i

I N T E R P R E T I N G J .  L .   AU S T I N

In this volume, Savas L.  Tsohatzidis brings together a team of

leading experts to provide up-to-date perspectives on the work of
J. L. Austin, a major figure in twentieth-century philosophy and an
important contributor to theories of language, truth, perception,
and knowledge. Focusing on aspects of Austin’s writings in these
four areas, the volume’s ten original essays critically examine central
elements of his philosophy, exploring their interrelationships, their
historical context, their reception, and their implications for key
issues of contemporary philosophical research. The volume deepens
our understanding of Austin’s philosophy while illustrating its con-
tinuing significance, and will appeal to students and scholars of mod-
ern philosophy, particularly to those interested in the philosophy of
language and epistemology.

Savas L.  Tsohatzidis is Professor of General Linguistics and the

Philosophy of Language at Aristotle University of Thessaloniki. His
previous publications include John Searle’s Philosophy of Language:
Force, Meaning, and Mind (Cambridge, 2007)  and Intentional Acts
and Institutional Facts: Essays on John Searle’s Social Ontology (2007).
iii

INTERPRETING
J .  L .  AU S T I N
Critical Essays

Edi ted by
S AVA S L .  T S O H AT Z I D I S
Aristotle University of Thessaloniki
 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06-04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107125902
DOI: 10.1017/9781316421840
© Savas L. Tsohatzidis 2018
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2018
Printed in the United Kingdom by Clays, St Ives plc
A catalogue record for this publication is available from the British Library.
ISBN 978-1-107-12590-2 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of URLs
for external or third-party internet websites referred to in this publication and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
v

For Olga and Sophia

vii

Contents

List of Contributors page ix

Acknowledgments xi

Introduction 1
Savas L. Tsohatzidis
1 Exploring Austin’s Galaxy: Searching for Truth through the
Lens of Ordinary Language 15
Marga Reimer
2 Levels of Linguistic Acts and the Semantics of Saying and
Quoting 34
Friederike Moltmann
3 On the Representation of Form and Function: Imperative
Sentences 60
Robert Fiengo
4 Uptake in Action 79
Maximilian de Gaynesford
5 Performativity and the “True/False Fetish” 96
Savas L. Tsohatzidis
6 The Vulnerability of Reality: Austin, Normativity,
and Excuses 119
Sandra Laugier
7 Berkeley and Austin on the Argument from Illusion 143
Robert Schwartz
8 Austin on Perception, Knowledge, and Meaning 165
Krista Lawlor

vii
i

viii Contents
9 Enough is Enough: Austin on Knowing 186
Guy Longworth
10 Knowledge and Knowledge-Claims: Austin and Beyond 206
Stephen Hetherington

Bibliography 223
Index 235
ix

Contributors

Robert Fiengo is Professor of Linguistics, Emeritus, at The Graduate

Center, City University of New York.
Maximilian de Gaynesford is Professor of Philosophy at the University
of Reading.
Stephen Hetherington is Professor of Philosophy at the University of
New South Wales, Sydney.
Sandra Laugier is Professor of Philosophy at the University of Paris
I Panthéon Sorbonne.
Krista Lawlor is Professor of Philosophy at Stanford University.
Guy Longworth is Associate Professor of Philosophy at the University
of Warwick.
Friederike Moltmann is Director of Research at the Centre National de
la Recherche Scientifique, Paris, and Visiting Researcher at New York
University.
Marga Reimer is Professor of Philosophy at the University of Arizona.
Robert Schwartz is Distinguished Professor of Philosophy at the
University of Wisconsin, Milwaukee.
Savas L.  Tsohatzidis is Professor of General Linguistics and the
Philosophy of Language at Aristotle University of Thessaloniki.

ix
xi

Acknowledgments

I am grateful to Hilary Gaskin of Cambridge University Press for welcom-

ing and supporting this book project, and to the contributors for their
thought-provoking essays.
Special thanks are due to Lucy Nusseibeh, J. L. Austin’s younger
daughter, for licensing the use of her father’s photograph on the book’s
cover.
To my wife Olga and my daughter Sophia I owe many more things than
my dedicating this book to them acknowledges.

xi
i
1

Introduction
Savas L. Tsohatzidis

J. L.  Austin published few papers in the course of his short life (1911–
1960) – they were brought together, along with a few until then unpub-
lished ones, in the 1961 collection of his Philosophical Papers  – and it is
unlikely that the two slim books published in 1962 on the basis of his notes
for two series of lectures – How to Do Things with Words and Sense and
Sensibilia – would have been published by him in the sometimes sketchy
form in which they have been preserved for posterity.1 But his surviving
output contains enough material for understanding the fascination that
his work has exercised on many of his philosophical contemporaries, as
well as the unease that it has provoked in some others among them.
The fascination was evidently related to the striking originality of the
proposals that he was led to make in discussing almost all of the time-
honoured topics he has written about, and to his remarkable skills in
working out his way to those proposals by noticing hitherto unsuspected
differences between deceptively similar phenomena, as well as hitherto
unsuspected similarities between superficially unrelated phenomena, often
on the basis of an uncommonly patient and perceptive examination of
the variety and complexity of the ways in which linguistic expressions
are related to the contexts in which they are used. The unease was prob-
ably due to the fact that his proposals appeared to constitute not so much
answers to the questions traditionally asked about the time-honoured top-
ics whose discussion had occasioned them, but rather invitations to pose
different kinds of questions regarding those topics, and to set out to answer
them without making the sorts of mistakes that, in his view, rendered the
traditional answers and the traditional questions suspect  – most promi-
nently, the mistake of seeking and formulating generalizations without

1
Throughout this volume, references to Austin’s three books are to their latest editions: third edition
(Austin 1979) for Philosophical Papers, second edition (Austin 1975) for How to Do Things with Words,
first edition (Austin 1962) for Sense and Sensibilia.

1
2 Savas L. Tsohatzidis
antecedently examining lots of relevant cases, and without paying atten-
tion to how differences and similarities between particular cases would
ordinarily, as opposed to academically, be described and understood.
Austin’s work has influenced various subsequent developments within
analytic philosophy and related fields, sometimes in ways that are not widely
known,2 both by providing fresh insights into familiar topics and by creat-
ing the conceptual space for the discussion of some unfamiliar ones. And
its impact has been increasing in recent years, as suggested by the number
of books in diverse philosophical subfields that rely on Austinian ideas at
key points in the development of their arguments (for example, Schwartz
2006; Crary 2007; Fiengo 2007; Travis 2008; Langton 2009; Fischer 2011;
Lawlor 2013; Bauer 2015; de Gaynesford 2017) or are specifically dedicated
to the examination of Austin’s own arguments (Gustafsson and Sørli 2011;
Laugier and Al-Salech 2011; Garvey 2014) and of their metaphilosophical
implications (Baz 2012; Laugier 2013; Maddy 2017).
The present volume is a further outcome of the current intensification
of interest in Austin’s work. Its ten original essays critically address key
aspects of his contributions in, primarily, four areas – the theory of truth,
the philosophy of language, the philosophy of perception and the theory
of knowledge –, aiming to deepen our understanding of those contribu-
tions and of their contemporary significance.
The volume begins with Marga Reimer’s essay, ‘Exploring Austin’s
Galaxy: Searching for Truth through the Lens of Ordinary Language’,
whose focus is on the account of truth that Austin presented in his 1950
article ‘Truth’ and further defended in his later, posthumously published
paper ‘Unfair to Facts’ – an account in which Austin eschews the idea of
an isomorphism between vehicles of representation and objects of repre-
sentation that was characteristic of traditional formulations of the corre-
spondence theory of truth, and proposes a novel formulation that aims to
make better sense of the way in which truth is commonly conceptualized.
According to Reimer, the basis of Austin’s elaboration of the correspond-
ence idea in these two papers is the observation that the often densely met-
aphorical vocabulary usually deployed in discussing the ‘fit’ between words
and world that the correspondence theory purports to elucidate represents
the ‘fit’ in question as a matter of degree; and the distinctive outcome of

2
To give just one example, Austin’s account of truth has been an important inspiration for the devel-
opment of situation semantics within linguistics and philosophy; see Barwise and Perry (1983)
and Barwise and Etchemendy (1987) for two key works in that tradition which acknowledge the
Austinian influence, and Kratzer (2014) for an overview of recent developments.
3

Introduction 3
Austin’s elaboration of that observation – and one that past and present
critics (and even allies) of Austin have, in Reiner’s view, missed, in part
because of outdated views concerning the significance of metaphorical lan-
guage and its interpretation – is that truth is not only a relational phenom-
enon (as correspondence theorists had always insisted) but also a ‘spectral’
phenomenon (that is, one that admits of degrees) – indeed, that it is a
spectral phenomenon precisely because the relation (of correspondence)
that constitutes its core is, literally understood, itself spectral: as Reimer
puts it, ‘the idea that truth is correspondence to the facts, when coupled
with the idea that correspondence admits of degrees, leads naturally to the
view that truth itself is a degreed or (in other words) a “more or less” kind
of phenomenon.’ Reimer contends that, by thus elaborating it, Austin
revolutionized the correspondence theory; and she suggests that the worry
that, in so doing, he might have lost sight of the distinction between what
truth is and how it is commonly conceptualized would be misplaced; for,
in her view, truth is in any case a kind of phenomenon whose nature can
plausibly be taken to be exhausted by its conceptualization, in the sense
that truths and falsehoods are not the sorts of things that the world would
contain if no one was thinking or talking about it, and about the relation
of one’s thought and talk to it.
Austin’s How to Do Things with Words is widely recognized as a classic
work in the philosophy of language, through which an entirely new set of
topics and issues – those relating to speech acts – have entered that field.
The volume’s next five chapters are dedicated to some of these topics and
issues, exploring novel ways in which Austin’s ideas might be fruitfully
connected to issues of contemporary concern, re-examining the conditions
under which aspects of them have been assimilated and other aspects of
them have been discarded, and investigating how Austin’s overall approach
to language dovetails with his general conception of human action.
Friederike Moltmann’s essay, ‘Levels of Linguistic Acts and the Semantics
of Saying and Quoting’, is an extended demonstration of the relevance
of some central distinctions between act levels that Austin introduced in
How to Do Things with Words – in particular, the distinction between the
locutionary and the illocutionary level, and, within the former, between
the phonetic, the phatic and the rhetic level – to the construction of an
adequate formal semantics of speech act ascriptions, and of various forms
of quotation, in natural language. The central idea of her proposal is that,
contrary to what is typically assumed in formal semantics, speech act
ascriptions and other types of so-called propositional attitude ascription
should not be taken to contain references to abstract propositions, but
4 Savas L. Tsohatzidis
should instead be represented as containing references to various types of
non-enduring products (in the Twardowskian sense of the term) of the acts
and attitudes they ascribe; and that the object complements – including,
crucially, the sentential object complements  – of speech act verbs and
attitude verbs within such ascriptions should be taken to semantically
function as predicates attributing various sorts of properties to the act-
products or attitude-products therein referred to, the kinds of properties
being attributed depending on the kind of act or attitude to whose prod-
uct they are being attributed (in the case of speech act ascriptions, on
whether the act, and its product, is locutionary or illocutionary, and, if
the former, whether it is phonetic, phatic or rhetic). After outlining the
various types of semantic evidence favouring this general view against the
propositionalist view, and the way it applies to the semantic representation
of illocutionary act ascriptions, Moltmann concentrates on its applica-
tion to the semantic representation of locutionary act ascriptions, in their
phonetic, phatic or rhetic guises. And she contends that the representa-
tions provided by her semantics support, among others, two noteworthy
conclusions:  First, that, although several influential speech act theorists,
notably John Searle, have supposed that Austin’s category of rhetic acts
is theoretically dispensable (on the grounds that the content-specifying
role of rhetic acts can be accommodated within a re-conception of illo-
cutionary acts as acts of ‘attaching’ forces to abstract propositions), the
Austinian distinction between rhetic and illocutionary acts, interpreted
in a particular way, is both semantically indispensable and explanatorily
potent, as it can be shown to control a variety of crucial linguistic differ-
ences between the corresponding speech act ascriptions, without assuming
that either type of ascription contains ineliminable references to abstract
propositions. Second, that the Austinian distinction, within the locution-
ary level, between phonetic, phatic and rhetic acts, in conjunction with
the Twardowskian distinction between acts and products, and in combin-
ation with a three-dimensional conception of syntactic structure, enables
the construction of a unified, and fully compositional, semantic account
of the various forms of natural language quotation (pure, direct, indirect
and mixed), contrary to what most formal semanticists have so far deemed
possible.
Robert Fiengo’s essay, ‘On the Representation of Form and Function:
Imperative Sentences’, develops a critique of the thesis that a sentence and
the illocutionary act performed in using it can together be awarded a sin-
gle representation. As Fiengo points out, that thesis is foreign to Austin’s
conception of illocutionary acts, but has, in spite of that, been the leading
5

