggested, harder to find any

generic force associated with imperatival acts. At least

it is not clear that the force delimits an actual illocu­

tionary act. It would not be too surprising if this were

so. Schiffer was unable to find any imperatival acts

which were (F) ijj-identif iable. The content, the thought to

be entertained in most cases, was A's iji-ing. The signifi­

cant factor in individuating the imperative illocutionary

act types is the different reasons being provided for A's

i|^-ing. If there were a generic imperatival force that were

analogous to the assertive force it would be the case only

that S meant that A was to ip. The form of the reason would

have only to meet the minimal conditions for speaker

meaning, the utterance of x providing not just a cause, but

a reason for A, Given the rather rich variety of reasons

169

that can be provided for A's ip-ing when S means that A is

to, such reasons as those provided by ordering or begging

or demanding or pleading or advising or inviting it would

not be odd that there were no illocutionary act of just

meaning that A was to ip.

This would not mean that there would be no generic

force attached to an imperatival utterance such as "Shut

the door!" It would be the force of meaning that the

hearer was to shut the door. It would just be the case

that this would never exhaust speaker force.

However, I would like to argue that one force which

marks off an actual illocutionary act meets or at least

approximates the requirements for being the generic force

for a number of imperatival utterances. It is the force of

telling (to), of S telling A to ip. Again because of the

ambiguity of "tell," it is important to realize that I am

speaking only of the weak sense of "tell" already specified.

S utters to A, "Shut the door!," and means it,

although he may mean more. Then S told A to shut the door.

Suppose S had been ordering A when he said this. It does

not seem incorrect that when S told A to shut the door he

was giving an order. And it is not, as Cohen suggests, that

he was implying or suggesting that he was ordering. Similar

considerations apply with demanding. If S demands that A

shut the door by uttering the imperative sentence, it does

170

not seem incorrect to say that he told A (related to A)

that A was to shut the door.

However, a few snags develop when we consider

advising. There is some reason here for believing that it

is wrong to say that S told someone to do something when he

advised. Part of the feeling of oddity may be that advising

is usually contrasted with the, or one of the, strong sense

of "tell" so we expect a contrasting sense of "tell" to be

operative. However, we do sometimes advise with the simple

imperative sentences with which the force of telling is

associated, and sometimes describe what someone did when

they told us to do something as advising us. Suppose S is

a wise man to whom A has gone in order to determine what to

do with his life, S says, "Think pure thoughts!" A might

very well describe the encounter to a friend by saying, "I

asked the nut to advise me on what to do with my life, and

he told me to think pure thoughts." S advised A, but also

told A to do something. This last case exhibits the weak

imperatival sense of "tell" effectively.

It is not, it should be noted, directly a conse--

quence of the fact that simple imperatival utterances have

this telling force that every imperatival illocutionary act

which is performed has this force. The thesis is only that

those performed by the utterance of a simple imperatival

sentence do. All imperative acts might, but this would be

because of what S meant, and not from the formula used

171

unless that formula also has the force of telling. This

provides one way around a potentially troublesome case.

If S begs A to ip, then it would only be in very

unusual circumstances that anyone would say that S told A to

\p. The difficulty is resolvable in this way. The force is

supposed to attach to particular expressions, and would

emerge as the speaker's force in a literal uttering of that

expression. It is not obvious the force of telling would

attach to begging because it is not obvious that a speaker

can beg by issuing an imperatival sentence if the intonation

is that usually associated with imperatival forms.

That a force attaches to various utterances is not

an implausible view. There is some reason to believe that

simple non-performative indicative sentences have the force

of telling (that), simple imperative sentences have the

force of telling (to). Although other literal uses of the

sentences can be the performance of further illocutionary

acts, the force of telling does not disappear. When

ordering someone to shut the door by using an imperative

sentence S tells that person to shut the door. When S

advised A on what to do to overcome his vitamin deficiency

by saying, "Eat plenty of oranges!," S told A to eat plenty

of oranges.

Possibly in cases of advising where we describe S as

telling someone to do something in giving a piece of advice

S told A only conditionally. In advising A on what to do to

172

be respected by his fellows, S told A to do such and such,

but what should be said here is that S told him that in

order to be respected by his fellows then he should do

such and such.

It does not follow just because when S orders by

literally uttering an expression that has the telling force

he tells someone to do something that in every single case

of ordering someone to do something S tells the person to

do something. Nothing follows directly to the effect that

every order has telling as its force. Only those where a

simple imperative is uttered literally, where a speaker

means at least what is meant by his words, does it follow

as a consequence of the force that it has been argued above

attaches to certain sentences. However, I think that the

speaker force of the more specific acts includes the

generic force in almost every case. This is a consequence

of what it is to tell someone that something is the case.

Telling (that) will be the case I will work with.

It was argued that telling is (F)P-r identifiable, and that

the (F)P is just P itself, S told someone that P if and

only if S meant that P. In every assertive illocutionary

act S meant that (F)P. But this satisfies the requirement

of at least telling someone that (F)P. To explain P, in

the sense of offering an explanation for P, it was argued

that S must mean that Q is the reason for P. For instance,

if S explains to A that a thyroid condition is the reason

173

for the weight problem A has, then S told A that a thyroid

condition is the reason for A's weight problem. To deter­

mine in an illocutionary act what was told and what was not

look at how much is included in the (F)P. Telling does not

reach into the (F)r. Hence in estimating S did not tell A

that he had calculated the results although an estimate

involves this.

However, even some assertive illocutionary acts are

such that it seems not only that someone has underdescribed

what is going on when he describes S as having told A that

P, but that he has misdescribed it. Telling is simply too

strong. Suggesting that P is a case in point. If S

suggested that P, then S did not tell anyone that P.

If the earlier work is correct, then what it is to

offer a suggestion is to mean that P is a possible or P is

plausible or something to this effect. So, when S suggest

that P he does not mean that P. He means that P is possible

or likely or something to that effect.

S may offer a suggestion by using a simple sentence

which it was argued earlier has the force of telling, for

example, "Jones is the crook." However, it would not be

the case that S told A Jones was the crook if S was just

offering a suggestion. Although the utterance might mean

that, Jones does not, and although the utterance might have

that force Jones does not perform that illocutionary act.

S meant something to the effect that possibly Jones is a

174

crook. This is what he told A. Further, it is likely that

it is not possible to suggest that Jones is the crook by

using the simple indicative sentence, "Jones is the crook,"

if the speaker uses normal intonation. In fact it does not

appear it can be done with that sentence at all. But, if

it can be at least S must be using non-normal intonation.

Suppose that the expression is not a simple indica­

tive, but an explicit performative sentence or a sentence

with some other conventional device. What is meant is

effected by the presence of the performative verb. What

S's denying that P to A means is that P is not the case.

It is plausible to offer that S told A that P is not the

case. But note what the presence of "deny" does if S

utters "I deny that P," as far as determining what S told.

This operation on P seems to be part of the conventional

force that attaches to "denying."

In summary, the suggestion offered here is that it

is plausible to say that some force attaches to certain

expressions. That force gets conventionalized seems as

likely as that meaning gets conventionalized, and if

this is sufficient for saying that an expression has meaning

there is no reason to deny the expression has a force. In

the case of expressions which vary significantly as far as

the possible illocutionary acts which a speaker might

actually be performing in the literal utterance of that

expression the force is minimal, and will meet only the

175

requirement of being P-identifiable. Telling, in an

appropriately weak sense corresponding to relating that, is

a candidate for the force which attaches to simple indica­

tive sentences. The expressions have force, but it appears

they have it in virtue of meaning that P, And, on the view

of expression meaning presented here meaning that P for an

expression is just a matter of being a conventional way for

a speaker to mean that P. The force that non-performative

sentences or sentences with no other conventional device

which determines a specific force have in in virtue of the

fact that there is some P that the sentence means, that is,

in virtue of the sentence being meaningful. The telling

(to) cases run essentially the same way, but here that

meaning is that someone is to do something. "Shut the

door!" on this view would mean that someone is to shut the

door. In these cases conventionalized force or at least

the conventionalized force that a particular expression has

can be accounted for in terms of the meaning that the

expression has. And that is to be accounted for in terms

of speaker meaning,

How adequate is such an account for the increased

complexities of the performative and other sentences with

conventional devices? There are significant similarities.

The performative verb can be described in some cases as

conveying a certain meaning. See the example of "deny"

176

above. To say "I deny that P," seems to be a conventional

way of meaning that it is not the case that P.

The situation is not always this simple. It is not

just the expression meaning having the form (F)P which will

provide speaker force identification. Sometimes (F)r will

provide speaker force identification. But there is no

reason not to believe that a certain reasons clause is not

also conventionalized by a particular performative verb.

Ordering is a conventionalized way of appealing to the

relationship of authority that S has over A as a reason for

A to do something. The conventionalized force of "I

order," is that of an order, but this requires reference to

the reasons clause. But the reason has a slot within the

analysis of speaker meaning, and there is no reason not to

suppose that that part can be conventionalized also. The

next chapter will offer an account of how such information

can be exhibited in a representation of the meaning of the

expression.

Speaker Meaning, Expression Meaning, and Communication

How, then, does literal speaker meaning relate to

expression meaning and literal speaker force to expression

force? I will use p. 54 as the definition of speaker

meaning. It should be clear how the story would go with

the others. It was suggested that x means that P if it is

the case that it would be expected that anyone uttering x

177

might mean that P. What it is for a speaker to mean that P

is to intend in a certain way, by getting the audience to

recognize that this is what the speaker intends, to get

the audience to entertain the thought that P is true. So

what is expected if S utters x is that he intends someone

entertain a certain thought by recognizing that S so

intends. A specification of the meaning of x would be a

specification of the thought that a speaker would intend

his audience to entertain when he uttered x, S will mean

what the expression he uses means when S intends the

audience to entertain the thought that people would expect

in virtue of the fact he said x for them to entertain.

The speaker's force would be related to the utterance

force in the same way. To specify the force of an utterance

is to specify the force that it would be expected that

anyone who uttered the expression would intend. And, force

depends, in ways already specified, on meaning.

Suppose S utters, "I warn you that the bull is

about to charge." It would be mutually expected that

anyone who uttered this sentence would intend in the

specified way his audience to entertain the thought that

the bull is about to charge, and that this represents a

danger for the audience. Suppose that what S intends is

just what people would expect, that is, S intends that they

entertain the thought that the bull is about to charge, and

that this represents a danger to the audience. To mean

178

this is just what it is to warn someone that the bull is

about to charge. This is what S did.

From the hearer's point of view what it is to under­

stand the meaning of an expression in a language is to

understand what thought the rest of the linguistic community

would expect someone who uttered that expression might mean,

that is what thought a speaker would intend the audience to

understand via recognizing that this is what he intends.

To understand what a speaker meant is to recognize the

thought that he actually did intend the audience to enter­

tain by recognizing he so intended. A understands what the

speaker meant just in case the speaker's intention is

recognized. S intends A recognize that the bull is about

to charge, and that this represents a danger for A. S

utters the performative sentence, "I warn you that the bull

is about to charge." From the expression used this is what

A expects S might intend. Recognizing that S does so intend

is to understand what S meant. To mean this is just what it

is to warn, A has understood that S was warning A.

This account explains how it is that the meaning of

the expression used seems so basic in any account of

communication. Very often what the speaker meant is deter•<-

mined in part by knowing what the expression meant. This is

because the expression is a very good guide to the speaker's

intentions. It reveals what intentions to expect him to

179

have, and what intentions he expects the audience to expect

him to have. Still, speaker meaning is the basic concept.

A good deal remains to be said about the subject.

In those cases where S intends more or even something dif­

ferent than what his words would lead an audience to expect

S intended, how does the audience determine what S's

intentions are, and how does the speaker hope to achieve

this understanding in A? This is a large and difficult

question. In the next chapter I will discuss one class of

such cases.

Before turning to that question I will deal with

another question. Hopefully, it will let us get clear on

how much needs to be said before an answer to the question

in the preceding paragraph can be given. How much of the

illocutionary act information would be represented in a

grammar of a language, and just how might a grammar repre­

sent that information?

CHAPTER IV

ILLOCUTIONARY ACTS AND LANGUAGE

So far I have offered an analysis of the concept of

an illocutionary act, and explicated some of the relation­

ships between meaning and force. Next I will turn to the

more general question of the location of a theory of

illocutionary acts within a general view of language, noting

the connections between illocutionary acts and other facets

of language. I will start with the question of where speech

act information belongs in a theory of language, and from

that turn to the question of how illocutionary acts are

related to other phenomena that must be accounted for within

a theory of language use, I will center my discussion of

this latter issue ..on one topic, accounting for the possi*~

bility of indirect speech acts.

Illocutionary Acts and Grammar

A start has been made at determining answers to

these questions, Specifically, the relationship between

illocutionary acts as instances of speaker meaning and the

meaning of expressions in a language has been discussed, I

would like to continue in this area investigating the amount

of illocutionary act information that can be determined by

features such as the meaning of an expression or the

180

181

grammatical structure of that expression, and the possi­

bility of representing illocutionary act information within

a grammatical description. More specifically, in a sentence

which could normally be used to perform some illocutionary

act what information about that sentence which is connected

with it being used to perform that act would it be reasonable

to expect to find in a grammar? Further, what information

is best viewed as belonging somewhere else, and where does

it belong?

An obvious place to begin looking for illocutionary

act information would be in the meaning of the illocu­

tionary verb which names a given illocutionary act. What

is needed is some way of determining what information

belongs in the specification of the meaning of the verb.

The procedure to be used here to elicit this information

will be to compare performative and non-performative uses,

such as third person or past tense uses, of these verbs.

Sentences in which the verbs are not used in the first

person indicative active are not used in the performance of

that illocutionary act, and features that show up in these

uses should, arguably, be reflected somewhere in the

grammar. Some of the features that are features determined

by the meaning of the verb will be represented in the

reading of the verb. To claim that this information is

part of the meaning of the term in question or is determined

by the meaning of the term is the simplest explanation of

182

the fact that it occurs in both explicit performative and

non-performative sentences.

To sum up the procedure, it would work as follows.

In uttering "Last week I promised to be at your house," the

speaker is not promising. But there are certain syntactic

and semantic restrictions which govern the acceptableness

of the sentence. These same restrictions show up in the

performative sentences with this verb. Given that these

restrictions show up in both places, and cannot be accounted

for by other, for example, pragmatic considerations, it is

reasonable to suppose that information that there are such

requirements belongs in an adequate grammar of the language.

This would be the simplest explanation of the restrictions.

I will limit discussion here to sentences which have a

performative main verb and a complete embedded sentence.

In this way we afford the largest amount of data for the

grammar to account for, and insure that data are not read

into sentences as they might be if the sentences signifi-r

c^ntly underdetermined the act.

Searle (.1969) lists as one of the rules for the use

of the force indicating device of promising that the act the

speaker commits himself to doing be a future act, This

seems to be information that should be featured in the

reading of the verb although it should be modified to

reflect the fact that the act only need to be future with

respect to the time of the promise, and not, for example, to

183

the time of the reporting of the promise. Unless the future

requirement holds, sentences like, "Just this morning I

promised I would have the job finished by last week," would

not be anomalous. So, the dictionary entry for "promise"

will contain information to the effect that what S promises

to do must be a future act.

What is needed is some way to formally represent

such information. Katz (197 2) has offered one method for

representing semantic information. Very briefly, Katz

argued that the meaning of any expression can be understood

in terms of the concepts that make it up. Suppose that the

task is to represent the meaning of the phrase "dog runs,"

in this notation. The meaning of "dog" is viewed as made up

of certain concepts like, being an animal, being four<-legged,

and so on. Exactly which concepts are necessary to include

in the meaning will depend on theoretical considerations.

If including a concept gets certain other semantic proper­

ties and relations right, then it is to be included in the

meaning. That is, including being animal in the meaning of

"dog" will explain the analyticity of statements like,

"Dogs are animals." Each concept is represented in his

theory by semantic markers, They will be set off in

individual parentheses, for example, "(animal)/1 Since in

the expression 'Jdog runs" "runs" is being predicated of

"dog" the semantic representation should reflect this fact.

Except for parentheses that enclose words, parentheses

184

reflect modification. "(A) ((B) ((C)))" would indicate that

" (C)" is predicated of "(B)," and "(B) ((C))" of "(A)."

Next, "run" is such, on this meaning, that something

does the running, and what does it is animate. The reading

for "run" will contain a slot with a variable to indicate

that something must be running. That what does the running

is animate is reflected by placing a selection restriction

on the value of the variable. The reading that replaces

the variable must be marked "animate." I will depart from

Katz, and indicate this by subscripting the variable.

Supposing that the concepts that make up the meaning of

"dog" are being an animal and being four-legged, we get the

following intermediate reading,

( (animal) ((four-legged))) ((runs))

"Runs" would be broken down into its constituent concepts.

This is a radical oversimplification and barbarism of some

very sophisticated means of representation, but it should

suffice for present purposes.

A tree diagram could be constructed from any of the

readings should it be necessary to facilitate understanding

of the reading. If an item is enclosed in parentheses

indicating that it modifies the item to its left, as

described above, branches can be drawn from a node to the

item and the one that it modifies. The item on the right

branch of any node would modify the item on the left branch

185

of that node. Should these items themselves modify or be

modified by some other part of the reading a branch would be

drawn from the node above the two items either to the right

or left to another node to form a new branching with these

other parts of the reading.

A number of questions emerge at this point about the

interpretation of the formalism. There will be differences

that my general view might impose on any interpretation from

those that an alternative account might impose on the

formalism. I will postpone discussion of these differences

until I have discussed the grammatical representation of

this illocutionary act information. Then I will return to

the question as to exactly what is the grammar representing.

Borrowing loosely from Katz a means of representing

expression meaning, we can begin to construct readings.

Promising requires both that someone do the promising, and

that something is promised. What is promised is the

performance of some future act, I will represent the future

time reference restriction by subscripting "f" to the act

variable. Then, as a start we get something like the

following.

(S) ((promises) ((Pf)))

What replaces "P" will be some reading for the embedded

sentence. In the reading of "S promised that S would shoot

186

someone," what will occupy the position of the sentence

variable will be the reading for "S will shoot someone."

Part of what remains to be done is to spell out

what belongs in the reading of "promise" in the manner

described above. There are at least two difficulties in

giving an analysis in a Katz-like framework consistent with

the analysis of the act given in earlier chapters. One is

to develop a reading for the weakened intended effect. The

other involves determining how to include the information

from the reasons clause. This information should be

included in the readings since this information is such

that it too can be determined by examining third person

uses of the performative verb. I will begin with the latter

of the two problems, the proper treatment of the reasons

clause. If S promised to ip, then he meant that he would ip

because (A was to consider that S had as a reason) that S

had undertaken an obligation to i|), I will not attempt to

analyze the notion of undertaking an obligation, I will use

"because" to represent the rather complex conditional nature

of the reasons clause. The purpose is only facility of

representation. The other difficulty I will handle by

appealing to the weakened notion of "tell" indicated in the

third chapter as the sense of meaning that someone is to

entertain a thought. This is another simplifying procedure,

Promising to do such and such is something that S

does so the reading would reflect this fact by enclosing

187

all the information to the right of "S" in parentheses.

(S) C )

The first items inside these parentheses would be the

representation of telling A something, so everything after

"tells" or after "tells A" would be enclosed in parentheses.

((tells) ( ) )

What is told, of course, is that S will for such and such

a reason. I will keep the variable "P^" for S ip-ing. The

reason, "because such and such," will appear to the right of

the right of "P^" indicating it is why S ip1 s,

CPf) ( )

S tells that he will tp, and will have a certain reason for

iJj-<-ing just because he told them he would ip, The parentheses

which have not been filled out yet would represent informa­

tion to the effect that S intended someone to believe S

would ijj because S had undertaken an obligation, Placement

of the parentheses would be determined as it was for the

other parts of the reading. Using "because" as explained

above, and treating S's act of telling as a single concept

for the sake of simplification we would get the following

for this part of the reading.

(because) ((S's telling) ((obligated),..,,))

What remains to be done is to specify the proper relation^

ship between the obligation and S's ip-ing. I will assume

188

that it is just the case that S's i^-ing is what has become

obligated. I will treat it as a single concept.

(obligated) ((S's ijj-ing) . . .)

No information remains which belongs to the right of

"S/s ^-ing." The dots indicate only that a number of

parentheses from the other parts of the reading which were

used to indicate direction of predication would be included

in the reading.

I am not so much interested in the specific correct­

ness of this reading as I am in the possibility of repre­

senting that information. The readings should be indicated

with that proviso in mind. What the actual readings would

look like is an empirical question determined by what

information including a particular item would allow us to

predict.

The process of determining how much information

should go into a particular reading of the verb can be

determined not just by comparing performative and non-

performative sentences, but by comparing sentences involving

a particular illocutionary act type with sentences involving

another. That is, sentences with different performative

verbs, but which are otherwise alike are compared. The

features which it is necessary to include to account for

differences of other semantic properties, and which are

invariant between performative and non-performative uses

189

of the verbs it seems are justifiably claimed to be such

that a description of a language should account for and

include them.

To see how this might go consider some examples

mentioned earlier in giving analyses of various illocu-

tionary acts. Part of the difference between guessing and

predicting is that predicting involves evidence. This holds

across various uses of the two verbs, and would account for

the fact that "Jones estimated that the mountains were not

more than forty miles away, but being from the city and

not having anything to go on, Smith only guessed at their

distance," is not unintelligible as it would be if the

positions of "guess" and "estimate" were reversed. It

should be a fact that is reflected somewhere in a grammar,

and as the only difference is in the location of the verbs,

the most likely place to reflect the information would be

in the reading of the performative verb. If the earlier

analysis was correct, and this information belongs in the

reasons clause, then it would be handled in the same way

that it was with the reading for "promise."

In those cases in which the P is not what indi*-

viduates the act, the reading for P will just be the reading

of the embedded sentence. In (F)P-identifiable cases what

goes into this part of the reading will be the reading of

the embedded sentence modified in the way that the proper

identification of the act demands.

190

Suppose that S promises to attend church by

uttering, "I promise I will go to church." Assuming that

"I" here means "the speaker of the sentence" the grammar

would assign a reading to the sentence uttered which would

be characterized roughly by saying that the sentence meant

the speaker tells someone the speaker will go to church

because in telling that he would the speaker obligated

himself to go to church. Supposing that the sentence means

roughly this, then what is needed to insure that in uttering

this the speaker was promising is that he uttered it

literally, that is, that he meant what the sentence meant.

For a more detailed explication of this see Chapter III.

A significant part, then, of the information that

is placed, for example, by Searle (1969) in the rules for

the use of the force indicating device of promising is

information which would be considered information about or

information determined by the meaning of certain illocu-

tionary verbs. Specifying the limits of what can be

represented in this way is the job of a later section. In

it I will discuss information that seems not to belong

within the scope of a grammar.