Introduction 5
presupposition of purported incorporations of Austin’s ideas in generative
grammar, and has also been encouraged by the systematic conflation of
questions of form with questions of function in Searle’s philosophically
influential account of illocutionary acts, where it has taken the shape of
the thesis that the structure of every natural language sentence contains an
‘illocutionary force indicator’ encoding the sentence’s function. Fiengo’s
main negative thesis is that such indicators should be eliminated from
the representation of sentential structure, not only because the predictions
about sentential function that their postulation implicitly makes are often
empirically incorrect, but also because the tasks that they are expected to
fulfil, given their supposedly dual role, are sometimes conceptually inco-
herent (as when, for example, the negation of a so-called illocutionary
force indicator is expected to be interpretable as having either a sentence
and an act under its scope). Fiengo’s main positive thesis is that, by strictly
separating the representation of sentential structure from the representa-
tion of sentential function, one opens up the possibility of understanding
why certain types of sentential structure are better equipped than others to
fulfil certain types of sentential function, and of thus explaining the prefer-
ential correlations of particular structures with particular functions, when
such correlations indisputably exist. Aspects of both the negative and the
positive thesis are then developed in more detail by a close examination of
the use of imperative and indicative sentences in realising acts denoted by
verbs belonging to Austin’s category of exercitive illocutionary acts. Acts
belonging to this category, Fiengo argues, can be differentiated from each
other along two intersecting dimensions, resulting in a classification of
them into four subcategories. Of these four subcategories, only two (those
comprising acts whereby speakers exercise their influence on hearers) are
such that their members are preferentially accomplished by means of
imperative sentences, whereas the other two (which comprise acts whereby
speakers exercise their rights or privileges) are such that their members
are preferentially accomplished by means of indicative sentences. And in
both of these types of case, Fiengo argues, the preferences are explicable
not by reference to special ‘illocutionary force indicators’ that imperative
and indicative sentences respectively contain, but rather by reference to
structural differences between imperative and indicative sentences that a
grammar would have to specify independently of any considerations of
illocutionary force. In particular, imperative sentences, in contrast to indi-
cative ones, are sentences that, as a matter of grammar, lack Tense; because
they lack Tense, they can only describe types of activities, without being able
to refer to token activities exemplifying those types; and, Fiengo contends,
6 Savas L. Tsohatzidis
it is precisely because they cannot refer to token activities exemplifying
the types of activities they describe that they are interpreted by hearers as
attempts, by the speakers, to make hearers produce themselves tokens of
those activities; in short, what makes imperative sentences suitable for the
performance of illocutionary acts of exercising influence on hearers is not
that they have a special ‘illocutionary force indicator’ that encodes this illo-
cutionary role and that indicative sentences lack, but rather that they lack a
structural feature that indicative sentences have and that, as such, is quite
independent of illocutionary roles – namely, grammatical Tense.
Maximilian de Gaynesford’s essay, ‘Uptake in Action’, is a critical exami-
nation of Austin’s often quoted but less often scrutinized suggestion that
illocutionary acts require ‘uptake’ on the part of their audiences, as well as an
examination of a much discussed argument by Rea Langton that purports
to derive far-reaching social, political and legal implications from Austin’s
suggestion, by claiming that certain types of cases where a speaker’s audi-
ence does not supply the ‘uptake’ demanded by the speaker’s illocutionary
attempts must be analyzed as cases where the audience in question effec-
tively ‘silences’ the speaker. De Gaynesford argues that Langton’s analysis
cannot be accepted as it stands, since it faces a dilemma: if the absence of
‘uptake’ is understood as implying merely that the speaker does not succeed
in communicating her performance of an illocutionary act to an audience,
rather than that she does not perform the act in question at all, then the
analysis, contrary to its declared purpose, fails to be an analysis of anything
that could legitimately be regarded as a form of silencing; if, on the other
hand, the absence of ‘uptake’ is understood as implying that the speaker
does not merely fail to communicate to an audience that she is performing
an illocutionary act, but does not even succeed in performing one, then the
analysis does depict a possible form of silencing, but only at the cost of
relying on what, as de Gaynesford goes on to argue, is the false premise that
no illocutionary act can actually be performed in the absence of ‘uptake’.
In arguing that this premise is false, de Gaynesford introduces a distinc-
tion between two importantly different respects in which an illocutionary
act may be uptake-dependent  – it may require for its performance the
cognitively active presence of an addressee and it may require for its perfor-
mance the cognitively active presence of a witness – and points out that all
four of the logical possibilities that, given this distinction, could be envis-
aged are actualized: some illocutionary acts are uptake-dependent because
they are addressee-dependent without being witness-dependent; some
others are uptake-dependent because they are witness-dependent without
being addressee-dependent; still others are uptake-dependent because they
7

Introduction 7
are both addressee-dependent and witness-dependent; and finally, some
(and indeed not a few) others are absolutely uptake-free, in that they are
neither addressee-dependent nor witness-dependent. Apart from showing
that Langton’s analysis of silencing can be accepted only in a considerably
weakened form (namely, as applying just to cases where the illocutionary
acts involved are not uptake-free), this state of affairs, de Gaynesford sug-
gests, requires revisiting some general assumptions that have been com-
mon in philosophical accounts of speech acts since Austin. For although,
as de Gaynesford notes, Austin was more guarded than his successors in his
claims about uptake, and arguably ambivalent about the extent of its sig-
nificance, many of his successors have tended to uncritically assume either
that all illocutionary acts are uptake-dependent or that none is; and since,
as de Gaynesford also notes, the distinction between uptake-dependence
and uptake-freedom is not confined to a particular region of illocutionary
space but manifests itself within each of the major categories of illocution-
ary acts that Austin and others have proposed, a reconsideration of impor-
tant theoretical decisions taken on the basis of the uncritical assumptions
is called for.
In my own essay, ‘Performativity and the “True/False Fetish” ’, I crit-
ically examine the virtually unanimous rejection, by post-Austinian phi-
losophers of language, of a thesis about explicit performative utterances
on which Austin particularly insisted, but which he considered too obvi-
ously true to require defence – namely, the thesis that such utterances are
not truth-evaluable, despite being utterances of grammatically declarative
sentences, which have traditionally been regarded as paradigms of truth-
evaluability. If Austin’s thesis were right, explicit performatives would pose
a serious problem to the truth-conditional conception of linguistic content
that has come to be dominant in the philosophy of language, since their
non-truth-evaluability would entail, given that conception, that they have
no content at all. It is therefore not surprising that post-Austinian philoso-
phers of language were interested in rejecting Austin’s thesis. But whether
they have offered good reasons for rejecting it, and for embracing the anti-
Austinian view that explicit performatives do have truth conditions, is a
different question, which the essay proposes to address. In addressing it, I
begin by suggesting that the principal reason offered against Austin’s thesis
and in favour of the anti-Austinian view can best be represented as the con-
clusion of a kind of abductive argument in favour of the denial of Austin’s
thesis: the argument that unless one assumed, contra Austin, that explicit
performative utterances are truth-evaluable one could not explain what
everyone, including Austin, would regard as their most distinctive and
8 Savas L. Tsohatzidis
remarkable feature, namely, that, in issuing them, speakers can accomplish
the illocutionary acts that they thereby name. I then argue that attempts
to justify the denial of Austin’s non-truth evaluability thesis by producing
explanations of performativity that essentially depend on the hypothesis
that explicit performatives are truth-evaluable cannot succeed for at least
two types of reason: on the one hand, because utterances that, on the
proposed explanations, should be explicit performative ones turn out not
to be explicit performative ones; on the other hand, because utterances
that, on the proposed explanations, should not be explicit performative
ones turn out to be explicit performative ones. Since the source of both
of these explanatory failures turns out to be none other than the adoption
of the hypothesis that explicit performative are truth-evaluable, I sug-
gest that they strongly undermine the anti-Austinian view and vindicate
Austin’s thesis, in favour of which I then sketch an independent argument
based on the behaviour of explicit performatives in deductive inferential
contexts (specifically, on the fact that their behaviour in such contexts
could not be reconciled with the hypothesis that they are truth-evaluable
unless one denied the applicability, in those contexts, of certain logically
fundamental inference rules). My conclusion is that the Austinian thesis
that explicit performatives are not truth-evaluable can by no means be
regarded as having been superseded, and that Austin’s opponents might
even have to seriously consider adopting it if some of their own broader
interests were to be safeguarded.
Sandra Laugier’s essay, ‘The Vulnerability of Reality: Austin, Normativity,
and Excuses’, proposes, among other things, an original interpretation of
Austin’s philosophy of language in terms of his philosophy of action. Her
starting point is the claim that Austin’s thesis that explicit performative
utterances constitute realizations of certain actions rather than descrip-
tions of realizations of those actions can best be viewed as arising from
the conjunction of the observation that explicit performatives utterances
have normative force with the assumption that nothing descriptive can
have normative force. Extending that claim, she suggests that, in gener-
alizing his doctrine of performatives to his doctrine of illocutionary acts,
Austin is in effect proposing a resolutely anti-representationalist account
of language as a normative practice (and is not simply interested, as many
interpreters of him have assumed, in supplementing existing representa-
tionalist accounts with a detachable account of illocutionary force). She
then argues that the rationale of Austin’s envisaged anti-representation-
alist account of language as a normative practice can plausibly be traced
to the general conception of intentional action as a phenomenon that is
9

Introduction 9
constitutively tied to normative assessment, which Austin sketches in his
1957 essay ‘A Plea for Excuses’. According to the conception sketched in
that essay, as Laugier interprets it, nothing would be an intentional action
unless it could go wrong in specifiable ways, and consequently the best way
of finding out what a particular type of intentional action is is to find out
under what conditions an action of that type could go wrong, and so under
what conditions its performance, or attempted performance, would be sub-
ject to blame (the study of excuses, as Laugier notes, is, for Austin, essential
to the study of human action precisely because excuses reflect the various
respects in which various types of intentional actions are blameable, and so
the various respects in which they can go wrong). Now, Austin’s main con-
structive proposal, in How to Do Things with Words, as to how the study of
linguistic acts should proceed was that it should take the form of specifying,
for each type of act of this sort, under what conditions its performance, or
attempted performance, would be infelicitous – in other words, under what
conditions its performance, or attempted performance, could go wrong, and
so could be subject to blame. Austin’s main constructive proposal, then, can
plausibly be regarded, Laugier suggests, as an application to the special case
of language of his general view that intentional action is a phenomenon
that is constitutively tied to normative assessment. And it is in this sense,
according to Laugier, that the philosophy of language of How to do Things
with Words can plausibly be regarded as a special case of the philosophy of
action sketched in ‘A Plea for Excuses’.
Austin’s contributions to the philosophy of perception in Sense and
Sensibilia and to the theory of knowledge in his 1946 essay ‘Other Minds’
are prominent examples of his policy of refusing to discuss seemingly lan-
guage-independent philosophical issues without first examining whether
the language in which the issues are formulated and debated can be sup-
posed to have been given enough sense. But although it is clear which
positions and attitudes about such issues Austin was thereby aiming to
reject, it has remained unclear which ones he would be recommending.
The volume’s next four essays propose different answers to that question,
based on readings of Austin’s texts that are informed either by historical or
by contemporary considerations.
Robert Schwartz’s essay, ‘Berkeley and Austin on the Argument from
Illusion’, is an extended defence, presented from an unusual angle, of
Austin’s attack on the argument from illusion in Sense and Sensibilia.
Schwartz interprets that work as aiming to dissolve rather than to solve
the problems that the argument from illusion was supposed to raise for
the philosophy of perception in Austin’s time; and he further argues that
0