First, however, I would like to show that the

information that appears in sentences with a performative

main verb that can be motivated on purely linguistic grounds

is not just information that applies to performative

sentences. I do this to show the systematic character of

191

much of this information. Some of this information re­

appears in the form of restrictions that can be imposed on

any sentence which could in normal circumstances be used to

perform that act. By normal circumstances I mean the

speaker meaning at least what his words mean. The pro­

cedure for eliciting this information will be to look at

restrictions which can be made on the embedded sentence in

an explicit performative sentence. The previous idealiza­

tions are still operative; the discussion is limited to

those cases where the embedded sentence is a complete

sentence.

The restrictions on the embedded sentence are, it

appears, the same restrictions that can be placed on

complete sentences other than performative sentences which

are used to perform the illocutionary act named by the verb

from which the restrictions were drawn. If this relation­

ship holds for the bulk of non-conventional illocutionary

acts, then we will have a method of limiting the range of

possible illocutionary acts which a sentence can normally

be used to perform. That is, there will be a way of

determining what range of illocutionary acts a speaker

might be performing, barring special circumstances, when he

means (at least) what his words mean.

In short, the proposal is this. The literal

uttering of a particular non-performative sentence can be

associated with any one of a number of illocutionary acts.

192

However, there are limits. A given sentence cannot in

ordinary circumstances be used to perform just any illocu-

tionary act. The hypothesis is that there are certain

linguistically determined restrictions on the embedded

sentence in a performative sentence which are a conse*-

quence of which performative verb is the main verb.

Performative sentences are normally associated with only

one act type. Any non-performative sentence which can

normally be used to perform that illocutionary act is

subject to these same restrictions when it is being used

to perform that act. Hence, "I will be there," can be a

promise in normal circumstances, but, "Your feet stink,"

cannot be, and just looking at the embedded sentence in an

explicit performative sentence with "promise" as the main

verb would allow prediction of this.

The regularities picked out here apply only to those

acts which are not highly conventionalized ones, In these

cases rather specific formulae arise for the performance

of the act. Even in these cases certain features can be

predicted, but I will concentrate on the non-conventional

ones. There doubtless are more regularities than X will

deal with. The ones I will concern myself with are the

subject of the embedded sentence, the tense (time reference),

and the presence of an evaluative marker in the performative

verb.

193

Are there any restrictions on the subject of the

embedded sentence in sentences where "promise" is the main

verb? It might seem that the subject of the embedded

sentence must be in the first person, but this is not clear,

It is true that we do make promises by saying things like,

"I promise that he'll be there." Arguably such sentences

are elliptical, and what is really meant by the speaker is

that he promises he will see to it that someone is there.

Doubtless this is what we mean, or should mean if we are

thoughtful, but we do promise by issuing sentences such as

the above. Which is correct does not matter. If ''I promise

he'll be there," is accepted as a promise, then so would,

"He'll be there," be accepted. The parallel will hold. The

hypothesis is that such restrictions will hold in the case

of other sentences used to perform the act if they hold in

the embedded sentence.

There is a restriction on the time reference of the

embedded sentence. It must make future time reference.

Otherwise the sentence fails to be intelligible.

This second feature appears to be shared with

"predict" when it occurs in main-verb position in performa«"

tive sentences, although as noted earlier there is not

always the requirement of a future time reference, These

are instances where the time reference is future in an

epistemological sense, but if there are such cases they will

not affect the generalization. That is, if predictions

194

about a past event in an epistemological future are

allowable they will be allowable in both the performative

and non-performative sentences. But, utterances like,

"She wasn't there," which are claimed to be predictions may

be so only in an extended sense, and the performative uses

such as, "I predict she wasn't there," may be somewhat

strained. It may be that an explanation is required before

the sentence is deemed acceptable. Whichever way is taken,

the generalization holds, I will assume that there is a

future time reference requirement on both sentences,

"Answer" does not impose either of these restric­

tions on the embedded sentence.

Explicit performative sentences with "order" as the

main verb require future time reference. "Order" seems to

allow not only second person, but also third person subjects

for the embedded sentence. The third person is, I think,

the second person in ceremonial garb. There are orders

like, "I order Bill Jones to step forward," or "X order

the person who insulted the commanding officer's wife to

turn himself in." The first of these might be used in a

large crowd where the reference of a second person subject

would not be understood. The other might be used in

circumstances where the offender was not known, although

such sentences are used when the identity of the addressee

is known. That such uses are derivative off the second

person is suggested by the fact that it is permissible to

195

say, "I order you, Bill Jones, to turn yourself in." On

the other hand, "I order he (him), Bill Jones, to turn

himself in," is not acceptable.

Again, though, whatever the subject restriction on

the embedded sentence it reappears in other sentences used

to perform that act. Hence, "You will report to the mess

hall in ten minutes!" and "Bill Jones, turn yourself in!"

are both acceptable as orders.

Further evidence that the restrictions are

linguistically based in that the same restrictions appear

in sentences with a performative main verb which is in the

past tense. The restrictions govern the embedded sentences

of these sentences also. This is seen by noting the

unacceptability of sentences like, "I predicted the sun

disappeared," or "I promised I was there," rather than

"I predicted it would disappear," or "I promised to be at

the party,"

There is some information which appears to meet the

hypothesis, That is, certain requirements on the embedded

sentence are also requirements on other sentences used to

perform that illocutionary act. However, it is not clear

in these cases exactly what the status of the information in

the restriction is, One difference between performative

sentences which in their normal use are assertive acts and

those which are imperatival acts is that in the latter the

embedded sentence must, to speak roughly, describe an

196

action. Admittedly, there is something odd about "I order

the sun to set," and the oddity is preserved in the

imperative sentences used to perform that act. To have a

totally non-defective sentence, the embedded sentence must

be such that it describes an action. However, it is not

clear exactly what the restriction is, and whether it is

linguistically determined.

Notice that such sentences as "I order the grass is

green," will be blocked by the requirement of a future time

reference. Hence, there is another explanation of what is

wrong with it other than the fact that the grass being green

is not an action. The future time reference restriction

will not block, "I order the sun to set," and hence future

time reference cannot account for anything that is wrong

with it or with similar sentences.

Two factors disincline me toward placing a restric­

tion on the embedded sentence that it be an action, and

treating this as analogous to other restrictions. First,

it is not clear that the oddity of such sentences is

linguistically based or whether it is merely a feature of

our knowledge of the world. We know as a factual matter

that the sun does not have it in its power whether or not

it sets, In an earlier age where beliefs about the agency

of the sun or the grass might have been different such an

order may not have been odd. It is not clear th^t such a

restriction is a feature of the meaning of the main verb.

197

This raises a general question about how to distinguish

knowledge of the language from knowledge of the world, It

is not an easy one, but one which a semantic theory must

answer.

Secondly, even if there is reason to believe that

the restriction is justified on linguistic grounds there

might be some other way of getting at it. It might be

accomplished by placing a restriction on the subject of the

embedded sentence, for example, requiring that the subject

be marked as capable of performing actions such as "animate,"

This explains the difference is acceptability between

ordering the grass to remain green and ordering someone to

see to it that the grass remains green. It may be felt

that a future time reference and an animate-marked subject

is not sufficient. However, this is not clear. Even

something like, "I order you to be tired tonight," is not

clearly anomalous. To carry out the order the addressee

might spend the afternoon running around the block,

We may have just substituted one theoretical

question for another, The requirement that the subject be

marked "animate" might not be a linguistic, but only a

factual requirement. I will avoid this issue, and restrict

myself to the requirements I mentioned at the outset.

The generalizations seem to be holding up well so

far as a comparison of the following pairs should indicate:

"I'll be there," "I promise I'll be there"? "I was there";

198

*"I promise I was there"; "Leave town!" or "You will leave

town!" "I order you to leave town," If there is something

wrong with the embedded sentence on the left-hand side, then

any sentence which also has those characteristics responsi­

ble for the oddity will not be one with which a speaker

could perform the act in question. Notice that we can order

by using an imperative sentence that might be said to have

no subject at all. But the arguments that the underlying

representation of that sentence has both a second person

subject and a modal tense marker seem quite strong, Hence,

these are things that the grammar would say about the

sentences in question.

It might be claimed that the generalization runs

aground of considerations based on sentences like, "You

did it," and "You did it quite well," The problem they

present is that although either of these might be used to

inform someone or to state something, only the latter and

not the former can be used to congratulate in normal

circumstances, Likewise to reprimand something like, "You

did a bad job," and not just "You did a job," is needed,

Those cases where just saying "You did it," is sufficient

to congratulate involve special intonation or special

circumstances. But if all this is true, then a sentence

meeting the requirements imposed by a performative verb is

not sufficient to guarantee that it can be used to perform

the illocutionary act named by that verb.

199

The illocutionary acts of which this are true are

those like reprimanding, congratulating, and praising.

These are all, in part, acts of evaluating, and any account

of their meaning should reflect this fact. Arguments like

the ones used previously could be used to establish this

fact. Further, not only are all of these marked to indicate

they are evaluative, that marking is either pro or con. To

praise someone is not to express disfavor. This should be

reflected somewhere in an account of the meaning of the

evaluative illocutionary verbs. Hence, the explicit per­

formative sentence's representation in the grammar will have

an evaluative marker somewhere in that representation,

And, it is just this evaluation that is missing from

a sentence like, "You did it," However, sometimes intona­

tion can carry that evaluation, I suggest the following

generalization. If a performative verb has a pro or con

evaluative marker any sentence which is used to perform that

act would have that same marker, that is, be marked as

either a pro or a con evaluation, "Reprimand" would have a

negative marker in it as would, "You did a lousy job of

covering up the scandal, John," Therefore, this sentence

could be a reprimand in normal circumstances if it met the

other requirements which would possibly be that it must be

past or present tense, and have a second person subject

term, Sometimes the intonation or the context will make it

clear that the speaker is evaluating an action one way or

200

the other. In normal cases though the non-performative

sentence must be marked in the same way that the performa­

tive verb is,

Interestingly it is only sentences in which some­

thing receives an evaluation from the speaker's point of

view, that is, where what the speaker primarily is doing is

evaluating for which the point holds.

"Reprimand" may not be a suitable case in this

regard, It does not meet the requirements of an explicit

performative sentence, namely, the first person, present

tense, indicative, active use. The more ceremonial, "You

are hereby reprimanded," is much more common and natural

than "I hereby reprimand you," or "I reprimand you." It may

be best not to include it on any list.

Interestingly it is only sentences in which some­

thing receives an evaluation from the speaker's point of

view, that is, where what the speaker primarily is doing is

evaluating for which the point holds. Speakers warn

hearers about things which are purported to be not in the

interest of the hearer, but in warning the speaker is not

evaluating, and the requirement on other sentences used to

perform the act do not hold.

There are probably other requirements like this

last one which would emerge if a careful study were made of

utterances used to perform illocutionary acts. But, I

suspect the number of additional restrictions will be small,

201

and correspondingly the number of illocutionary acts that a

particular non-performative sentence can be used to perform

can be quite large. This is just the way the facts are. In

Figure 2 I have charted the regularities that seem to hold

between explicit performative sentences and other sentences

which in normal circumstances can be used to perform the

act that the performative sentence is normally used to

perform,

The chart works as follows. Suppose we have a

sentence with a second person subject and future time

reference, such as "You will shut the door," Special

circumstances aside, then according to the chart which is

constructed just off information from the readings of

performative sentences, the uttering of that sentence

cannot be a reprimand or a congratulating, but it might be

a promise, prediction, order or answer among other things,

This seems correct.

If these regularities hold across the more non-

conventional acts, then a means is available for deter­

mining the range of illocutionary acts which a given

sentence might be associated with. The means is to check

the restrictions which a grammar would put on performative

sentences.

In a work alluded to earlier Katz (1972) has

attempted to say something about what illocutionary act

information might be expressed in a semantic theory, and how

202

Person of Subject (Time

Tense Reference)

Evalua­tion

Act 1 2 3 Past Pres Future Pos Neg

Promise X ? X X

Predict X X X X

State X X X X X X

Request X ? X

Order X •p X

Claim X X X X X X

Advise X •p X

Congratulate X X X X

Inform X X X X X X

Answer X X X X X X

Figure 2, Illocutionary Act Sentence Restrictions

203

much might go elsewhere, for example, in a theory of con­

versation. The methodology used for constructing readings

used above as well as the form of the readings are borrowed

from his work.

Since my methodology has this debt, and since he

has had something to say on the subject of speech act

information and grammar it should be beneficial to examine

his views on the subject and compare them with what has been

said here,

Katz (197 2, p, 151) begins by examining three

sentences. "Someone stole the British Crown Jewels."

"Who stole the British Crown Jewels?" "Someone, steal the

British Crown Jewels!" He says they share the same pre­

supposition, namely, there exists someone and there exists

something which is the British Crown Jewels and is unique

in being such. Further, they share the relation, "x steals

y at time t." This he labels the "condition," The pre­

supposition and condition together make up the propositional

content. Where they differ is in the propositional type.

The three types above are, respectively, assertive,

erotetic, and requestive.

Nothing in the analysis presented in the first

chapter corresponds exactly to Katz's distinction between

propositional type and propositional content, It is not

just the F of the (F)P and (F)r in my analysis to which

propositional type corresponds as the F picks out much more,

204

individuating particular illocutionary acts within Katz's

propositional type, What it appears to correspond to is

whether the act is assertive or imperative. In the

analysis I have presented I have assumed that questions

belong in the imperatival class. Since there is a differ*-

ence on my account between assertive and imperatival cases

it is important to see if there is any formal way of

representing the differences. I will begin by going through

Katz's account to see how compatible his characterization

of the differences is with the distinctions which I have

drawn. In addition I will note his comments on the illocu­

tionary act information which a grammar should record,

Katz's view is that sentence type determines

illocutionary force to the extent that this is determined

by grammatical structure. However, since sentence type

does not guarantee a particular propositional type Katz

(1972, p, 152) says, "Thus we have to say that the meaning

of the main verb determines propositional type when this is

not determined by the sentence type." He suggests that

grammatically the class of performative verbs can be

distinguished from other verbs by the fact that they play

this role in determining propositional type. Their

determining propositional type is, of course, contingent on

certain other conditions being met. The tense of the

auxiliary must be present, and the subject must be first

person. The propositional types "Q," "Imp," and so on will

205

be represented in the reading of the performative verb.

Katz views the mechanism of the conversion of sentence type

to be the selection restrictions on the variables cate­

gorized for the subject-main verb relation and for the

tense-main verb relation. The selection restriction will

require that the readings have markers, respectively, for

speaker of the sentence and for present tense. If either

selection restriction is not met the type is assertive. If

both are satisfied it takes the type specified in the

reading of the performative verb.

Katz motivates this account, as is his usual wont,

by its ability to explain further data, in this case the

difference between performatives and causatives. Whether

this is the most efficient way of handling this information

is a question for semantic theory, and beyond the scope of

this discussion.

The propositional type operates on the condition of

the sentence, The propositional type converts the uncon­

verted condition into either a truth condition (assertive),

answerhood condition (erotetic), or a compliance condition

(requestive) ,

There are two areas of contrast between the views

of Katz and myself on these matters. First, he stops the

contribution of the grammar at this point. Other informa­

tion, such as the fact that a certain non-performative

sentence can be used to perform a certain illocutionary act

206

belongs in some other theory. I have already indicated what

further information about sentences and illocutionary acts

can be predicted just from information a grammar would

provide.

The other contrast is with his general view of

propositional type. In distinguishing assertive and

imperatival illocutionary acts I characterized the differ*-

ence as whether S mean that P or S meant that A was to ,

It is not sufficient to say that the difference is that in

the latter cases what is meant is that someone is to do

something, A formal characterization of this fact is

needed, Just the characterization above is not correct

because that someone is to do something can also be the

content of an evaluative act. S can report that Jones is

to go to the office. Barring some other formal charac~

terization of the difference Katz's method will win by

default. What he has said on the matter is not incompatible

with the theory outlined in the first chapter, but a good

amount of work needs to be done to show that it is not,

Rather than doing this work I would like to offer two

proposals that might suffice for distinguishing the

imperative from assertive acts, roughly, for distinguishing

different propositional types which would be revealed in a

semantic description of the performative sentence, I think

that there will be such a way because it seems that the

propositional type is in every case determinable from

207

knowing completely what was meant and what reasons A was

to have.

Suppose we had a semantic representation of the

following two sentences, "I state that you will shut the

door." "I order you to shut the door." Suppose, although

this has not been shown, that at some level of representa­

tion the content of the order and the content of the stating

were indistinguishable. Is there any possible way the

requestive nature of the order might be revealed in the

reading? It was argued earlier that the information from

the reasons clause would appear in the reading of the verb.

One hope, but only a hope, is that the character of the

reasons for someone doing something is not the character of

truth-supporting reasons (Schiffer, 1972, pp, 57-58), and

so would differ from the reasons for believing something.

The different character of the reason, then, would be

featured in the reading of the imperatival performative

verb. At this point such a proposal is largely an article

of faith, but there is some support of it, namely, that some

h^ve argued that reasons for believing are different than

reasons for doing. It was argued for each illocutionary act

the reason was either a reason to believe something or a

reason to do something. This is the fact we would seek to

reflect in the reasons clause of the readings. In this way,

the fact that the intended effect involves an action in

some assertive cases will not eliminate the distinction

208

between act types. For example, that S has evidence would

be a reason for believing something, but not of itself a

reason for doing anything. Whether such distinctions can be

maintained would require a detailed examination of reasons

for believing and doing, and the way in which the reasons

are reasons.

Let me return to the intended effect in the asser­

tive and imperatival acts, and see if there is something

there which will separate the cases. The difficulty is that

there are sentences that meaning approximately, "I tell you

that you are to see the principle." But someone who meant

this might mean either he was telling someone that this was

the case, an assertive act, or telling him that he was to

do this, an imperatival act. It seems that "are to" is

ambiguous in these two cases. At the level of semantic

representation this difference would be represented, This

difference in representation would allow distinguishing

propositional types. They can be distinguished by whether

someone meant that something was the case or that someone

was to do something on a reading that would not be

equivalent to meaning that something was the case.

There is, then, a great deal of information about

illocutionary acts that is revealed or would be revealed by

an adequate grammar of the language. Such information will

not insure on any given occasion that a speaker performed

a given illocutionary act. To know this is to know what he

209

intended, and such knowledge is not to be gleaned from a

grammatical description of the expressions he utters, but

from the circumstances in which the utterance of the

expression took place. What a description of a language

will tell us is what illocutionary act a speaker performed

if he meant what his words meant in the case of explicit

performatives, and in other cases what is the range of

illocutionary acts he might be performing if he meant at

least what his words meant.

The question emerges as to how a speaker can mean

more than his words mean, and how an audience can grasp

what he meant. How do all of these acts which a grammar

can not determine take place? How is the gap between what

is meant and what is said bridged? To request an answer to

these questions is to request a whole theory of communica­

tion, I will not offer such a theory. Rather, X will

discuss one small corner of such a theory, the cases of

indirect speech acts. Before turning to these I will make

the promised detour into the issue of interpretation of

semantic theories,

Interpretation of Grammars and Speaker Meaning Theories

For Katz (1972, pp. 38-39) semantic markers repre­

sent concepts, and concepts are abstract entities, Since I

have appended a version of his formalism to an account which

makes speaker meaning basic, and which offers a specific

210

account of what it is for a speaker to mean something, there

might be differences required in the interpretation of the

formalism.

Sentence meaning on the theory I suggested in the

previous chapter is to be understood in terms of what it

would be expected that someone who uttered the sentence

meant. Leaving out the complexities of the reasons clause

here, speaker meaning is a matter of intending someone

entertain the thought that such and such is true. Sentence

meaning is to be understood, then, in terms of what thought

would be expected that someone who uttered the sentence

intended the audience entertain.

What the formalism such as that I gave for "I

promise I will go to church," represents is the structure

of the (F)P to be entertained. To this point I have just

schematized this as (F)P, but said nothing about what it is

a schematization of. To have a name for what is being

schematized, let me call it a proposition, Then, S meant

that P if S intended A entertain the thought that the

proposition P is true. The formalism represents the

internal structure of the proposition.

It is important that nothing I have said commits me

to a particular way of viewing propositions, On the one

hand, I am free to follow Katz in viewing them as abstract

entities, What I cannot do on my theory is identify there

being a proposition with there being a meaning. That is,

211

there could be propositions populating every corner of the

universe in my view, but no meaning. Meaning would arise

when a speaker intended in a certain way to get someone to

consider whether the world was the way that a certain

proposition represented it as being. Sentence meaning

arises when people have certain expectations about these

intentions of speakers.

The following situation is somewhat analogous.

Speakers understand, communicate, and state laws of logic.

This does not force anyone to adopt a particular view on

what laws of logic are. Nor does it mean that there must

be understanding of, communication about, or stating of

them in order for them to exist. Just the fact we do these

things with them says nothing about their status. Likewise,

that speakers intend in a certain way for certain people

to consider whether a certain proposition is true does not

require viewing propositions in a certain way, What I have

said about meaning is that it depends on speakers, It

does not follow that everything that is necessary for them

to do this must be understood in terms of the speaker's

mental life. Certainly some of what is involved in meaning

must be interpreted in this way. Meaning involves in one

way or another intentions, expectations and beliefs.

There is nothing in a speaker meaning theory which

is incompatible with a formal semantics, Even if speaker

meaning is defined as the intention to create beliefs it is

212

open to give those beliefs the requisite internal structure,

and to say what is necessary to say about what beliefs are.

What could not be done would be to identify there being

beliefs with their being meaning. On the view I have

outlined what cannot be done is to identify the proposition

and meaning. The systematic nature of meaning phenomena

which linguistic and logical semantical theories charac^-

terize is just one aspect of what a complete account of

meaning involves.

Although I suspect my account is compatible with the

abstract entities many semantic theories find necessary, I

am even more suspicious that there are no abstract entities,

I do not know what will do the work that they are said to do

in accounting for the phenomena, The accounts that are

usually contrasted with the abstract entity accounts are

often labeled mentalist accounts, As I use the term it

here refers only to a view which maintains that the

formalism of a theory can be viewed as representing some­

thing essentially mental, either mental events or mental

states or whatever the proper description is, I do not have

a complete mentalist account, but I do not have it obviously

false that a mentalist account will never do the work of an

abstract entity account, With the understanding that

nothing in adopting a formal representation of some meaning

phenomena commits me to either view of the nature of

propositions, or to any other view for that matter, I would

213

like to offer some suggestions as to what a mentalist, as

I have defined "mentalist" might say about propositions,

Mentalism in semantics is supposed to run aground

of the facts. It cannot get them all right. If, to speak

incorrectly, but understandably, a proposition is what a

speaker means there must be an infinite number of proposi-^-

tions since there are an infinite number of things that a

speaker might mean. If propositions are abstract entities

we can be assured of always having enough of them so we can

mean whatever it is that we wish to mean.