10 Savas L. Tsohatzidis
the considerations that Austin was drawing upon for that purpose can be
redeployed for similar purposes in relation to some new problems that the
argument from illusion is being supposed to raise for contemporary phi-
losophy, even after the demise of the particular variety of foundationalism
that was the immediate object of Austin’s attack. What makes Schwartz’s
construction of his interpretation unusual, and gives it added historical
depth, is his parallel argument that each one of the crucial steps that he
discerns in Austin’s attack on the argument from illusion can plausibly
be taken to have been prefigured in, and to receive further support from,
Berkeley’s work on perception, especially in his Essay Towards a New Theory
of Vision. Both Austin and Berkeley, according to Schwartz, deny that the
various metaphysical and epistemological questions about so-called objects
of perception that philosophers have been accustomed to raise can ever
be usefully answered; and both of them suggest that they cannot be use-
fully answered because they rely on assumptions that have no place either
in common sense or in scientific reasoning about the nature of percep-
tual experience, and derive exclusively from philosophers’ tendency, itself
indicative of a mistaken view of how language works, to treat claims that
can only make sense when relativized to context as if they would continue
to make sense even when not so relativized. Given that Austin’s passing
references to Berkeley represent him, in a way that has been and still is
common in discussions of perception, as one of the main architects of the
argument from illusion that Austin attacks, Schwartz’s interpretation sug-
gests that the historical terms in which both Austin and several of his oppo-
nents tend to frame their dispute over the import of the argument from
illusion need to be revised: as far as that particular dispute is concerned,
Schwartz contends, Berkeley is clearly, and significantly, on Austin’s rather
than on his opponents’ side.
Krista Lawlor’s essay, ‘Austin on Perception, Knowledge, and Meaning’,
rejects the view, which is standard among Austin’s critics, and not uncom-
mon among his supporters, that Austin’s contribution to the philosophy
of perception in Sense and Sensibilia has been, and was meant to be, purely
negative. On Lawlor’s interpretation, what Austin aims to do in that work
is rather to use his critique of sense-data theorists as an opportunity for
exposing his positive views as to how philosophical issues should be framed
and investigated, and then to exemplify his recommended approach to such
issues by applying it to particular problems in the metaphysics and episte-
mology of perception, where he delivers substantive results. According to
Lawlor, two guiding assumptions of Austin’s recommended approach are,
first, that, since the role of philosophy is to make explicit and to rationally
11

Introduction 11
reconstruct the common-sense understanding of the world, it is essential
for philosophers to develop an adequate account of the truth conditions
of ordinary claims about the world where that understanding is embodied;
and secondly, that such an account cannot be adequately developed unless
it is governed by the principle that the truth conditions of such claims are
not determined solely by the meanings of the sentences used to make them,
but also by the circumstances in which the claims are made. In previous
work that she briefly summarizes, Lawlor has argued that this ‘situation
semantic’ approach that Austin pioneered in the analysis of language can
be profitably applied to the analysis of knowledge claims, leading to a dis-
tinctive version of the so-called relevant alternatives account of knowledge.
In her present essay, she argues that that account, combined with Austin’s
approach to the interpretation of perceptual reports in Sense and Sensibilia,
leads to a position from which some central issues in contemporary phil-
osophy of perception can also be profitably addressed. The key feature
of Austin’s treatment of perceptual reports that needs to be taken into
account for this purpose, Lawlor claims, is that, although Austin denies
that sentences reporting perceptual experiences can, as such, be divided into
those that are corrigible and those that are incorrigible, or can, as such,
be divided into those that can and those that cannot be conclusively veri-
fied, he does not deny, and indeed insists, that particular uses of sentences
reporting perceptual experiences, can, relative to particular circumstances
in which the reports are made, be correctly characterized as incorrigible
or as conclusively verified. Combining these Austinian notions of de facto
incorrigibility and de facto conclusive grounds with her situation-theoretic
version of the relevant alternatives account of knowledge, Lawlor is then
in a position to argue that a recognizably Austinian response is available to
two key questions in contemporary philosophy of perception – first, the
question of whether one’s seeing something, as opposed to one’s seeming
to see something, provides a special sort of ‘factive’ reason or justifica-
tion (that is, a reason or justification that one could not have unless one’s
perceptual belief was true); and secondly, the question of how, if seeing
does not provide such a ‘factive’ reason, it can nevertheless be sufficient
for knowing. The Austinian perspective, Lawlor contends, allows one to
appreciate that seeing can fail to provide ‘factive’ reasons of the required
error-proof sort (and that, consequently, contemporary epistemological
disjunctivism relies on a false characterization of the opposition between
seeing and seeming to see), but, at the same time, allows one to maintain
the thesis that seeing can be sufficient for knowing (and has the resources
necessary for defending that thesis against closure-related objections).
2

12 Savas L. Tsohatzidis
Guy Longworth’s essay, ‘Enough is Enough: Austin on Knowing’,
proposes to shed new light on the conception of knowledge underlying
Austin’s paper ‘Other Minds’, by addressing two puzzling features of that
paper: that most of it is apparently unconcerned with its ostensible topic,
being dedicated to questions about knowledge in general rather than to
questions about knowledge of other minds in particular; and that, in
addressing those general questions, it proposes substantive restrictions on
admissible challenges to knowledge claims, which may be pre-theoretically
appealing but whose rationale is left inexplicit, and which, moreover, may
seem to conflict with certain theses about knowledge that Austin cannot
plausibly be regarded as wishing to reject (in particular, the thesis that
knowledge is factive, and the thesis that it is closed under known entail-
ment). The key to the resolution of both puzzles, Longworth suggests, lies
in reading Austin’s paper in its historical context – specifically, in reading
it against the background of, and as a contribution to, the epistemological
program of so-called Oxford Realism, whose fundamental commitments
Longworth outlines in the form of six theses gleaned from the writings
of its main exponents. The first puzzle can then be resolved, according
to Longworth, by taking into account the fact that a basic commitment
of Oxford Realism to which there is evidence that Austin subscribed is
that knowing is a distinctive state of mind incompatible with falsity and
different in kind from the state of believing. Given that commitment, to
ask, as Austin does in pursuing his general questions about knowledge,
how one can know that someone else knows something, is tantamount
to asking how one can know that someone else is in a particular kind of
state of mind. And if that is so, Austin’s paper can plausibly be regarded
as being concerned in its entirety with questions about (a particular region
of ) our knowledge of other minds, contrary to what initial appearances
might suggest. The second puzzling feature of Austin’s paper can be elu-
cidated, according to Longworth, by taking into account the fact that
another basic commitment of Oxford Realism to which there is evidence
that Austin subscribed is that knowing is a state of mind that is primitive,
in the sense that it cannot be constructed out of elements distinct from
knowledge. That commitment precludes the existence of independently
specifiable sufficient conditions for knowledge, but does not preclude the
existence of independently specifiable necessary conditions (and, in par-
ticular, the existence of independently specifiable necessary conditions that
would help mitigate the threat of dogmatism that the commitment would
tend to engender, if conjoined with certain other Oxford Realist views).
And it is precisely in trying to articulate one such necessary condition,
13

Introduction 13
according to Longworth, that Austin introduces his proposed restrictions
on admissible challenges to knowledge claims. Longworth then argues that
the necessary condition implicit in these restrictions is best articulated as a
condition of doxastic responsibility in John McDowell’s sense – namely, as
the condition that a knowledge claim is sustainable only if the claimant is
appropriately sensitive to non-question-begging reasons for or against the
truth of what she claims to know – and points out that, thus articulated,
the condition is consistent both with the thesis that knowledge is factive
and with the thesis that it is closed under known entailment. In addition,
Longworth argues that his proposed interpretation accords better with
Austin’s presumable commitments and explicit pronouncements, and sup-
ports an intrinsically better motivated necessary condition on knowing,
than some alternative interpretations of Austin’s position that have recently
been proposed.
Finally, Stephen Hetherington’s essay, ‘Knowledge and Knowledge-
Claims: Austin and Beyond’, explores the relevance of Austin’s treatment
of knowledge claims in ‘Other Minds’ to some contemporary epistemo-
logical projects, and in particular to his own project of developing a so-
called practicalist conception of propositional knowledge on the basis of
the idea that all forms of knowing-that can be analyzed as constellations
of particular forms of knowing-how (where the latter is analyzed in terms
of ability). To have the propositional knowledge that p, for a given p,
is, on this conception, to have the ability to engage in a variety of actions the
common denominator of which is that each is, in a way that is characteris-
tic of actions of its type, ‘directed’ at p and capable of having a bearing on
the truth of p, and would thus be actions actually or potentially manifesting
or expressing the knowledge that p. Assuming this practicalist conception
of propositional knowledge, Hetherington argues that it can be connected
to Austin’s work in both a constructive and a critical spirit. The construct-
ive connection arises from the observation that many, and probably most,
of the actions that, on the practicalist conception, would constitute actual
or potential manifestations or expressions of propositional knowledge are
illocutionary acts in Austin’s sense; and that, in particular, all the acts that
Austin places in the first and the last of his five classes of illocutionary
acts – the class of verdictives (e.g. grading, ranking, assessing, diagnosing,
estimating, etc.) and the class of expositives (e.g. testifying, informing,
conceding, correcting, explaining, etc.) – are acts of this sort. According
to Hetherington, this makes Austin’s account of illocutionary acts an espe-
cially well-suited framework for describing the fine structure of the particu-
lar kinds of knowing-how that, on the practicalist conception of knowledge,
4

14 Savas L. Tsohatzidis
are constitutive elements, rather than causal by-products, of knowing-that;
and it also suggests, according to Hetherington, that the practicalist con-
ception of propositional knowledge can, in its turn, offer a perspective for
a more sympathetic than usual reading of Austin’s controversial claim that
first-person, present-tense ascriptions of propositional knowledge are not
descriptions of one’s inner states, but rather instruments for the perform-
ance of illocutionary acts of giving one’s word. The critical connection
arises from the observation that since, on the practicalist conception, prop-
ositional knowledge is ultimately a set of abilities to act, and since abilities
to act are in general fallible, the practicalist conception invites a resolutely
fallibilist view of propositional knowledge. Hetherington argues, however,
that it is doubtful that the sort of fallibilism about knowledge that Austin
would be prepared to tolerate – whether expressed in his own terms or
in terms of contemporary ‘relevant alternatives’ accounts of knowledge,
which he regards as descending from Austinian ideas – amounts to the
resolute fallibilism that practicalism about knowledge recommends; and
he consequently suggests that, paradoxically, Austin’s variety of fallibilism
about knowledge would not allow him to enjoy the full advantages of an
epistemological view that his distinctive account of language would be
especially well suited to articulate.
Readers of this volume will find in the essays that follow many more
ideas and arguments than my brief summaries have attempted to con-
vey; they may also find that their authors do not ‘bite off more than they
can chew’, as Austin once complained about some famous authors he was
reading.
15

Ch apter  1

Exploring Austin’s Galaxy
Searching for Truth through the Lens
of Ordinary Language
Marga Reimer

Is it true or false that Belfast is north of London? That the galaxy

is the shape of a fried egg? That Beethoven was a drunkard? That
Wellington won the battle of Waterloo? There are various degrees and
dimensions of success in making statements:  the statements fit the
facts always more or less loosely, in different ways on different occa-
sions for different intents and purposes.
Austin, “Truth”

1 Preliminaries
What is truth? I have no answer of my own and don’t propose to advo-
cate for a particular theory of truth. However, by appealing to a num-
ber of insightful points made by J. L. Austin in his papers “Truth” (1950;
1979:  117–133) and “Unfair to Facts” (1979:  154–174), I  will suggest that
truth, as ordinarily conceived, is not only a relational phenomenon, but a
spectrum phenomenon as well. All of this will be in the spirit of Austin and
in the spirit of the correspondence theory of truth, yet will fall short of an
explicit endorsement of that or any other theory of truth. I will also sug-
gest that, contra Strawson (1950), Searle (1998a), and Neale (2001), “cor-
respondence to the facts” is neither a “misleading idiom,” nor an “empty
metaphor,” nor an “idiomatic form of ‘is true.’ ” Indeed, truth as “corre-
spondence to the facts” wears its semantic heart on its linguistic sleeve: it
conveys a word/world relation of correspondence (conformity, fitness, etc.)
that – like correspondence more generally and more literally – admits of
degrees. Hence the idea of truth as a spectrum phenomenon. In place of the
ideas of Strawson, Searle, and Neale, I will suggest that truth as correspon-
dence (fitness, conformity) to reality (the world, the facts) is akin to what
linguists call a “conventional metaphor” – an idea that comports well with
Austin’s own views on truth. I will, however, take issue with one of Austin’s
15
6

16 Marga Reimer
more controversial claims – that “is true,” when predicated of a statement,
is not “logically superfluous.” I will suggest that, although logically potent
in some contexts, the phrase is logically superfluous in others. But even
in contexts of the latter sort, the phrase is (I will suggest) rich in terms of
potential pragmatic (or conversational) implications (or Gricean “implica-
tures”). After a brief discussion as to how the views proposed herein com-
port (or fail to comport) with traditional theories of truth, I will conclude
by suggesting that the “lens” of ordinary language is a reflective lens turned
inward, but is not for that reason a lens incapable of revealing insight into
what Austin calls the “problem of truth.”