The large number of things that might be meant

cannot be the real difficulty with mentalism as it should

be obvious there is no difficulty here. That it is not the

real difficulty will emerge shortly. Consider this case,

There are, certainly, an infinite number of things that a

person might believe, beliefs he might have. He will only

have some subset of these beliefs, but he could have any of

an infinite number. Beliefs are mental events whatever it

is that mental events turn out to be-<-brain processes, bits

of behavior, dispositions to behave, or dramas on an

ethereal stage. But there is no difficulty in supposing

that person might have any of an infinite number of beliefs.

Just that there are an infinite number of propositions

someone might consider the truth of does not show there is

any reason to suppose that propositions cannot be mental

214

events. The problem of mentalism in semantics is not that

there are a lot of things to mean.

The real problem is that the mentalist must, get his

infinite number of propositions from a finite store in a

systematic way. What a Katz-like formalism does is show

how a finite number of concepts can be combined in syste­

matic ways to get the infinite number of senses, and allow

the proper predictions about semantic properties,

This raises three questions that a mentalist account

must attempt to handle. If propositions are mental events/

states of some sort, then we can ask: (1) What sort of a

mental event/state is a proposition? (2) What is the status

of the items that the formalism represents as their

constituent parts? (3) How do we get from the finitude of

one to the infinity of the other?

Certainly I can do no more than just outline the

kind of answer that a mentalist might give at this point,

Nor will I bother to defend the answers against the

criticisms that might be raised, I will not concern myself^

either, with any of the modifications what might be neces­

sary for imperative cases.

Propositions might be viewed as (mental) repre­

sentations of possible states of affairs, That there are

such representations is not sufficient for their being

meaning, As was said, meaning will emerge at that point

215

where speakers intend hearers to consider whether the

representation corresponds to an actual state of affairs,

I do not intend the mental representation to be

understood as a mental image, Fodor, Bever, and Garrett

(1974, pp„ 156~16Q) summarize the devastating objections to

an imagist view. There is no need to view representations

in this imagist way. There are a large number of accounts

available on how to view representation. Rosenberg (1974)

discusses a number of such accounts in the process of

forming an account with empiricist underpinnings.

But to insure that propositions get the facts right

it appears necessary to go inside the proposition, If a

mentalist adopted a Katz-like formalism for presenting

propositional structure, what might he say about whatever

it is that semantic markers stand for?

Clearly on this view they represent the repeatable

semantic content of the propositions. This is not to say

what they are, No harm arises from following the terminology

of Katz, and saying that semantic markers represent concepts.

So, what are concepts on the mentalist story? What might

be said at this point is that to have a concept of something

is to have the ability to identify particular instances of

that thing, either something concrete like a dog or a

property or relation such as being red or something kicking

something, To have a concept is to have a certain

knowledge. To have the concept of a dog is to be able to

216

distinguish dogs from other sorts of things, and to identify

all of those things which are dogs. It is important to

distinguish having the concept from just correctly picking

out dogs. Someone may have gone blind, and lost his sense

of touch, but would still be said to have the concept of a

dog. Having the concept is whatever it is in virtue of

which someone can identify the individual members. This

account would need to be applied to concepts like running.

The remaining question is how do these individual

concepts get composed into the infinite number of proposi­

tions which a person might wish another to consider to see

if it represents an actual state of affairs, If rules can

describe the compilation of semantic markers into the

representations of the senses of sentences, why not suppose

that rules can describe the compilation of concepts into

propositions? Then, one need only assert that speakers,

in some sense of "know," know these rules, I do not wish

to enter the discussion as to whether there can be knowledge

of this, probably, unconscious sort, The point is that this

is what the mentalist can say, and Fodor (1968) at least

shifts the burden of proof to those who wish to criticize

this way of talking,

There is a further question as to the meaning of

"composition" here, The composition of words into sentences

is understandable^ but at some point an answer needs to be

provided as to what the combining that could be a combining

217

of mental (or ultimately physical) items would look like or

what mental processes would be functionally equivalent. It

is not obvious that the current theoretical vocabulary of

any psychology will suffice.

Possibly there is a mentalist interpretation of the

features of an account of meaning which seem to defy

mentalist explanations, I do not propose that any of this

is correct, and as the correctness of nothing X have said

previously depends on it I will not examine it further

beyond noting it does have the consequence that it suggests

a rather intimate connection between our representing the

world and our speaking about it. What we can mean depends

on what we can represent about the world,

I will leave the discussion of mentalism and

abstract entities in semantics at this point, and return to

the subject of illocutionary acts, In many indirect speech

acts it appears that a speaker can mean more than he says,

How might he be able to do this?

Illocutionary Acts and the Use of Language

The issue I wish to discuss involves certain

indirect speech acts, illocutionary acts which are performed

by uttering something which in virtue of its form would

normally lead one to assume that it was being used to

perform a different illocutionary act than the one which it

is actually being used to perform, I will look at cases

218

where not only is a different illocutionary act type

involved, but where a different propositional type, in the

sense of propositional type used by Katz (1972), is

associated with that expression. An example would be a

sentence like, "Do you have the time?" Ostensibly the

utterance of this would be a question, and would require a

"yes" or "no" answer on the part of the hearer. However,

this is not normally a question about whether a hearer

either knows or does not know what time it is. It is a

request that the hearer tell the speaker what time it is,

I will list a number of possible indirect speech

acts here. Each group is arranged in an order determined

more or less by whether the expression is normally used to

perform an act different from what its form or meaning would

seem to demand, or whether some stage setting is required

for the new act type. More will be said of this later. The

first group consists of questions which might be used to

request that someone do something, the second of statements

which also share the feature of being associated with certain

requests.

The first group includes: "Do you have the time?"

"Will you shut the door?" "Will you go to the movie?"

"Would you pass the salt?" "Can you pass the salt?"

"Could you teach me to play chess?" "Can you fly the

plant?"

219

The second group includes: "I would like the salt,"

"Your cigar is annoying." "I cannot hear you." "You're

standing on my foot." "Your room is a mess," "You'd find

this boring," "This is grown-up talk, children,"

There are differences in the above cases, as was

said, as to how directly they are tied to different

propositional types. Part of the task here will be to

state and account for the differences.

How is it that a non-performative utterance comes

to be associated with a different propositional type than

its grammatical structure would lead one to expect? There

are a number of possible answers, and I wish to discuss

some of the more commonly given ones to see how adequate

they are as explications of what is going on in these

cases. I should state at the outset that it would be quite

surprising if one answer handled all of the cases, given

the apparent differences between the cases.

The first possibility is that the expressions in

question just mean what a full blown imperatival sentence

or what a performative sentence used to perform the act

would mean. In other words, the cases are cases of

expression meaning. Another possibility is that these

cases are just like certain normal kinds of cases of a

speaker meaning more than his words mean. Just what kinds

of cases will be spelled out later. In short, they do not

differ in kind from the case where a speaker says, "He'll

220

be there," meaning that Albert will be at the party. A

third possibility is that there is some kind of implicature

(Grice, 1964, 1967) which does the work of carrying the

additional information, or at least that a significant

resemblance holds between these cases and cases of

implicature.

Grice distinguishes two broad different categories

of implicatures, conventional and conversational, Many of

the difficulties which would confront the former of these

as an explanation of the data are the same difficulties

which would confront an account given in terms of expression

meaning, so I will restrict the discussion of implicatures

here to conversational ones. Then, is there any light to

be thrown on the data by an examination of implicatures?

If a speaker implicates that such and such is the

case, then according to Grice (19 64, 19 67) the speaker does

not say that such and such is the case, but hints of

implies or suggests that it is, In cases of conversational

implicatures he is able to convey the information he

implicates because the speaker and hearer are both aware of

general rules or maxims which govern conversation and, in

the case of particularized conversational implicatures,

certain specific aspects of the context in which the

utterance is made,

Grice (1967) says that these implicatures arise when

the participants are engaged in cooperative discourse,

221

Supposing that people are engaged in cooperative discourse,

then they will be subscribing to various maxims of conversa­

tion which govern discourse. These include maxims of

quantity enjoining the speaker against making a contribution

which is either too informative or not informative enough;

maxims of quality enjoining against contributions which the

speaker believes false or for which he lacks adequate

evidence; a maxim of relation which enjoins against

contributions to the exchange which are irrelevant,

irrelevance here being determined by the point of the talk

exchange. There are other sorts of maxims, but they need

not be mentioned here,

Implicatures arise when a speaker, whom the audience

supposes to be cooperating in the conversation, violates one

of the maxims governing conversation, A way this might

happen, one of Grice's (1967) own examples, goes as

follows, A teacher, S, in recommending a pupil says about

the pupil only that his command of English is good and that

he has attended tutorials, Working out the implicature

appears to come to bringing the contribution in line with

being a cooperative contribution. The contribution violates

the maxim requiring that the contribution be as informative

as necessary, S cannot, though, be implicating that he does

not know any more about the pupil since it is his pupil, If

he is cooperating he must be trying to convey information he

does not wish to put down. There is no reason to believe

222

he would not wish to put down good things about a pupil.

Hence, he must be implicating that the pupil is no good at

philosophy.

An additional requirement is imposed on anyone who

attempts to use something akin to a conversational implica-

ture to explain the performance of certain indirect speech

act. Since someone performed an illocutionary act if and

only if he meant something, and a complete specification

of what is meant will identify the illocutionary act

performed, care must be taken to insure that the explication

of how someone was able to perform that act allows for him

to have meant what is implied.

Among the cases which it seems more plausible to

treat this way is the following. It has an interesting

feature which will figure into later discussion. The

circumstances are special enough so that on subsequent

utterances A does not appear to have to work out, in the way

he initially worked out, the implicature, To provide an air

of realism, I can assert that it actually did occur.

S, an army drill sergeant, asks A, a basic trainee

who has not removed his hat on entering a building, "Is your

head cold?" A is initially puzzled, but comes to see that

S meant is that A is to remove his hat. If it is via a

conversational implicature that S meant this, then according

to Grice it should be capable of a detailed working out

along the lines of the one given in the recommendation case1,

223

There is in this case a real question about which maxim is

violated, It is either the maxim of relevance or a maxim

of quality, In the case of questions the two are closely

related. If S asks A out of the blue if A is sick, then

normally A will think that S has some reason to think that A

is sick, for example, that A does not look well. It is an

erotetic version of the maxim enjoining against saying what

you lack adequate evidence for. Otherwise, there is a

question as to why the remark was made.

Now, the following working out schema would be

sufficiently detailed to explain how the implicature could

be gotten across.

1. S asked A if A's head was cold,

2. The remark violates a maxim as there is no apparent

reason to suppose A's head is cold, This is a

feature of the context,

3. Hats are sometimes worn to keep the head warm (not

cold). This is general background knowledge.

4. If A's head were cold he might wear a hat. This

is from 3.

5. A is wearing a hat. This is a feature of the

context.

6. That S is wearing a hat may be the reason that S

asked if A's head were cold.

7. S cannot be asking if A's head is cold. This is

from 2, It is known that A's head is not cold, and

224

the remark violates the erotetic version of the

maxim of evidence.