2 Austin’s Galaxy and the “Problem of Truth”

Midway through “Unfair to Facts,” Austin asks: Why raise this cry “Unfair
to Facts”? – to which he responds:
The expression ‘fitting the facts’ is not by any means an isolated idiom in
our language. It seems to have a very intimate connexion with a whole
series of adverbs and adjectives used in appraising statements  – I  mean
‘precise’, ‘exact’, ‘rough’, ‘accurate’ and the like. . . . All these are connected
with the notion of fitting and measuring in ordinary contexts, and it can
scarcely be fortuitous that they, along with fitting and corresponding, have
been taken over as a group to the sphere of statements and facts. (Austin
1979: 161)
Austin acknowledges that the expressions in the aforementioned “series”
might be used nonliterally in ordinary everyday assessments of statements.
Yet he does not think that this undercuts the importance of their analysis
when addressing the “problem of truth.”
Now to some extent the use of this galaxy of words in connexion with
statements may be a transferred use; yet no one would surely deny that
these constitute serious and important notions which can be, and should be
elucidated. I should certainly go much farther and claim . . . that these are
the important terms to elucidate when we address ourselves to the problem
of ‘truth’. (Austin 1979: 161)

Yet, as Austin regretfully notes:

All these terms are commonly dismissed along with the supposed useless
‘fitting the facts’. (1979: 161)
Let’s consider these three (connected) passages one by one, beginning with
the first.
17

Exploring Austin’s Galaxy 17
I am in complete agreement with Austin that “fitting the facts,” as well
as “corresponding to the facts,” are not “isolated idioms,” but are rather
part of a “galaxy” of words and phrases used in connection with the ordi-
nary everyday assessment of statements (Reimer 2006). But we should
then ask: Why has this “galaxy” of expressions been “taken over as a group
to the sphere of statements and facts”? Austin admits that this intriguing
phenomenon is “scarcely fortuitous.” But again, what explains it? Perhaps
truth, as ordinarily understood, is a relational phenomenon and, just as
importantly, one that admits of degrees. It would therefore be of a piece
with the expressions in Austin’s galaxy, expressions that suggest a spectral
relation of statements to facts – expressions like “precise,” “exact,” “accu-
rate,” “rough,” “loose,” and “approximate.” As Austin puts it in the quote
that opens the present chapter:
There are various degrees . . . of success in making statements: the statements
fit the facts always more or less loosely. (Austin 1950: 124; 1979: 130)
With respect to the second passage, I am again in complete agreement
with Austin. The use of expressions in the aforementioned “galaxy” is
arguably not always strictly literal. Such use (or uses) may sometimes be
“transferred” – that is, metaphorical or otherwise nonliteral. Thus, while
clothes literally fit persons to varying degrees, statements metaphorically
fit facts to varying degrees. And while maps literally correspond, more or
less, to the terrain, statements metaphorically correspond, more or less, to
the facts.
As to the third passage, I am once again in complete agreement with
Austin. Not only Strawson (1950), but decades later Searle (1998a) and
soon after that Neale (2001) have wrongly dismissed the notion of “fitting
the facts” as useless if not downright misleading. The obvious question is:
Why the dismissive attitude toward such ordinary language expressions,
when these are not used in the strictest and most literal sense possible?
Perhaps because of analytic philosophy’s traditionally dismissive attitude
toward language that isn’t strictly literal. Although this attitude finally
began to change with the seminal work of Max Black (1962) and Monroe
Beardsley (1962), it can still be seen in the works of the three philosophers
discussed in the following section. This dismissive attitude toward nonlit-
eral language is a particularly dangerous one as metaphors are often used
to capture philosophically, scientifically, politically, and otherwise theoreti-
cally important concepts (like that of truth) in cases where the resources
of literal language appear not up to the task. Indeed, why else resort to
nonliteral language? (Reimer and Camp 2006).
8

18 Marga Reimer

3 Criticisms: Misleading Idiom, Empty Metaphor, or

Philosopher’s Invention?
Here, I will examine a couple of objections to Austin’s view that truth is
appropriately conceptualized in terms of the notion of “correspondence
to the facts.” I will look, in particular, at concerns expressed by Strawson
(1950), Searle (1998a), and Neale (2001). I  have chosen to discuss these
particular objections for two main reasons. First, they are representative
of views that are the polar opposite of Austin’s views: while Austin thinks
the ordinary language locutions in his “galaxy” – including “fitting” and
“corresponding to” “the facts” – are crucial to an understanding of truth,
Strawson, Searle, and Neale regard them as irrelevant if not deeply mis-
leading with regard to such understanding. Second, the views of the latter
philosophers (as noted previously) represent outdated thinking about the
philosophical importance of nonliteral language that needs to be exposed
as such.

3.1 Strawson (1950): A Misleading Idiom

According to Strawson, “correspondence to the facts” is a mere idiom and
so not to be taken at face value – that is, as implying a relation (of corre-
spondence) between statements and facts. Indeed, according to Strawson,
the requirement that there be things in the world to which true statements
“correspond” is misguided, and “facts” are nothing more than pseudo-
entities invented to satisfy this wrong-headed requirement.
I concur with Austin’s twofold response to Strawson’s objections.
Regarding the idea that “corresponds to the facts” is a misleading idiom,
I would agree with Austin when he says:
This seems to me quite implausible  – why should not we be meaning by
it that there is some sort of relation between something and something?
(1979: 159–160)
Regarding the idea that facts are ad hoc pseudo-entities invented by phi-
losophers, I would again agree with Austin when he says:
[T]his is to treat a wholesome English expression as though it were a phi-
losopher’s invented expression; to treat ‘facts’ as though they were in the
same position as ‘propositions’. (1979: 159)
And, since we are talking about a “wholesome English expression,” the
definition of “fact” provided by Wikipedia (vs. by a professional philoso-
pher) is particularly instructive. There, it is reported that:
19

Exploring Austin’s Galaxy 19
A fact is something that has really occurred or is actually the case. https://
en.wikipedia.org/wiki/Fact
Not surprisingly, the ordinary everyday sense of “fact” comports with the
etymology of the expression. The Online Etymology Dictionary reports the
word’s etymology as follows:
1530s “action, anything done,” especially “evil deed,” from Latin factum “an
event, occurrence, deed, achievement,” in Medieval Latin also “state, con-
dition, circumstance,” literally “thing done”. www.etymonline.com/index.
php?term=fact
Thus, the Austinian view of facts as “things in the world” comports with
the current usage of “fact,” as well as with the word’s etymology.
In response to Strawson’s claim that facts are pseudo-entities invented
by philosophers, Austin rejoins: “How come that English has invented so
unhappy an expression?” (1979: 159). He then offers, only to reject, several
responses that Strawson might provide. One of these responses, discussed
in section (c) of this chapter, is that “corresponds to the facts” is idiomatic
for “is true.” But what might Austin say in response to his own question?
No doubt, something to the effect that ordinary speakers conceptualize
the notion of truth, rightly or wrongly, as involving a correspondence or
“fitness” between statements and “the facts” – just as Wikipedia suggests.
No wonder, then, that we have “invented” the happy expression “corre-
spondence to the facts.” It “fits” precisely our ordinary everyday conception
of truth.

3.2 Searle (1998a): An Empty Metaphor

According to Searle, the “correspondence metaphor” is (i) of no theoretical
value and is (ii) completely empty. In support of point (i), he claims that
“the metaphor of correspondence is certainly no clearer than the notion
of truth itself, and so the correspondence theory should not be seen as
an attempt to reduce complex and obscure notions to simpler and clearer
ones” (Searle 1998a: 382). In support of point (ii), he claims that “cor-
respondence is a useful metaphor because it is so empty.” As he explains,
The fact that we can use so many other non-synonymous expressions – “fit,”
“state,” “describe,” and even “square with” – to do the same job, should be
a clue that “correspondence” is not being used literally. (Searle 1998a: 394)
Let’s consider these points in turn, beginning with the first. The “corre-
spondence metaphor” arguably has considerable theoretical value, as the
notion of correspondence is indeed clearer than that of truth. Pace Searle,
0

20 Marga Reimer
the notion of truth is not so much “complex and obscure” as it is simple
yet unanalyzed. “Correspondence,” when used in connection with the
ordinary everyday assessment of statements, might not be used literally.
However, its nonliteral meaning is surely grounded in its literal meaning,
which is certainly clear and simple. Thus, maps correspond, literally and
to varying degrees, to the terrain they are maps of. Similarly, clothes fit,
literally and to varying degrees, the individuals who don them. Although
the “transferred” uses of the expressions in question might not be as clear
and simple as their literal counterparts, such is the nature of metaphor
and, I would add, such is the nature of the phenomenon (truth) that the
“correspondence metaphor” seeks to capture. I  would also suggest, pace
Searle, that the metaphor is so popular because it is so intuitive and it is so
intuitive because the notion of (literal) correspondence is one that we are
all familiar with. Although arguably used nonliterally, “correspondence to
the facts” is grounded in a familiar and concrete phenomenon.
The “correspondence metaphor,” as Searle calls it, also has explanatory
value insofar as it can make sense of the frequent use we all make of the
myriad of expressions in Austin’s “galaxy” when assessing ordinary every-
day statements. All of the terms in that galaxy, including “correspondence”
and “fitness,” are terms appropriate to relational word/world phenomena
that exist on spectra.
What now of Searle’s second point:  that correspondence is a useful
metaphor because it is so empty? It certainly sounds odd to say that an
expression is useful because it is empty. After all, empty expressions lack
(semantic) content. Of what use could they possibly be? Perhaps Searle is
trying to say that the metaphor is “empty” in the sense that it implies noth-
ing, or at least nothing that is nontrivial. But we have already seen that this
is a mistake. Moreover, how is this putatively “empty” metaphor useful,
according to Searle? What “job” is it alleged to perform? Presumably, the
job performed is affirmation  – or rather reaffirmation of (the truth of )
some statement, as in:
(1) “Phoenix is the capital of Arizona” corresponds to the facts.
But then wouldn’t Searle have to say that the “correspondence metaphor”
is useless because it’s redundant, as (1) says no more than the more concise
and more direct (2)?
(2) Phoenix is the capital of Arizona.
The emptiness of the correspondence metaphor is evident, according
to Searle, from the fact that it is interchangeable with a host of other,
21

Exploring Austin’s Galaxy 21
non-synonymous expressions. However, I would counter that some of the
putatively non-synonymous expressions enumerated by Searle (“fits the
facts,” “squares with the facts”) are informative insofar as they suggest that
truth is a relational phenomenon that admits of degrees – and is therefore
a spectrum phenomenon. Thus, truth would appear to involve word/world
relations that are variable with respect to the degree to which the former
corresponds to (or “fits”) the latter. So, although the expressions in ques-
tion are not, on their literal interpretation, synonymous, they are close
enough in metaphorical meaning to be interchangeable (salva veritate).
Thus, if a statement “corresponds to the facts,” it also “fits,” “squares with,”
and “jibes with” those facts. Again, such expressions are anything but
“empty” as they suggest relational phenomena that exist on spectra: that
(in other words) admit of degrees.

3.3 Neale (2001): Idiomatic for “Is True” or a Philosopher’s Invention?

According to Neale,
It is no more illuminating to be told that a sentence is true if and only if it
corresponds to the world than to be told that a sentence is true if and only if
it is true, states a truth, says the world is as the world is, or fits the facts. For
the last of these phrases, perfectly ordinary as it is – unlike the philosopher’s
invention ‘corresponds to a fact’ – seems to be an idiomatic form of ‘is true’.
(Neale 2001: 63)
There are two issues to address here: one concerning the (semantic) con-
tent of certain putatively “idiomatic” phrases, the other concerning the
idea that “corresponds to a fact” is a “philosopher’s invention.” Let’s con-
sider these in turn.