8. It is impolite (wrong) to wear a hat indoors. This

is general background knowledge.

9. S has brought it to A's attention that A is doing

something wrong. This is from 6 and 8.

10. Drill sergeants tell basic trainees what to do when

the trainee does something wrong. This is back­

ground knowledge.

11. Hence, S must mean that A is to take off his hat.

This is from 9 and 10 and from the definition of

"telling" given earlier.

12. Drill sergeants do not request or suggest; they

order. This is general background knowledge that

S and A have.

13. Therefore S was ordering A to remove his hat.

The reasoning is in the open in that there is not

any inference which S intended A to make which he did not

intend A realize S intended A to make. So, S could have

meant that A was to remove his hat.

There are genuine questions about the real status of

the working out schema. It is not clear that A consciously

goes through anything like these steps and in a particular

order, so that at any given point he could tell the inference

which he just made. Whether A has gone through some such

225

deductive calculation or whether the inference is sub­

conscious or whether there is only some neurological

analogue of an inference or/and such statements are merely

rational reconstructions is a larger more general issue of

significant psychological importance, but one which I will

not go into. It seems reasonable to suppose that this case

can genuinely be described as one whereby implicature S has

performed some illocutionary act.

I will go through one more case, but X will not go

through the working out schema in the detail that I did

above. This example consists of a remake of a scene from

Giant, modified to make the reasoning less complex. S, a

male, is sitting with a group of men who are determining

what needs to be done to insure that they control the up­

coming election. A, voluptuous and wife of S, wanders away

from the women, and into the midst of the group of men. The

discussion comes to a halt, and S says to A, "This is just

man talk, dear." A does nothing, and S tries again with,

"You would not find it interesting." When she still does

not leave he finally tries, "It'll be boring." What S

meant in this case was that A was to leave, S requesting

that she leave, but in the sense of "say" which is tied to

the words uttered S did not say that she was to leave.

Here, then, is another candidate for a conversational

implicature. It would seem to be a particularized implica­

ture since it is rather closely tied to particular kinds of

226

circumstances. On the other hand there would seem to be a

suggestion of some sort with most claims about what a

person would find boring or uninteresting that S is

attempting to dissuade him from doing it. There is the

suggestion with claims that something is interesting that

S is attempting to persuade someone to do something. This

may be a consequence of a more general belief about pro­

viding person with reason either for or against doing

something.

I will suggest how this case might be understood

in a way similar to a conversational implicature. Each of

these steps is such that it would not be incorrect to say

that S intended A to realize that S intended A to realize

that S intended A to reason in this way. This will show

that it is possible that S might have meant, then, what A

is to work out. I will leave this prefix off of each of the

steps as I did in the previous case.

Providing someone with a reason to do something is

one way of attempting to get that person to do something.

This is general knowledge. The fact that staying and

listening to the conversation would bore A is a reason for

A not to stay and listen. Hence, S has provided A with a

reason not to stay, that is, to leave, There is a general

requirement of politeness to the effect that it is impolite

to tell someone to do something in order to get it done.

Accordingly, when being polite imperative sentences and

227

performative sentences are avoided, or "please" is affixed

or prefixed to the imperative. Hence, if S intended to get

A to do something and was being polite S would probably find

another way of doing it besides telling the person, that is,

besides using the imperative or performative forms used for

telling. Providing A with a reason is just such a way.

There is no reason to suppose S is not being polite. If S

were just pointing out to A that the discussion would not

interest A intending her to make up her own mind whether

she wished to or not, S would not have needed to make the

point repeatedly as there is no reason to suppose that A did

not understand what his words meant. Then he was doing

more, most likely, than just telling her it would be boring

or offering her a suggestion to consider whether or not she

wished to remain. Hence, S was requesting or telling A to

leave. More may need to be said to insure that the implica-

ture could be worked out, but I think it is reasonable to

think it would go along these lines.

The problem with a conversational approach to all

indirect speech acts is that no matter what we finally say

about the status of the working out of the implicature it

does not seem correct to say of some of the cases that they

are worked out or intended to be worked out by the hearer in

anything like the way it seems reasonable to say it about

the above cases. Suppose S asks A, "Do you know the time?"

The connection between the use of this expression and

228

requests that the hearer tell the speaker seems more

intimate, and less context dependent, than is the connection

between the expression and the request in those cases worked

out above.

There are other possible explanations for the cases

where the connection seems more intimate. The two mentioned

were the cases being cases of expression meaning or certain

particular cases of a speaker meaning more than his words

meant. The cases worked out above were cases of a speaker

meaning more than his words meant, but this is not the type

of case I wish to discuss. Some cases of a speaker meaning

more than his words meant do not appear to be cases where a

speaker needs to make an involved calculation to determine

what was meant. Such a case would be one where a conversa*-

tion the speaker uttered, "He'll be there," meaning that

Jones will be at the party. The meaning of the expression,

the timeless expression meaning (Grice, 1969) is not that

Jones will be at the party. However, the speaker did not

conversationally implicate that Jones would be at the party,

Determining what the speaker meant is not that involved.

Are the cases of indirect speech acts which resist analysis

in terms of implicature just these kinds of cases where a

specification of the meaning of the words underdetermines a

specification of what the speaker meant?

In the example used above what the speaker means

over and above what the expression timelessly means involves

229

providing a reference for the indexical items in the expres­

sion, Jones for "he" and the party for "there." Other cases

might involve in addition intended a particular meaning for

some ambiguous term. Intending a particular reference and

a particular meaning for any ambiguous term is a large part

of what is involved in the cases of speaker meaning like

the one above. The context usually makes it clear which

person, place, thing, or time is being picked out by the

indexical items. There is no need for appealing to a

working out schema to determine the reference or the

particular meaning of some item, and where A must do this

it is not intended. It is not clear though that intending

a meaning and reference is sufficient to mark off a class of

cases of a speaker meaning more than his words meaning

where A's understanding is not intended to be worked out or

calculated as above. This is because of the presence of

words like "do," S decides to take A's advice and tells A,

"All right, I'll do it," meaning that he will ask Mary Sue

Sweetwater to the prom. What S meant is not just a matter

of intending that a particular linguistic meaning be

operative. "Do" is not ambiguous between fixing a flat

tire, asking someone to a dance, or any of the countless

things that a person might mean. However, I think there are

reasons for treating the phrase "do it" as or analogous to

the way that indexical items are treated, "I did it,"

means roughly that the speaker performed some (contextually

230

specified) action. Determining what the speaker means over

and above what his words mean involves determining the

specific act he is talking about doing. The speaker meant

that he would do a specific act out of the countless ones he

might have been talking about, This is a more plausible

account of what is involved here than maintaining that he

said that he would do it, but implicated or implied that

he would take Mary Sue to the prom. He did not implicate

it. Like the other case discussed of a speaker meaning more

than his words mean S was making the reference of something

that can refer to more than one thing specific.

Supposing, then, that a certain type of case of

expression meaning underdetermining speaker meaning can be

marked off in this way, does it characterize any of the

cases of indirect speech acts which were listed at the

outset? Very clearly it does not. None of the new

propositional types depend either on ambiguity or on

specifying a reference of a particular item. Even though

in these cases a specification of what the speaker meant

would differ from a specification of the meaning of the

meaning of the expressions, this would not be due to the

fact that a particular meaning of some ambiguous term or a

particular reference of some term that can have more than

one reference. This does not appear to be what is at stake.

Then, has an expression like, "Can you pass the

salt?" just come to mean that the speaker is requesting the

231

salt. Among the reasons which incline one toward an

affirmative answer is the immediate, apparently non-

inferential, nature of A's understanding. Is the best

explanation of the data to be given in terms of expression

meaning?

I think there are stronger considerations against

making the new propositional type a part of the meaning of

the expression in question. First, there is the fact that

these cases lack the systematic character of many other

instances of expression meaning. They have to be learned

independently of learning the grammatical form and the

meaning of their constituent words. It can be argued that

on having mastered a grammatical form, say imperatival,

and the meaning of certain words and the contribution they

make to the meaning of the sentence in that form we can

substitute new words into that form to make new requests

or understand new requests that are made with that form.

Once we understand, "Give me the milk!," then we go on to

produce new requests by substituting other nouns for

"milk" or verbs for "give" or pronouns for "me," The new

sentences become requests to give the new item to someone

or to perform some other act such as selling the item to

someone. But if we suppose that "Do you know what time it

is?" in virtue of its meaning is a request we are confronted

with the fact that on substituting new lexical items into

the sentence we do not have new requests involving different

232

things. Nor do questions about people's abilities auto­

matically become requests for a person to do what the

speaker is asking he can do. "Can you run a four-minute

mile?" is not normally a request for a person to run a

four-minute mile even if the person is about to run a mile.

At best, if the new propositional types are matters of

expression meaning the cases are somewhat idiomatic.

This non-systematic character of the cases is one

problem with maintaining that a conventional implicature is

what is involved in these cases. If the implicature was

conventional then there should be some feature of the

utterance which is responsible for the implicature that S

is requesting something. It is not easy to find a candidate

for this feature of the utterance. It does not appear, as

the above discussion illustrates, that the grammatical form

of the expression can play this role. Nor does there appear

to be anything about the word "salt" or "time" or "know" in

the examples that will serve.

Even the suggestion that the examples mean such and

such, but are idiomatic, which explains why they lack the

systematic character of other cases of expression meaning,

does not excuse them from other questions that can be

raised about attempts to argue that meaning is what is

involved. The examples do not seem to bear the proper

relationship to other expressions which it would be

233

reasonable to expect they would if the job they were used to

perform they performed because of their meaning.

For example, often utterances which express wants

or desires of the speaker are used to request. In fact,

questions about what a person intends to order in a

restaurant are often phrased in just this way. The question

put to the customer is, "What would you like?," and the

response to the effect, "I'd like the special." However,

there is nothing inconsistent or odd or, I have been told,

even uncommon about the response, "I'd like the steak, but

bring me the diet plate." Such a conjunction of two

imperatives or performative requestives is either unintel­

ligible, inconsistent, or (maybe) a request for both.

The way around this objection is to claim that the

expression in question is ambiguous. There is no incon­

sistency because the speaker is using another meaning of

the expression, "I would like the steak." The phrase is

ambiguous between an assertion about the speaker's wants on

the one hand, and an idiom of requesting on the other. If

they are ambiguous then a hearer should have no trouble

with, "I would like the salt, but I would not like the

salt."

It does not seem that such expressions are

ambiguous. How they seem is not, obviously, a telling

argument. But it counts for something so what needs to be

done is to try to find another account of what is involved.

234

Before attempting to find a way of not multiplying meanings

I will consider one more reason for avoiding the explanation

in terms of expression meaning. Like this one, it appeals

to how the facts appear.

To hold out for an account of the data in terms of

meaning forces acceptance of specifications of meaning which

just seem wrong. Like the previous point this is not con­

clusive. It does not seem, though, that if someone were

asked to provide a specification of the timeless meaning of

"Pass the salt!" and "Can you pass the salt?" that he would

provide the same specifications even though one is often

used in place of the other. Nor does it seem that a

specification of the meaning of "I would like some tea,"

would be to the effect that the hearer is to get some tea

for the speaker.

On the one hand it does seem to be the case that

these expressions are closely tied to the performance of

illocutionary acts of requesting. They are too closely

tied to be understood as or analogously to conversational

iraplicatures, On the other hand I do not wish to say that

they mean, for example, that the speaker requests that the

hearer shut the door. Is there some point between requiring

to be worked out a la conversational implicature and

expression meaning which is not a matter of intending a

certain sense and reference?