3.3.1 Idiomatic?
My interest is in the two phrases “corresponds to the world” and “fits the
facts.” Both are arguably “idioms” in the ordinary sense that they are com-
mon turns of phrase – in the sense of “idiom” invoked by Austin when he
denies that “fitting the facts” is an isolated “idiom” in our language. But
neither phrase is an idiom in the technical (and traditional) sense that their
respective meanings are not compositionally determined. Neither phrase
is what Austin refers to as a “fused idiom” (1979: 159). Thus (pace Neale),
neither phrase is logically equivalent to “is true.” Indeed, both phrases –
when prefaced with “A sentence is true if and only if it . . .” suggest that
truth involves a relation between words and the world. Moreover, given the
appropriate sort of preface, both phrases suggest that truth – like (literal)
2

22 Marga Reimer
corresponding and (literal) fitting – is a spectrum phenomenon. The idea
that the expressions in question lack compositionally determined mean-
ings is incredibly counterintuitive. Real idioms, so-called fused idioms,
have meanings that have to be learned individually, rather than “figured
out.” Thus, while no one has trouble figuring out what it means to say that
a sentence “fits” or “corresponds to” the facts, one has to learn what it
means to say that it’s “raining cats and dogs” or that you are “the apple of
my eye.” That’s why idiom dictionaries are so popular among those seeking
to learn a new language.

3.3.2 A Philosopher’s Invention?

To claim that the phrase “corresponds to a fact” is a “philosopher’s inven-
tion” is frankly misleading. Perhaps it is true that only a philosopher would
say something like:
(3) A sentence is true if and only if it corresponds to a fact.
Yet such a statement is wrongly characterized as the “invention” of a “phi-
losopher” insofar as it derives its intuitive force, which is considerable,
from the ordinary everyday idea that truth involves “correspondence to
the facts.” And, although this latter notion is indeed an ordinary one – just
as is “fitting the facts” – and was perhaps “invented” by ordinary folk, it
captures a very important notion: that of a “match” between one’s beliefs
(or words) and the facts (reality, the world, the environment). So con-
ceived truth is important, if not essential, for our material survival and
flourishing. Thus, “corresponds to the facts” is no more “idiomatic” for
“is true” than is “fits the facts.” Both expressions, however commonplace,
ought to be taken at face value: as suggesting that truth involves a rela-
tion between words (or beliefs) and the world, a relation whereby the for-
mer corresponds, more or less, to “the facts” – i.e., reality, the world, the
environment, etc.

3.4 A Conventional Metaphor?

So if “corresponds to the facts,” like “fits the facts,” is neither a mislead-
ing idiom, nor an empty metaphor, nor idiomatic for “is true,” what then
is it? Perhaps such expressions are best understood as what linguists call
“conventional metaphors”: that is, metaphors commonly used in a given
culture’s ordinary, everyday language to give structure to some aspects of
that culture’s conceptual scheme. Here, the relevant aspect of the con-
ceptual scheme is, of course, that which concerns truth. Perhaps Austin’s
23

Exploring Austin’s Galaxy 23
hesitation to say that correspondence and fitting, when used to character-
ize truth, involve “transferred” uses of those notions stems from the fact
that these are arguably conventional metaphors. They are so common, they
are so ingrained in our thinking about statements and facts, that they don’t
sound metaphorical; rather they sound literal. But a little reflection suggests
that they are not literal; clothes fit (more or less) those who don them in a
sense that is patently literal; the same cannot be said of statements “fitting”
the facts. Maps correspond (more or less) to the terrain they map in a sense
that is patently literal; the same cannot be said of statements “correspond-
ing to” the facts. So, although conventional (that is, commonly used), the
expressions in question are nonetheless nonliteral; they are metaphorical.

4 Truth as a “Spectral” Phenomenon

The idea that truth is correspondence to the facts, when coupled with the
idea that correspondence admits of degrees, leads naturally to the view that
truth itself is a degreed or (in other words) “more or less” kind of phenom-
enon. To get an intuitive sense of this view, let’s return to the quote that
opened this chapter:
Is it true or false that Belfast is north of London? That the galaxy is the shape
of a fried egg? That Beethoven was a drunkard? That Wellington won the
battle of Waterloo? There are various degrees and dimensions of success in
making statements: the statements fit the facts always more or less loosely,
in different ways on different occasions for different intents and purposes.
(Austin 1950: 124; 1979: 130)
This passage is in response to the idea that “is true” is logically superfluous –
a claim for which I  will offer qualified support in section 7. However,
I  certainly agree with the claim that immediately precedes the forego-
ing passage: that there are cases “where it is pointless to insist on decid-
ing in simple terms whether a statement is ‘true or false.’ ” Austin then
nicely illustrates this point by way of four questions. Indeed, with respect
to these particular questions, my inclination would be to agree (with
Austin) that a simple “true or false” answer would be “pointless” (if not
misleading). With respect to each of Austin’s questions, I would probably
respond (respectively) with something like:
(4) That’s roughly true, but Belfast is really more northwest of London.
(5) That’s a colorful way of putting it, but it’s kind of true, I suppose.
(6) I guess that’s more or less true; but I wouldn’t put it that way. I’d say
Beethoven drank to excess.
4

24 Marga Reimer
(7) That’s partially true; Wellington actually won the battle along with
some Prussian general.
The general idea is clear enough:  truth, as ordinarily understood, is a
“degreed” phenomenon: statements and beliefs vary in the degree to which
they are true. Indeed, I would go further and suggest that truth value is a
degreed phenomenon:  statements and beliefs exist on a spectrum from
absolutely true to absolutely false. Logical and mathematical truths would
be on the far left; logical contradictions would be on the far right. Ordinary
truths and falsehoods would lie somewhere between these two extremes.
This sort of approach is not inconsistent with Austin’s own words (quoted
immediately above):
There are various degrees and dimensions of success in making statements: the
statements fit the facts always more or less loosely, in different ways on
different occasions for different intents and purposes. (Austin 1950:  124;
1979: 130)
I would extend this insightful point by saying:
There are various degrees of success or failure in making statements:  the
statements fit or fail to fit the facts and this success or failure is always a
matter of degree.
Thus, just as a size 7 engagement ring might fit perfectly an “average”
(adult female) ring finger, a pair of pants, sized for a toddler, might fail
to fit – and to a dramatic degree – a grown man with a forty-inch waist.
These are analogies for factual (vs. logical or metaphysical) statements that
are patently if contingently true on the one hand (“Southern Arizona gets
very hot in the summer.”) and patently if contingently false on the other
(“Billionaire businessman Donald Trump is known for his politically cor-
rect views”).
As to why we might have a tendency to think that (truth evaluable)
statements and beliefs are either true or false simpliciter, there are any
number of possible explanations. First, most ordinary everyday statements
and beliefs are, in fact, either clearly true or clearly false. Second, we are
taught by our parents and later (if indirectly) by our teachers with their
T/F quizzes and exams that statements are either true or false simpliciter.
There’s no “in between.” And for philosophers, there may be a couple of
additional reasons: we are perhaps misled by subjects like mathematics or
logic, or by fields like analytic philosophy where the most (philosophi-
cally) interesting statements are often logically/metaphysically true or log-
ically/metaphysically false. But this just means that such statements are
25

Exploring Austin’s Galaxy 25
at the far left of the spectrum or at the far right of the spectrum: they are
“absolutely” true or “absolutely” false. Also, philosophers have tradition-
ally thought of (and perhaps justified) their field as involving a “search
for Truth.” Such sought-after Truth (the object of certainty) is thought
to be, in some important sense, beyond all doubt – much in the spirit of
Descartes’ enormously influential Meditations on First Philosophy. So con-
ceived, truth does not admit of degrees – the notion of something’s being
“more or less true” is oxymoronic.

5 Correspondence Theories as “Nonstarters”

for the Austinian?
Thus far, I have been presenting Austin as an advocate of the correspon-
dence theory of truth. Is it possible that I am mistaken about this central
point? It might seem so. Charles Travis has recently claimed that corre-
spondence theories are “nonstarters” for “the picture Austin tries to paint”
(Travis 2013: 45). So what is going on here? Let’s have a look at what Travis
has to say on behalf of his intriguing (if not startling) view.
According to Travis,
it is essential to the picture Austin tries to paint that correspondence theo-
ries of truth be non-starters. (2013: 45)
Travis’ main argument relies on a standard and, as I think Austin would
agree, outdated conception of the correspondence theory of truth:
as such theories conceive things, there are two distinct autonomous
domains, each of whose denizens are what they are independent of any
such considerations as what it would be fair, or just to say, or what would
mislead, or what might be a better or worse description of the facts. Truth
is then merely a matter of whether, for any item in the first domain, there is
an item in the second which is a match. (Travis 2013: 45)
Given such a construal, the conclusion is as inevitable as it is obvious:
There is no room here for evaluations to depend on Austinian consider-
ations. (Travis 2013: 45)
Here, we can think of the “Austinian considerations” to which Travis refers
as Travis himself characterizes them: “any such considerations as what it
would be fair, or just to say, or what would mislead, or what might be a
better or worse description of the facts.” In short, attending to such factors
involves attending to the particular circumstances in which the statement
is made.
6

26 Marga Reimer
However, the importance of these circumstances is easily accommo-
dated by Austin’s spectrum model of truth, on which statements can be
“more or less” true (or false, I  would add) depending on the particular
circumstances in which the statements are made. Statements are then
appropriately regarded as true, more or less, depending on the degree to
which statements and facts “match,” which will in turn depend on the said
circumstances.
Let’s consider more closely the idea that truth is a relational notion/
phenomenon – one that involves (using Travis’ own words) a “match” of
sorts between the “denizens” of two domains. Is this an Austinian view?
Absolutely!1 As we just saw, the Austinian “denizens” – statement and fact –
of the domains in question (words and world) – do indeed “match” or
“correspond” in cases of true statements, but importantly, the match is a
“more or less” kind of thing and the degree to which there is a match will
depend on the very Austinian considerations to which Travis refers.
Moreover, there is an abundance of textual support for the proposed inter-
pretation of Austin, on which he can be seen as a correspondence theorist who
revolutionizes the theory insofar as he undercuts the nature of the “match”
between true statements and facts (or better, circumstances or events).
Consider first Austin’s claim that the coherentist and pragmatist con-
ceptions of truth fail to:
appreciate the trite but central point that truth is a matter of the relation
between words and world. (1950: 124, n. 24; 1979: 130, n. 1) (emphasis added)
Thus, for Austin, truth is a matter of there being a “match” between words
(statements) and world (“the facts”) but again, this matching relation is a
matter of degree – one’s words can match, more or less, with the world,
with “the facts.” There is considerable (additional) textual evidence for this
view, both in “Truth” and Austin’s follow-up paper “Unfair to Facts.” We
have already seen some (but not all) of this evidence. Let’s consider these
two seminal works in turn, beginning with the former.

5.1 “Truth”
In “Truth,” Austin asks:
When is a statement true? The temptation is to answer ‘When it corre-
sponds to the facts’. And as a piece of standard English this can hardly be

1
It is doubtful that these domains are, for Austin, “autonomous,” given that statements are pat-
ent worldly phenomena and can affect the circumstances in which other statements are made and
assessed.
27

Exploring Austin’s Galaxy 27
wrong. Indeed, I must confess, I do not really think it is wrong at all: the
theory of truth is a series of truisms. Still, it can at least be misleading.
(1950: 115; 1979: 121)

How can it be misleading, according to Austin? Part of the trouble has to

do with the notion of “correspondence.”
‘Corresponds’ . . . gives us trouble, because it is commonly given too
restricted or too colorful a meaning, or one which in this context [the
“problem of truth”] it cannot bear. (1950: 118; 1979: 124)

If “corresponds” requires a unique and perfect fit between statement and

fact, its meaning is arguably too restricted. As Austin explains, some of the
mistakes in correspondence theories are the result of thinking that:
for every true statement there exists ‘one’ and its own precisely correspond-
ing fact – for every cap the head it fits. (1950: 117; 1979: 123)
If “corresponds” involves “reading back into the world the features of lan-
guage,” it is perhaps too “colorful,” insofar as it proposes a mirroring of
the world by our words, something of which Austin is highly skeptical
(1950: 119; 1979: 125–126).
In either case, whether too restrictive or too colorful, “corresponds”
has taken on a meaning inappropriate to the context of the philosophical
“problem of truth.” The only appropriate meaning to that context is one in
which correspondence (and thus truth) is a degreed phenomenon.
The relevant notion of correspondence, the one relevant to the philo-
sophical “problem of truth,” involves neither a perfect fit between state-
ments and facts nor a “mirroring” of the world by statements. Rather the
relevant notion of correspondence construes that notion (and thus the
notion of truth itself ) as a degreed phenomenon.