235

How is it that A can grasp what S means immediately

by the expression without the expression itself coming to

mean that? S may intend A work it out, but if S is aware

that someone has used the expression with A to mean that A

is to take off his hat S need not intend that A work it out

to understand what S meant. Each could know that in the

past the expression has been used by people to tell other

people to remove a hat so that must be what is going on now.

Notice how what we would say about this is different

from what we would say about an expression like, "At ease!"

A would never be able to work out what S meant when he

uttered the expression by attending to facts about the

expression, A will either have to be told or figure it out

by watching others respond. When S utters it, if he expects

A to understand immediately what he meant, it is not because

of A's having worked anything out previously, but solely

by what is mutually expected if the expression is uttered.

Suppose that H had merely observed S ask A if A's

head was cold, and had seen A subsequently remove his hat,

H was not able to work it out himself, and saw no possible

way it might have been worked out. H witnessed others

respond in the same way A had, by removing their hats.

Unlike S and A, H might very well believe that, at least

among those in the army, that "Is your head cold?" means the

same as "Take off your hat!" Like, "At ease!" it is just

a way of giving a command. Should H someday become a drill

236

sergeant instructing other soldiers not unlike himself the

expression may come to mean this, at least for his platoon.

Should H get a job teaching other drill sergeants or writing

drill manuals it might soon come to mean this for the whole

army.

What is the difference between this last case and

the way in which S and A understand the expression? Let us

look back to the definition of non-composite utterance type

meaning given in the previous chapter to see how well it can

capture the difference (Schiffer 1972, p. 130). It is an

analysis of non-composite meaning, but that should not

matter here. The first clause of the definition appears to

be met by the subsequent situation of S and A. It would be

mutually known* that someone who intended someone to remove

his hat might utter, "Is your head cold?" Clearly this

would be true after the members of G had all worked it out

or witnessed someone working it out, and knew that everyone

had done this. This is just the situation we wish to

discuss. Then, what about the second clause? Would anyone

in G who uttered x in order to mean P (the values here

should be obvious) intend to realize that state of affairs

E which includes the fact that x is such that there is a

precedent in G for uttering x and meaning P? If everyone

had previously worked it out, and people supposed that

current members of G could then grasp what was meant without

working it out, then would not there be a precedent for

237

uttering the question and meaning that the hearer is to

remove his hat? Unfortunately, on normal meanings of

"precedent" there would be such a precedent.

However, it does not seem we would wish to say

that it means this. New members of G who learn the language

will have to work it out before there would be a precedent

for them. The trick is to capture this in the analysis. I

think the following requirement will do this, and might

also keep generalized implicatures from meeting the require­

ments for being cases of utterance meaning. What needs to

be added to Schiffer's analysis is that there is a

precedent in G for uttering x and meaning that P solely in

virtue of uttering x and meaning that P. That is, the

precedent exists just because people utter the one to mean

the other. This modification does not appear to have

adverse effects on the analysis.

In the case where a person is told to remove his

hat by being asked if his head is cold it is not the case

that there is a precedent for uttering x and meaning that P

just in virtue of uttering x and meaning P. There is a

precedent for uttering the one to mean the other partly

because the members of G have worked it out that when

someone uttered one the other was what he meant. The

reflexivity of the intentions in speaker meaning have a

counterpart in utterance meaning.

238

Another possibility for modifying the analysis of

utterance type meaning is to rule out these troublesome

cases with a negative existential clause. The definition

would require that there is a precedent for uttering x and

meaning that P, but add that it is not the case there exists

a precedent in virtue of some feature such as S's having

worked it out. This might produce less problems in going

on to construct a definition of composite meaning.

Without going into the detail I did in the case

above I would like to offer a suggestion on how a case like,

"Can you pass the salt?" is capable of analogous treatment.

If the general view outlined above is correct then which

cases are instances of what is an empirical question.

Since it is an empirical question whether what follows is

correct I offer it only as a possibility to show there are

analogous ways of handling these indirect speech acts.

There is a general requirement of politeness that

we do not tell people to do things. Children are taught

to ask rather than to tell. Like the request for someone

to remove his hat, understanding what was meant becomes

immediate rather quickly. In the case where the question

about the person's head being cold is a request, working out

the speaker's meaning is no longer needed after the second

time, and quickly after that people develop mutual expecta^

tions about other responses to the utterance of the expres­

sion. Questions like, "Can you pass the salt?" are, first,

239

much more common than are questions about head temperature;

second, not nearly as complicated as far as working out

goes; and third, learned very early in life. This would

account for their close tie to the requestive meaning.

Asking this question is one thing we would expect

someone to do if he were to request that someone pass the

salt. However, this does not show that requesting is part

of the meaning as we might expect a speaker to ask about

head temperature to get someone to remove a hat. However,

the request is not part of the meaning. Once sensitized

to the working out other questions about the head are

immediately understood as requestives, for example, "Is your

head hot?" is an order to put a hat on, "Is your head

crooked?" an order to straighten a hat. These items would

not mean that someone were to do such and such.

Analogously, some long lost working out schema which

we went through early in the learning of the language might

be involved in the "conventional" indirect speech acts.

Whether or not this is correct, the point is that there are

ways to account for indirect speech acts which are tied to

certain expressions other than by claiming that the speech

act is part of the meaning of the expression. There are

certain features of the expression in virtue of which it

could be worked out that a speaker meant such and such by

it. Once this is done then on subsequent occasions there

will be no need to work it out, S and A can come to have

240

mutual expectations that A will grasp it without working it

out because speakers have meant such and such by uttering

it before. Some cases may be such that the child is just

told to substitute one expression for the other, and in

these cases we might be tempted to attribute the new speech

act to the meaning of the expression. Not all seem to be

matters of meaning, and the foregoing account may help to

explain why.

Performative Utterances

Since Austin (19 62) argued that performative

sentences are neither true nor false the issue has gathered

proponents on each side, Since an account has been

developed both of illocutionary acts and, although to a

much lesser extent, utterance meaning in the preceding

pages an examination of the issue is warranted.

Austin maintained that performative sentences were

neither true nor false. According to him utterances like,

"I order you to leave," although indicative in form are not

statings (constative), and so cannot be either true or

false. To perform an illocutionary act by using the

explicit performative sentence is to perform the act named

by the main verb, and not to state that one is performing

the act. The performative prefix is said to make explicit

what the speaker is doing. It does not serve to allow him

to state what he is doing. Those siding against Austin

241

would say that while the speaker is not just stating that

he is doing something, but also doing it, it is not correct

to conclude that he is not also stating he is performing the

act. Hence, what he stated here, as in other statings, is

capable of being true or false.

Note that the dispute is about performative

utterances used to perform a particular illocutionary act,

and not about any performance of that act. Everyone would

agree, hopefully, that someone who uttered "Leave the

room!" had not uttered something that could be either true

or false. Anyone who wishes to argue that explicit

performatives are constative must be able to account for the

fact that in one performance of the act the speaker is doing

something which in another performance of the act he is not

doing, namely stating what it is he is doing.

Given that explicit performatives would by their

grammatical form appear, when uttered literally, to be

statements what about them would suggest to anyone that they

were not statements. Not many explicit arguments have been

forthcoming on the issue. I would like to offer some

reasons that might have been behind what lay behind the

view.

One apparent reason is that the speaker is trying to

do something besides stating, and this something else is

what he is primarily attempting to do. Were he stating

this would mean the speaker was in the production of one

242

utterance doing two things at once, one of them a stating

about what he was doing. This may have seemed paradoxical

to some. However, other utterances have this feature, and

it is not considered at all paradoxical. The utterance of

"I am calling you a liar," has just this characteristic of

being a doing and also an announcement that one is doing

something. Someone who tests a public address system by

saying, "testing" might both be testing as well as letting

anyone in hearing range know what it is that he is doing.

Another example (due to Harnish [1975]) would be, "I am

speaking English." There is nothing odd about using an

utterance both to do something, and to tell what it is one

is doing with that utterance.

Another reason for withholding claims of truth and

falsity might be constructed along the following lines. If

a performative utterance which genuinely is the performance

of that speech act is true, then those performances which

fail to be the performance of that speech act should be

false. If, to update an example of Austin's (1962), at the

christening of a ship at the appropriate moment S pushes the

official out of the way and says, "I christen thee the S,

S. Chairman Mao," he does not christen the ship. However,

is the correct description of the situation that what S

said is false? But if the sentence would have been true if

he had christened why should there be any hesitancy to say

it is false when he did not?

243

Two responses might be made. Those with a penchant

for the taste of bullets might just maintain that if S did

say this then S said something that was false.

Another response that might admit that just to say

S's utterance in this case was false is not quite correct,

is to develop an explanation of the hesitancy that is felt.

Comments that Austin (1962) has made on the infelicities to

which performatives are subject might be revealing in this

respect. He notes that some of the problems are very akin

to the difficulties that beset constatives when the pre-^

supposition of that statement fails to hold. Some, Austin

among them, have suggested that in uttering a sentence like,

"All of my children are bald," S failed to make any state­

ment, and has therefore said nothing that can be either

true or false, if S has no children. But, if S has children

all of whom are bald, then what he said is true.

Certain sorts of conditions are necessary for the

act performed to have been one of christening a ship, among

them that the person performing the act had been granted the

right to do so. These sorts of conditions might be con^

sidered analogously to the presupposition for S's statement

about his children. That in this case some of the condi­

tions necessary for what was done to be christening were not

met, and so that it is not correct to say that what he said

was false does not show that in other cases what he said

might not be true. Just because what S said is sometimes

244

neither true nor false does not show that it is never true.

The consequence of denying this is that almost every state­

ment ever uttered could never be true, since possibly for

every contingent statement at one time its presuppositions

were not met.

There are, then, two responses to this reason for

believing that performatives are neither true nor false.

The latter involves entering the dispute on presupposition.

The other is to just argue that the statement in question

just is false.

A rather ingenious argument that performatives are

not used to make statements is due to Schiffer (1972). The

argument will be clearer after developing an account of how

performatives might be taken to be statings that one is

ordering or warning or so on.

In order to determine if the literal utterance of

performative sentences might also be statings it is neces­

sary to provide some account of what it is to state some­

thing. There are a number of dimensions of difficulty to

producing a difficulty. If explicit performatives are

statings then in uttering, "I warn you that Harry is looking

for you," S stated that he was warning, but in warning by

uttering, "Harry is looking for you," S did not. This can

be refined further. With the kinds of theories I think have

the most chance of being correct S performed an illocutionary

act if and only if he meant something, and stating is an

245

illocutionary act. Illocutionary acts are identifiable

either by what was meant, the values of (F)P, or by the

reasons A was to have. Hence, there will be one of these

elements present in the former example that was not present

in the latter. I will assume that it is in the (F)P that

the additional information is to be found as it does not

seem that in stating S intends any special reason be opera­

tive.

In the two warnings S meant in one that he was

warning, but in the other did not. Those who argue that

performatives are either true or false must explain why in

one case the vehicle of meaning that (F)P is true and in

the other it is not.

Another difficulty to be met in producing an

analysis of stating is that stating seems more closely tied

to what is said than are some other illocutionary acts. Not

every way S might come up with of meaning (F)P or P will

count as stating. For example, S can request that someone

do something by implicature, but to use "state" to describe

the situation requires that it not be the case that what was

stated was implicated or suggested. If (F)P was only

implicated even though this is what S meant, S did not

state. In a large number of cases S can perform the

illocutionary act by implicating something. The proper

analysis of stating must insure that it was not meant via

some mechanism of implicature.

246

On the other hand it is not enough that what S said,

the words he used, meant P. If S was being sarcastic when

he said, "Nixon is a great pianist," it would be incorrect

to accuse him of stating that Nixon was a great pianist.

Further, what is stated by S can be more than what

S said even though he meant, at least, what he said. If S

uttered "He'll fix your radio," meaning that Jones would

fix the radio, then in circumstances where it was clear

what he meant then we do wish to say that S stated that

Jones would fix the radio. In determining what was stated

we must determine what the speaker meant. This will some'-

times be more than what S's words themselves meant, How­

ever, not every case of a speaker meaning such and such

beyond what his words mean will be a stating.

Admittedly there will be no clear boundaries in

ordinary language as to where the boundaries are to be drawn

between statings and implicatings on the one hand and

statings and sayings, in the sense of "Tell me exactly

what he saidi," on the other. Intuitions will clash at

the borders. I adopt the working assumption that the

theory that gets the most material right in the most

systematic way is free to settle the boundaries.

I propose the following account of stating, It is,

like telling, merely P and not (F)P-identifiable. In

uttering "Jones is a crook," S stated that Jones is a

crook, and what S meant is that Jones is a crook. Now,

247

some way must be found to tie stating more closely to the

utterance actually produced. Suppose that we can identify

a special class of cases of speaker meaning where what the

speaker means over and above what his words mean can be

determined just from the reference he intends for any term

that might have more than one reference, and the meaning he

intends to be operative for any term that might have for

than one meaning. (See the previous section for a more

detailed account of these cases.) For example, suppose in

uttering, "He is the squealer," S meant that Jones (not

anyone else) is the person who talked to the police (not

the person whom emitted high-pitched noises). Here deter^

mining what the speaker meant is determinable from the

utterance off the two items mentioned above, I will label

these particular cases of speaker meaning, "c-meaning" for

"constative meaning." Then S stated that P if and only if

S c-meant that P. Notice that if S c-meant that P he also

meant that P, so the view that S performed an illocutionary

^ct if and only if he meant something still holds up.

Suppose S warns A by saying "Jones is after you."

Then S meant that Jones is after A, and that this

constitutes some danger for A. Clearly S did not here state

that he was warning A, and this is just what the above

analysis of stating would say about the case. What it would

say that S stated is that Jones is after A. It is not

248

unreasonable to say that S stated this, although he also

did something more.

In a sentence like, "Jones is after you," the con­

ventional force would be constative. This is what the

indicative form comes to. If Jones uttered this sentence

literally, meant what the sentence meant, then he would

have constated something. In virtue of meaning what the

indicative sentence meant S stated. Saying that an

utterance has such and such a conventional force is to say

that if someone uttered it literally, meant what it is

expected someone who uttered the sentence would mean, then

he would have said something with such and such a force.

Likewise, we can see how the conventional force of an

imperative sentence would be the force of meaning that

someone were to ip.

Now we are ready to see if performative utterances

might be used to make statements. Suppose that instead of

"Jones is after you," S says, "I warn you that Jones is

after you." If S uttered this sentence literally, then we

should suppose him to have meant what his words meant. Let

us suppose that the sentence means that the speaker warns

the hearer that such and such. If the speaker uttered this

sentence literally, and the indicative form carries the

constative force, then he would have meant that he was

warning the hearer than P. S would have stated that he was

warning. Again, S was warning the hearer, but if the

249

indicative form carries the force of meaning that P, then on

parity of reasoning with the simple indicative example S

would have meant that he was warning. And, he would have

c-meant it. So, S stated that he was warning. If S uttered

"Jones is tired," literally then it is correct to say that

S meant that Jones is tired. Hence S stated. But, if S

uttered "I warn you that Jones is after you," and what this

sentence means is that the speaker warns the hearer someone

is after the hearer, then did not S meant that he was

warning the hearer that someone was after him. If there is

not a reason to deny that the indicative form carries the

force of meaning that P, then S stated he was performing the

illocutionary act.

But, Schiffer (1972, pp. 108-109) has constructed

an argument that this cannot be so, that S could not have

stated that he was warning. I will modify Schiffer's argu­

ment so that it will apply to the analysis I have given,

and apply it here. The basic form of the argument is kept

intact. Schiffer argues that it should always be possible

to make the force of a person's argument explicit, but if

in uttering, "I order you to leave," S stated that he was

ordering then it should not be possible to make the force

of this utterance explicit. To do this S would have had to

utter, "I state that I order you to leave." If he meant

this, then it should be possible to make this force

explicit, S would have had to utter, "I state that I state

250

that I order you to leave." An infinite regress is upon

us. Since Schiffer concludes that it is possible to make

the force of an utterance explicit, performatives are not

uttered with their full conventional force.

The problem, to repeat, is this. Suppose S means

what his words mean. If his words exhaust what he means

then a specification of what S means would be the conven­

tional meaning of the utterance. But then we would not say

that S means what the words mean for they do not mean that

S means such and such. The conventional force of an

utterance is the force the utterance would have if S uttered

it and meant it. The meaning of the expression will always

lag one step behind what S means.

I think I can show that the argument fails. I will

do this by showing that it has consequences that we would

not wish to accept. Consider again, "Jones is tired," It

is apparently constative in logical form, the kind of

expression whose conventional force would be that of a

statement, Schiffer (1972, p, 108) says, "If it were true

that in uttering, e.g., 'I (hereby) order you to leave' one

were asserting that one was ordering it would not be possible

for one to make explicit . . . the illocutionary force." I

suggest that if this is the case then no expression can ever

be uttered with its full conventional force. If the con­

ventional force of the sentence, "Jones is tired," is

constative, then it should be capable of being made explicit

251

that S means that Jones is tired. (It cannot be argued that

the indicative mood makes this conventional force explicit

already for the following reason. If it does, then the

conventional constative force of "I order you to leave," is

already explicit, and Schiffer's regress argument stops.)

For the force of the utterance to be made explicit S would

have had to utter, "I state that Jones is tired," But

Schiffer's principle comes back to haunt him. To make the

force of this explicit S would have had to state that he

was stating that he was tired. It would follow that even

ordinary indicatives are not uttered with their full conven­

tional force. Hopefully, however, a theory would have the

consequence that the literal uttering of them is a statement.

I would not wish to claim that in uttering, "I

state that Jones is tired," S stated that he was stating.

But if the reasoning in the case where he uttered, "Jones

is tired," is applied to the former utterance, if S meant

what his words meant then in the former sentence he did

state that he was stating. This is because of the conven­

tional force of the indicative form. If S did not state

that he was stating there are two explanations of this.

First, the presence of the performative verb, "state" just

is equivalent to the indicative form without the verb, and

in some way its presence blocks the further constative-

making character of its own indicative form. The second

choice is to follow Schiffer, and to say that performative

252

sentences are not uttered with their full conventional

force. This has questionable consequences.

Here is the problem for anyone who would take the

former approach, and argue that other performative sentences,

such as those with "order" are constative and either true or

false. If they are they must operate differently than the

constative performative verbs like, "state," "tell," and

"assert." An explanation is needed of this difference. I

do not have an explanation of how this occurs, but it is

worth pointing out that something akin to the performative

analysis (Ross 1970) could explain the fact. Explicit

performatives with the constative performative verbs are

surface realizations of underlying performative forms.

Such a linguistic supposition would offer support

for the view that performatives are capable of being either

true or false by handling one of the difficulties of

maintaining that they are. Even the presence of an under­

lying constative performative form would not decide the

issue of whether all performative sentences are either true

or false. There might remain other reasons for withholding

this dimension of assessment. The theory developed in this

study does not require deciding the issue in any particular

way. A more comprehensive theory of language use would.

I suppose that the way the issue is to be decided is to

continue to examine the consequences of supposing that

performative sentences are or are not true or false, to see

253

which features they share with sentences that clearly are or

are not either true or false, and to see what arguments can

be mounted on either side of the issue. In the first pages

of this section I offered that the arguments presented so

far to show performative sentences are not susceptible of

truth and falsity have been found wanting. The dispute

remains to be settled. That I leave as an exercise for the

reader,

REFERENCES

Armstrong, D. M. "Meaning and Communication," The Philosophical Review, LXXX (October, 1971), pp. 427 477.

Austin, John L. How to Do Things with Words. Oxford: Oxford, 1962.

Byerly, Henry. Department of Philosophy, The University of Arizona, personal communication, 1975.

Chomsky, Noam. Aspects of the Theory of Syntax. Cambridge Mass.: The M.I.T. Press, 1965.

Cohen, L. Jonathan. "Do Illocutionary Forces Exist," Jay F Rosenberg and Charles Travis, eds., Readings in the Philosophy of Language. Englewood Cliffs, N. J.: Prentice-Hall, 1971, pp. 580-599.

Cowan, J. L. Department of Philosophy, The University of Arizona, personal communication, 1975.

Fodor, Jerry A. Psychological Explanation: An Introduction to the Philosophy of Psychology. New York: Random House, 1968.

Fodor, Jerry A., Thomas G. Bever, and Merill F, Garrett, The Psychology of Language: An Introduction to Psycho-Linguistics and Generative Grammar. New York: McGraw-Hill, 197 4.

Frye, Marilyn. "Force and Meaning," The Journal of Philosophy, LXX (May 24, 1973), pp. 281-294.

Grice, H. P. "Meaning," The Philosophical Review, LXVI (1957), pp. 377-388.

Grice, H. P. "The Causal Theory of Perception," Robert J. Schwartz, ed., Perceiving, Sensing, and Knowing. Garden City, New York: Doubleday, 1964, pp. 444-451

Grice, H. P. The William James Lectures. Given at Harvard University, 1967.

254

255

Grice, H. P. "Utterer's Meaning and Intentions," The Philosophical Review, LXVIII (April, 1969), pp. 147-177.

Harman, Gilbert H. "Stephen R. Schiffer: Meaning," The Journal of Philosophy, LXXI (April 18, 1974), pp. 224-229.

Harnish, R. M. Department of Philosophy, The University of Arizona, personal communication, 197 5.

Katz, Jerrold J. Semantic Theory. New York: Harper and Row, 197 2.

Rosenberg, Jay F. Linguistic Representation. Boston: Reidel, 1974.

Ross, John R. "On Declarative Sentences," Roderick A, Jacobs and Peter S. Rosenbaum, eds,, Readings in English Transformational Grammar. Waltham, Mass.: Blaisdell, 1970, pp. 222-272.

Schiffer, Stephen R. Meaning. Oxford: Oxford, 1972.

Searle, John R. Speech Acts. Cambridge: Cambridge, 19 69.

Searle, John R. "What is a Speech Act," John R. Searle, ed., The Philosophy of Language. Oxford: Oxford, 1971, pp. 39-53.

Searle, John R. "Indirect Speech Acts." Unpublished manuscript, Department of Philosophy, University of California, Berkeley, 197 5.

Strawson, Peter F. "Intention and Convention in Speech Acts," John R. Searle, ed., The Philosophy of Language. Oxford: Oxford, 1971,

Wittgenstein, Ludwig. Philosophical Investigations. New York: Macmillan, 1953.

Ziff, Paul. "About What an Adequate Grammar Couldn't Do," Jay F. Rosenberg and Charles Travis, eds., Readings in the Philosophy of Language. Englewood Cliffs, N. J,: Prentice-Hall, 1971, pp. 548-556.