5.2 “Unfair to Facts”

In “Unfair to Facts,” Austin advocates a conception of truth that is
patently relational. There, as we have seen, Austin considers the view
that:
‘Corresponds with the facts’ is a ‘fused’ idiom precisely equivalent to ‘is
true’. It is in no way to be understood as meaning that anything ‘corre-
sponds’ with anything in the way that that word ordinarily means, or as
implying that there are such things as ‘facts’, or that anything is related in
any way to anything. (Austin 1979: 159) (emphasis added)
8

28 Marga Reimer
To which he replies:
This seems to me quite implausible – why should not we be meaning by
it that there is some sort of relation between something and something
(no doubt not so simple as it sounds). (Austin 1979:  159–160) (second
emphasis added)
The relation is not simple because it’s not dichotomous (word/world match
or no such match); it’s spectral (more or less such a match). Truth thus
exists on a spectrum for Austin. Austin attributes the implausible view that
“corresponds” is not relational to Strawson:
Strawson actually allows that when I say a map corresponds with the topog-
raphy of the countryside I  am talking about a relation between something
and something, yet still contends that I am not when I say a statement corre-
sponds with the facts. But how is this plausible? (Austin 1979: 160) (second
emphasis added)
It is thus clear that, contra Travis, Austin views truth as involving a rela-
tion between words and the world – albeit not a simple relation of the sort
envisioned by more “traditional” advocates of correspondence theories
(like Russell and Moore) who view truth as a dichotomous (vs. spectral)
notion.

6 Relevance to Traditional Theories of Truth

The idea that truth is a spectrum phenomenon is consistent not only
with the correspondence theory of truth (as advocated by Austin), but
also with other traditional theories of truth, the coherence and pragmatic
theories in particular. It is not, however, consistent with the deflationary
theory of truth. Let’s begin by having a (very) quick look at these four
theories.
According to the correspondence theory, truth consists in correspondence
to the facts. According to the coherence theory, truth consists in coherence
with a system of beliefs/propositions. According to the pragmatic theory,
truth consists in usefulness, fruitfulness, or utility of some sort. According
to the deflationary theory, there is nothing in which truth “consists”  –
hence the label “deflationary” theory. On any such theory, to assert that a
statement (or sentence) is true just is to assert the statement (or sentence)
itself. Thus, to assert that “Tucson is in Arizona” is true is (logically) equiv-
alent to asserting that Tucson is in Arizona.
Although Austin is an advocate of the correspondence theory of truth,
he concedes that the “more or less” nature of truth is consistent with other
29

Exploring Austin’s Galaxy 29
theories as well. In a footnote to the quote that opens the present chapter
and is cited again in the preceding section, he writes:
Here there is much sense in “coherence” (and pragmatist) theories of truth
despite their failure to appreciate the trite but central point that truth is a
matter of the relation between words and world. (Austin 1950: 124, n. 24;
1979: 130, n. 1)
The point is that coherence and usefulness are degreed phenomena, just
like (ordinary and Austinian) correspondence. Thus, Austin’s observations
about the “more or less” nature of truth are compatible with the idea that
truth consists in a kind of coherence, or in a kind of utility, for coherence
and utility are qualities that admit of degree. However, those same obser-
vations are not compatible with the deflationist’s idea that truth has no
nature – for there is nothing in which truth “consists” – and which might
therefore be said to admit of degree.

7 The Logical Status of “Is True”

As noted in section 4, the passage that opens the present chapter is in
response to the (deflationary) idea that “is true” is (as Austin puts it)
“logically superfluous.” Let’s look at what Austin says in opposition to
this idea. He points out that there are other expressions in the same class
(as “true”), expressions concerned with word/world relations, and that
are clearly not logically superfluous: e.g., “vague,” “bald,” “rough,” “mis-
leading,” “not very good,” “rather general,” or “too concise.” Where such
characterizations are in order, it is pointless to insist on deciding, in sim-
ple terms, whether the statement is “true or false” (Austin 1950: 123–124;
1979: 129–130).
Austin then illustrates this idea with the series of questions (just dis-
cussed) that open the present chapter:
Is it true or false that Belfast is north of London? That the galaxy is the shape
of a fried egg? That Beethoven was a drunkard? That Wellington won the
battle of Waterloo? (1950: 124; 1979: 130)
I think there is a clear (if philosophically uninteresting) sense in which
Austin is right: “is true” is not logically superfluous in all conversational set-
tings. However, I also think that there are settings – some conversational –
in which “is true” is indeed logically superfluous. Yet in these latter settings,
despite its logical vacuity, “is true” is pragmatically quite potent. Let’s con-
sider these three points in turn, beginning with the first.
0

30 Marga Reimer
Imagine that two philosophers of language come upon a blog devoted
to the life and times of Ludwig van Beethoven. The blog contains the fol-
lowing rather insulting remark:
(8) Beethoven was a drunkard.
In response, one philosopher says:
(9) The statement that Beethoven was a drunkard is misleading  – back
then lots of Europeans would consume an entire bottle of wine with
dinner.
Alternatively, she might say (more concisely if less precisely):
(10) That statement is misleading  – back then lots of Europeans would
consume an entire bottle of wine with dinner.
The other philosopher counters that, on the contrary:
(11) The statement that Beethoven was a drunkard is true; it is well known
that he was a raging alcoholic.
Alternatively, she might say (more concisely if less precisely):
(12) That statement is true; it is well known that Beethoven was a raging
alcoholic.
Thus, consistently with what Austin says in the two passages cited earlier in
this chapter, “is true” is not (always) logically superfluous; there are indeed
conversational settings in which “is true” is just as logically potent as (e.g.)
the predicate phrases “is misleading”  – or even “is false.” Such settings
would include those in which the topic of conversation is the accuracy of a
particular sentence or statement. None of this appears to be controversial –
nor, for that reason, particularly interesting.
However, consider a statement like:
(13) It is true that Beethoven was a drunkard.
Is the phrase “It is true that . . . ” logically superfluous? In other words, is
(13) logically equivalent to (8)? I would say yes, but would add that the
phrase is pragmatically potent; it might, in other words, have pragmatic
(vs. logical or semantic) implications. In this respect, (13) might be like
(14)–(18):
(14) Beethoven was in fact a drunkard.
(15) Beethoven was indeed a drunkard.
31

Exploring Austin’s Galaxy 31
(16) Actually, Beethoven was a drunkard.
(17) Beethoven really was a drunkard.
(18) Beethoven was a drunkard.
Utterances of any of (14)–(18) would pragmatically implicate that not
everyone would agree with (8). The implication would presumably be
effected by way of one or more Gricean conversational maxims, perhaps
quantity or manner, as the speaker is saying more than is necessary (quan-
tity violation) and/or is being somehow redundant (manner violation).
Why not simply say, “Beethoven was a drunkard”? The extra words (or
italics) – so-called intensifiers – conversationally implicate that the speak-
er’s claim that Beethoven was a drunkard has been (or might be) ques-
tioned. Thus, such remarks might naturally elicit responses like:
(19) Are there actually people who deny that Beethoven was a drunkard?
So, although there may be some settings in which “is true” is not logically
superfluous, even in settings where it is, it may be pragmatically potent.
The question is:  what does all this mean for Austin’s views on truth? It
certainly doesn’t establish that the deflationary theory of truth is superior
to the correspondence theory. I would suggest it has little if any relevance
for Austin’s advocacy of his particular version of the correspondence theory
of truth.

8 Through the Lens of Ordinary Language

In addressing what he called the “problem of truth,” Austin suggested that
we explore ordinary, everyday expressions in a particular “galaxy” of expres-
sions: a galaxy centered around the notions of fitting, corresponding, and
measuring. Such exploration, he thought, would be far more fruitful than
obsessing over the single word “truth,” for truth is:
a bare minimum or an illusory ideal (the truth, the whole truth and noth-
ing but the truth about the battle of Waterloo or the Primavera). (Austin
1950: 125; 1979: 130)
Austin also presumably thought that such exploration would lend credence
to the correspondence “theory” of truth. I put scare quotes around “the-
ory” so as to reflect Austin’s own views about the nature of such a theory:
When is a statement true? The temptation is to answer . . . : ‘When it cor-
responds to the facts’. And as a piece of standard English this can hardly be
wrong. Indeed, I must confess I do not really think it is wrong at all: the
theory of truth is a series of truisms. (Austin 1950: 115; 1979: 121)
2

32 Marga Reimer
One might protest that a theory of truth, even if only a “series of truisms,”
is not sufficiently supported by mere “conceptual analysis”: by (e.g.) the
“exploration” and “elucidation” of the expressions in the Austinian “galaxy.”
But any such criticism would be misguided. By focusing on concepts
other than that of truth, by focusing on concepts in Austin’s galaxy (“cor-
responds to,” “fits,” “exact,” “precise,” “rough,” “loose,” etc.), we arrive
at the idea that truth is but one of a host of concepts that are not only
relational (word/world), but that are also spectral. Although perhaps ini-
tially counterintuitive (especially for those trained in analytic philosophy
or mathematics), the idea that truth is a “more or less” concept comports
well with how we actually assess truth-evaluable statements in ordinary,
everyday contexts.
But in what sense is this spectral notion of truth anything more than
“spectral” – anything more than a philosopher’s phantasm? What does it
have to do with the truth about . . . well, Truth? In answering this question,
perhaps we should reflect on what Austin says in the opening paragraph
of “Truth”:
‘What is truth’ said jesting Pilate, and would not stay for an answer. Pilate
was in advance of his time. For truth itself is an abstract noun, a camel, that
is, of a logical construction which cannot get past the eye of even a gram-
marian. We . . . ask ourselves whether Truth is a substance . . . or a quality . . .
or a relation. . . . But philosophers should take something more nearly their
own size to strain at. What needs discussing is rather the use, or certain uses,
of the word ‘true’. (1950: 111; 1979: 117)
I am largely sympathetic to this way of thinking about ordinary, everyday
truth  – vs. that “camel of a logical construction,” Truth. Perhaps I  can
explain my sympathy by returning to the title of the present chapter, which
makes reference to the “lens” of ordinary language, a lens with which one
might explore such language in the search for truth. I would like to sug-
gest that this lens is a reflective one, but that it is turned inward toward the
viewer. The viewer thus sees nothing that is “outside” because she can’t;
she is effectively looking into a mirror. In exploring Austin’s galaxy of the
fitting/measuring notions used in the assessment of statements, we are
exploring how we conceptualize truth. But it’s not as though there is any-
thing more “substantive” or “objective” beyond this conceptualization  –
something “out there” that the reflective lens of ordinary language prevents
us from seeing. Truth is arguably exhausted by how we, as ordinary think-
ers, conceptualize it, but it is not for that reason any less interesting than
it would otherwise be. Indeed, to appreciate how we, as ordinary speak-
ers, conceptualize truth is to understand something very important about
33

Exploring Austin’s Galaxy 33
ourselves: it is to understand how we view the relationship between what
we say or believe about the world and the world itself. Might we be wrong
about this relationship? Might truth be other than we conceptualize it to
be? Because truth is not “out there” in the world, existing independently
of us (like the sun, the moon, and the stars), I don’t see how the answer
to this question could possibly be other than unambiguously negative.
Of course, those statements that we ultimately, and perhaps consensually,
deem “true” might be part of an interconnected system of statements, or
they might be those that are (in some sense) useful. Indeed, there might be
no “facts” of the sort that we naively suppose there to be – pace Wikipedia.
But so what? We still learn much about ourselves once we appreciate how
we understand – or misunderstand – the nature of the relation between
our words and the phenomenal world in which we live, both of which
are uncontestably real – in the only sense of “real” that really matters: the
ordinary, everyday sense.
4

Ch apter  2

Levels of Linguistic Acts and the Semantics of

Saying and Quoting
Friederike Moltmann

1 Introduction
In How to Do Things with Words, Austin (1975) introduced not only the
notion of an illocutionary act, such as an act of asserting, requesting,
promising, or asking a question, but also the notion of a locutionary act,
which consists in various acts ‘below’ the level of an illocutionary act.
A locutionary act includes what Austin calls a ‘rhetic act’, an act char-
acterized, roughly, as the act of uttering the words in a sentence with a
specific meaning and reference. A locutionary act also includes a ‘phatic
act’, an act of uttering words, and a phonetic act, an act of producing
sounds.
This chapter will outline a novel semantics of verbs of saying and of
quotation based on Austin’s distinctions among levels of linguistic acts.
Austin’s notion of a rhetic act is not a very clear one and tends to be consid-
ered problematic and insufficiently motivated. This chapter will propose a
particular way of understanding the notion of a rhetic act and argue that
it is extremely well reflected in the semantics of natural language, in par-
ticular in the semantics of verbs of saying and of certain sorts of plural NPs
in English (and German). This chapter will furthermore outline a novel
semantics of quotation, making crucial use of Austin’s distinctions among
lower-level linguistic acts, a semantics that promises a unified and com-
positional semantics of quotation. Two ideas guide that semantics. First,
quotations convey properties related to lower-level linguistic acts, Austin’s
phonetic or phatic acts; second, such meanings of quotations are strictly
based on syntactic structure, namely, a lower-level linguistic structure as
part of the syntactic structure of the sentence that is input to interpreta-
tion. Such lower-level linguistic structures may be phonetic, phonological,
or morpho-syntactic structures. They will be interpreted, roughly, as prop-
erties of utterance tokens and as such contribute, in one way or another, to
the semantic composition of the entire sentence.
34
35

On the Semantics of Saying and Quoting 35

This semantics of verbs of saying and of quotation is cast within a more
general act-related semantic theory of sentences and of attitude reports.
While this theory ties in with recent act-based theories of sentence mean-
ing ( Jubien 2001; Moltmann 2003; Soames 2010; Hanks 2015) in focusing
on acts rather than abstract propositions, it differs fundamentally from
those theories, in ways that will be crucial for the semantics of verbs of say-
ing and of quotation. Most importantly, unlike Soames (2010) and Hanks
(2015), it does not consider sentences as standing for cognitive proposi-
tions, that is, types of cognitive acts (acts of predication). Rather it takes
sentences to semantically act as predicates of various sorts of objects. These
include entities such as claims, promises, requests, decisions, thoughts,
and ‘words’, which are considered ‘products’ of illocutionary, cognitive,
and locutionary acts in the sense of a product of Twardowski (1912, see also
Moltmann 2013, 2014, 2017b). As such they are bearers of truth or satisfac-
tion conditions yet are particulars and dependent on the act that estab-
lished them. In addition to such products, sentences may act as predicates
of mental states (intentions, beliefs, hopes, etc.) and of modal objects such
as needs and permissions (Moltmann 2015b, 2017b).
This view of the meaning of sentences has various conceptual and empir-
ical motivations, especially in regard to the semantics of attitude reports
(Moltmann 2014, 2017a), modal sentences (Moltmann 2015b, 2017b), and
truth predicates (Moltmann 2015a). The semantics of verbs of saying and
of quotation is another important application of the view, promising a
novel, unified, and compositional semantic analysis.
An attitude report such as (1a) on that view has the logical form in
(1b), where claim takes as its implicit Davidsonian event argument an act
of claiming and product is a function mapping that act onto its product,
John’s claim:
(1) a. John claims that Mary is happy.
b. ∃e(claim(e, John) & [that Mary is happy](product(e)))
The semantic value of the that-clause, [that Mary is happy], is a property
of attitudinal objects, specifying their satisfaction conditions. Crucially,
speech act reports such as (2a) with a quotational complement will have
the very same logical form, as in (2b):
(2) a. John says ‘Mary is happy.’
b. ∃e(say(e, John) & [‘Mary is happy’](product(e)))
However, the event argument of say is just a locutionary, not an illocution-
ary act, and the semantic value of the quotational complement, [‘Mary is
6

36 Friederike Moltmann
happy’], is a property of locutionary products, specifying, roughly, their
form and semantic composition.
The form-related property that the direct quote in (2a) conveys is based
strictly on linguistic structure, understood roughly in terms of Generative
Syntax. The idea is that quotation involves a lower-level linguistic structure
(a phonetic, phonological, or morpho-syntactic structure) as part of the
syntactic structure of the sentence that is input to semantic interpreta-
tion (LF). I will suggest that this can be understood in terms of a three-
dimensional conception of syntactic structure, which permits lower-level
linguistic structures to be represented on a separate plane and thus escape
ordinary rules of syntax.
Direct quotes will have an ordinary syntactic structure as well as such
a lower-level linguistic structure, and the interpretations of the two struc-
tures together will serve to characterize locutionary products in terms of
a phonetic/phatic and a rhetic component. Mixed quotations involve a
similar complex syntactic structure and dual semantics. Pure quotations,
by contrast, will have only a lower-level linguistic structure as part of the
overall syntactic structure of the sentence.
Unlike in standard formal semantics, the act-related conception of
meaning permits an interpretation of a lower-level linguistic structure
as part of the syntactic structure that is input to semantic interpretation,
namely, as a property of utterances (phonetic or phatic products). This
enables a compositional semantics of quotation based strictly on linguistic
structure.
In the following, I  will first outline the attitudinal objects theory of
attitude reports with its main linguistic motivations, that is, the theory on
which clausal complements act as predicates of attitudinal objects (illocu-
tionary or cognitive products or mental states). I then extend the theory
to verbs of saying (including think) based on the notion of a rhetic act.
I  propose a way of understanding the notion of a rhetic act and show
that it is explicitly reflected in natural language. Finally, I outline the act-
based semantics of quotation with a very brief sketch of a novel, three-
dimensional conception of their syntax.

2 The Background: Sentences as Predicates

of Attitudinal Objects
The semantics of verbs of saying and of quotation this chapter will develop
takes as its background the view that sentences do not stand for abstract or
cognitive propositions, but rather act as predicates of a variety of objects.
37

On the Semantics of Saying and Quoting 37

In the case of independent sentences, these are entities such as assertions,
promises, questions, and demands, that is, illocutionary products. In the
case of sentences embedded under attitude verbs, these will be various sorts
of attitudinal objects. Attitudinal objects include illocutionary products as
well as cognitive products, entities such as thoughts, judgments, and deci-
sions. In addition, attitudinal objects include mental states such as beliefs,
hopes, and intentions.
Attitudinal objects and kinds of them are extremely well reflected in
natural language, namely, in the most common sorts of nominalizations
of attitude verbs, assertion, request, promise, thought, decision, belief, inten-
tion, etc. Whereas John’s assertion and John’s intention stand for particular
attitudinal objects, the assertion that S and the intention to do X stand for
kinds of attitudinal objects.
It is common to consider nouns like assertion, promise, and belief to
be polysemous, standing either for an act or for a proposition (as, for
example, in Searle 1968). However, there are good grounds for consider-
ing the entities they stand for as sui generis entities – attitudinal objects
(Ulrich 1976; Moltmann 2003, 2013, 2014, 2017b). Attitudinal objects
are distinguished from acts and propositions in three main respects,
well reflected in the applicability of the relevant predicates to products
but not to the corresponding acts or, in general, to the corresponding
propositions.
First, attitudinal objects have truth or satisfaction conditions, reflected
in the applicability of various sorts of predicates of satisfaction, which
would generally not apply to acts or propositions:
(3) a. John’s claim is true.
b. ??? John’s act of claiming is true.
(4) a. John’s command was satisfied / complied with.
b. ??? John’s speech act was satisfied / complied with.
c. ??? A proposition was satisfied / complied with.
(5) a. John fulfilled / broke his promise.
b. ??? John fulfilled / broke his act of promising.
c. ??? John fulfilled / broke a proposition.
Here ‘???’ (and the weaker ‘??’ and ‘?’) means ‘semantically unacceptable’ or
‘cannot possibly be true’.
Second, attitudinal objects (of the same type) enter the relation of exact
similarity (expressed by is the same as) just in case they are the same in
content, whereas other things need to be the case for acts to be the same.
8

38 Friederike Moltmann
Thus, John’s claim is ‘the same as’ Mary’s just in case the two claims are the
same in content, whereas for John’s act of claiming (or John’s speech act)
to be the same as Mary’s more needs to be the case than that. The same
holds for commands versus acts of commanding and promises versus acts
of promising.
Third, attitudinal objects have a part structure based on partial content
and not the temporal part structure of acts or states. This is reflected in
the way part of is understood with products:  part of the assertion/promise/
request/command can pick out only partial content, not a temporal part of
an act.
Beliefs, intentions, and hopes are not products of acts, but mental states.
Yet they share the characteristic properties of products, as do modal objects
such as obligations, permissions, and needs, the entities playing the same
sort of roles in modal sentences (Moltmann 2015b, 2017b).
A special class of attitudinal objects is illocutionary and cognitive prod-
ucts. They are to be understood as the (nonenduring) products of illocu-
tionary or mental acts, in roughly the sense of Twardowski (1912) (see also
Moltmann 2013, 2014, 2017b). In general, products but not actions are the
bearers of representational or normative properties. Just like attitudinal
objects in general, but unlike acts, products have truth or satisfaction con-
ditions, enter similarity relations based on shared content, and have a part
structure based on partial content, rather than the temporal part structure
of acts. Yet products have properties of concrete objects as well, such as
having a limited life span and the ability to enter causal and perceptual
relations, and illocutionary products come with a physical realization (an
utterance). Thus claims, the products of acts of claiming, can be made and
overheard, and make people upset.
The action-product distinction also applies to phonetic and phatic acts
as well as rhetic acts, which will be relevant for the semantics of verbs say-
ing and of quotation.
For the semantics of attitude reports, I make use of Davidsonian event
semantics (Davidson 1967). This means that cognitive or illocutionary acts
are the implicit arguments of (nonstative) attitude verbs. Cognitive and
illocutionary products then are the products of those Davidsonian event
arguments, and as such can be obtained by a function product applied to
acts. Sentences embedded under an attitude verb will act as predicates of
the product of the event argument:

(6) a. John claimed that S.

b. ∃e(claim(e, John) & [that S](product(e)))
39

On the Semantics of Saying and Quoting 39

Mental states will themselves be Davidsonian arguments of mental state
verbs, and the product function will then simply map a mental state onto
itself.
The semantics of attitude reports in (6b) is (almost) overtly reflected
in the corresponding complex-predicate construction in (7), involving
explicit reference to an attitudinal object (or to a kind of attitudinal object):
(7) John made the claim that S.
In fact, languages tend to show an alternation between the simple and the
complex-predicate construction, which further motivates the semantics
in (6b).
Not all attitude verbs have their clausal complement characterize the
product of their event argument. Clausal complements of some attitude
verbs instead serve as predicates of a contextually given attitudinal object,
for example, a claim relevant in the discourse context. This is the case with
response-stance verbs such as repeat, confirm, agree, and remind.1 Thus (8a)
has the analysis in (8b) for a contextually given claim or hypothesis d:
(8) a. John confirmed that S.
b. ∃e(confirm(e, John, d) & [that S](d))
This difference in the semantics of the two sorts of attitude verbs is sup-
ported by the observation that response-stance verbs but not volunteered-
stance verbs generally permit substitution of the clausal complement by
an ordinary NP (John confirmed the hypothesis that S vs.??? John claimed the
proposition that S). Moreover, response-stance verbs permit a reading of
adverbs like partially quantifying over parts of the content of the contextu-
ally given attitudinal object, a reading unavailable for volunteered-stance
verbs (Moltmann 2017a):
(9) a. John partly confirmed that Mary is incompetent.
b. ??? John partly claims that Mary is incompetent.
The same sort of difference will become relevant again for the semantics of
verbs of saying.
The involvement of attitudinal objects in the semantics of attitude
reports is also reflected in the semantic behavior of quantifiers and pro-
nouns that can take the place of clausal complements. Such ‘special’ or
‘nominalizing’ quantifiers include something, everything, that, and what

1
The notion of a ‘response-stance verb’ is due to Cattell (1978), who distinguishes it from that of a
‘volunteered-stance’ verb such as claim, request, think, etc.
0

40 Friederike Moltmann
(Moltmann 2003, 2013). The relevant observation is that predicates acting
as restrictors of such quantifiers generally are understood as predicates of
attitudinal objects rather than of propositions:
(10) a. John asserted something shocking.
b. John dreamt something nice.
c. John demanded something impossible to comply with.
Shocking, nice, and impossible to comply with express properties that can be
attributed to attitudinal objects (assertions, dreams, and demands), but
not abstract propositions (which cannot be ‘shocking’, ‘nice’, or ‘impos-
sible to comply with’). That is, they are properties that go with the rel-
evant nominalizations (John’s assertion, John’s dream, or John’s demand). This
motivates the analysis of special quantifiers as nominalizing quantifiers, as
below for (10a) (Moltmann 2003, 2013, 2017b):
(10) d. ∃e′∃e (assert(e, John) & shocking(e′) & e′ = product(e))
Here the quantifier ‘∃e’ is introduced by something and ‘∃e′’ goes with the
Davidsonian event semantics of the verb. On this analysis, special quantifi-
ers do not provide arguments of the predicate, but, as nominalizing devices,
introduce a new domain of entities on the basis of the meaning of the verb.
Furthermore, reports of sharing as below involve a special pronoun, the
relative pronoun what:
(11) a. John asserted what Mary asserted.
(11a) reports the sharing of a kind of attitudinal object, ‘the assertion that
S’, rather than of a particular attitudinal object such as ‘John’s assertion
that S’, as in the analysis below (Moltmann 2003, 2013, 2017b):
(11) b. ∃e′∃e′′∃e(assert(e, John) & e′  =  product-kind(e) & assert(e′′,
Mary) & e′ = product-kind(e′′))
Such reports of sharing will also be relevant for the semantics of verbs of
saying.
There are constraints on reports on sharing discussed in Moltmann
(2003, 2013, 2017b), which support the analysis in (11b). Roughly, the con-
straint is that the two attitude verbs need to involve the same force, but
may differ in certain other respects such as strength:
(12) a. ??? John promised what Mary asserted, that he will come back.
b. ??? John asserted what Mary demanded, that he will be back in
an hour.
41

On the Semantics of Saying and Quoting 41

(13) a. John suggested what Mary asserted.
b. John requested what Mary demanded.

When the two verbs are not identical, a decomposition of their mean-
ing takes place into a more general attitude and a modifier, so that the
shared object will be a kind of attitudinal object of the more general sort
(Moltmann 2003, 2013).
If sentences act as predicates of attitudinal objects, the question is of
course: what sort of property do they express? In their role as predicates of
attitudinal objects, sentences need not specify a structured content since
various sorts of attitudes need not involve any structure, such as implicit
belief or knowledge (and neither do modal objects such as needs). Rather
sentences can just specify satisfaction conditions of attitudinal objects, of a
sufficiently fine-grained sort. Satisfaction conditions may be taken to con-
sist in a set of worlds in which the attitudinal object is satisfied. However,
using situations instead will give a more adequate, fine-grained notion of
content. For that purpose, I will adopt notions from Fine’s (2017) recent
truthmaker semantics, which takes the content of a truth bearer to be
bipartite, consisting of a set of exact truthmakers and a set of exact false-
makers. The role of sentences specifying the satisfaction conditions of atti-
tudinal objects can then be captured by assigning sentences the following
meaning (Moltmann 2015b, 2017b):
(14) For a sentence S, [S] =  λd[∀i(i ╟ d → S is true in d) & ∀i(i ╢ d
→ S is false in d)]
Here ╟ is the relation of exact truthmaking that holds between a situ-
ation s and an attitudinal object d in case s is wholly relevant for the
satisfaction of d.  ╢ is the relation of exact falsemaking, which holds
between a situation s and an attitudinal object d if s is wholly relevant
for the violation of d. Thus a situation of Mary’s happiness is a satisfier
of both John’s claim that Mary is happy and John’s desire for Mary to be
happy, and a situation of Mary’s being unhappy a violator of both that
claim and that desire.2
One may object that this still does not give a sufficiently fine-grained
notion of content. Verbs of saying and of occurrent thought may involve

2
(14) treats declaratives and imperatives alike. However, imposing constraints on the properties
and types of satisfiers, say differentiating between situations and actions, may allow distinguishing
between the illocutionary act types associated with declarative and with imperative sentences, as well
as, perhaps, between that-clauses and infinitival complements.
2

42 Friederike Moltmann
complements that specify also a choice of words or concepts and their
order, as in the sentences below:
(15) a. John literally said that he was really incapable of doing the job.
b. John was thinking that he was really incapable of doing the job.
But this is precisely where the notion of a locutionary act comes into play.
That-clauses may have an additional role and that is to characterize a verbal
or mental act in terms of the choice and combination of concepts or words
(with an intended meaning or referent), that is, in terms of a structure that
results from a rhetic act. In other words, that-clauses will have two differ-
ent meanings: [1] a property of attitudinal objects specifying their satisfac-
tion conditions and [2] a property of the structured content associated
with rhetic acts. The satisfaction-related meaning of that-clauses applies
to attitudinal objects, whereas the structure-related meaning applies to the
product of rhetic acts (which may be part of attitudinal objects, namely,
illocutionary or cognitive products).
The distinction between actions and products also applies to rhetic
and phatic acts, in tune with Twardowski (1912).3 The product of a rhetic
act, but not the act itself, carries intended meaning-related and structure-
related features (a notion discussed in greater detail in the next section).
The product of a phatic act (a ‘token’), but not the act itself, carries only
relevant form-related features (say phonological or morpho-syntactic
features).4

3 Locutionary Acts and the Semantics of Verbs of Saying

3.1 Simple Verbs of Saying

We will now see that the same semantics in (14) carries over to verbs of
saying, though they involve locutionary, not illocutionary acts. The focus
will be on ‘simple’ verbs of saying, and only a few remarks about complex
verbs of saying will be made later.
What I call ‘simple verbs of saying’ includes not only the verb say, but
also write and think. There is good evidence that the semantics of these
verbs involves locutionary acts, as parts of illocutionary acts, rather than

3
For phatic acts, this has been explicitly pursued by Twardowski’s student Ingarden (1931).
4
The notion of a product of a phatic act in fact matches the familiar notion of a token (as opposed to
an utterance act). A token has only relevant properties, properties of the linguistic structure the act
is meant to realize, such as phonological, morphological, or syntactic features.
43

On the Semantics of Saying and Quoting 43

full illocutionary acts. First of all, reports of saying such as (16) leave open
what sort of illocutionary act was performed, say an assertion, a threat, or
a promise:5
(16) John said / wrote that he will leave.
Say, write, and think may also take direct quotes as complements, of all
three sentence types, which one might take to indicate that those verbs
involve illocutionary rather than locutionary acts in their semantics:6
(17) a. John said / wrote / thought ‘I will leave.’
b. John said / wrote / thought ‘Leave!’
c. John said / wrote / thought ‘What should I do?’.
The direct quotes are complements of the verb (Munro 1982): they fill in
an obligatory argument position taking the very same place as could be
taken by a special quantifier or pronoun, as in (18) or in the pseudocleft
construction in (19) (Grimshaw 2015):
(18) a. John said / wrote / thought something, namely ‘Leave!’.
b. What John said / wrote / thought was ‘Leave!’.
(19) a. What John said / wrote / thought was ‘I should leave!’.
b. What John said / wrote / thought was ‘What should I do?’.
In fact, direct quotes of the three sentence types may satisfy the semantic
selectional requirements of illocutionary verbs such as tell, demand, and
ask (Grimshaw, ms). As such, they alternate with that-clauses (assertions),
infinitival clauses (imperatives), and interrogative clauses:
(20) a. John told Mary that he will come / ‘I will come.’
b. John demanded to be given one more day / ‘Give me one more
day!’.
c. John asked where the exit was / ‘Where is the exit?’.
However, there are good arguments to the effect that say, think, and write
do not involve illocutionary acts, but just locutionary acts – and thus that

5
Searle (1968) disputes the existence of such neutral occurrences of embedded sentences, but see
Green (2015).
6
All three verbs also allow for quotations with parentheticals and quotational inversion, constructions
not available with non-quotational clausal complements:
(i) a. ‘I will leave,’ John said / wrote / thought.
b. ‘I will leave,’ said / wrote / thought John.
4

44 Friederike Moltmann
direct quotes having the form of declaratives, imperatives, or interrogatives
may just reflect locutionary and not illocutionary acts.
First, say, write, and think do not take interrogative or infinitival com-
plements for the purpose of specifying a question or a request as the con-
tent of the attitude described by the embedding verb:
(21) a. * John said /wrote / thought what he should do.
b. * John said / wrote / thought for Bill to leave.
Second, reports of sharing with say, write, or think and illocutionary verbs
are impossible. Thus (22a) is impossible as a report of sharing relating to
(22b) and (22c), and the same goes for (23)–(25):
(22) a. ??? John asserted what Mary said.
b. John asserted that Bill won the race.
c. Mary said that Bill won the race.
(23) a. ??? John said what Mary demanded.
b. John demanded that Bill should leave.
c. Mary said that Bill should leave.
(24) a. ??? John said what Mary asked
b. John said ‘Did Bill win?’.
c. Mary asked ‘Did Bill win?’.
(25) a. ??? John promised what he said.
b. John promised that he would help Mary.
c. John said that he would help Mary.
As discussed earlier, reports of sharing are about sharing kinds of products,
and say and write just cannot share kinds of illocutionary products, but
only kinds of locutionary products.
Also, reports of sharing with think and decide are impossible:
(26) a. ??? John thought what Bill decided.
b. Bill decided that they should leave the house / ‘Let’s leave the
house!’.
c. John thought that they should leave the house / ‘Let’s leave the
house!’.
Decisions are cognitive products on a par with illocutionary products such
as promises or demands, by carrying satisfaction conditions with a world–
word/mind direction of fit, to use Searle’s (1969, 1983) term.
This means that the clausal complements of say, write, and think,
including those that are direct quotes, serve to characterize locutionary,
45

On the Semantics of Saying and Quoting 45

not illocutionary products.7 The Davidsonian event argument of verbs of
saying will thus be just a locutionary act, though that act may be part of
an illocutionary act.8 The overall semantics of locutionary act reports will
then be the same as that for attitude reports:
(27) a. John said that S.
b. ∃e(say(e, John) & [that S]loc(product(e)))
That S here does not have the meaning given in (14), but rather a secondary
locutionary meaning, a property of rhetic products.

3.2 The Nature of Rhetic Acts

Clausal complements of verbs of saying, I have argued, serve to character-
ize products of locutionary, not illocutionary acts (or their mental counter-
parts). For Austin, locutionary acts consist of rhetic, phatic, and phonetic
acts. Rhetic acts are meaning-related acts below the level of illocutionary
acts. They are characterized as acts of using words with a specific meaning
or reference.9 As stated, this would not really be a single act involving the
use of a sentence, but a plurality of acts involving the words or relevant
constituents of the sentence.10 Clearly, though, a rhetic act could not in
fact be a mere plurality of acts of using the words in the sentence. Rather
it should be a coordinated or structured plurality, namely, of acts of using
expressions with particular meanings and with semantically relevant rela-
tions that will lead to the composition of the meaning of the sentence.
Let us take the meaning of a sentence to be the property of illocutionary
products given in (14). Then a rhetic act will in fact be a plurality of acts
of conveying semantic values of subsentential expressions as entering rela-
tions leading to the composition of the meaning of the entire sentence as
the property in (14). The product of such a plurality of acts will in turn be
a plurality of products, namely, products of meaning-related acts involving

7
According to Austin (1975), indirect quotes, that is, that-clause complements of verbs of saying,
characterize rhetic acts, whereas direct quotes characterize phatic acts (though Austin is not always
consistent in what he takes indirect quotes to characterize, cf. Searle 1968).
8
Note that as with belief, verbs of saying may describe states rather than acts, as given below
(Grimshaw 2015):
(i) The sign says, ‘It is forbidden to drive here.’
In that case, the product function applies to a state, as a carrier of representational properties, and
maps it onto itself.
9
Austin actually gave various not entirely consistent characterizations of the notion of a rhetic act.
I will just focus on one of them. See Searle (1968) for further discussion.
10
This was noted by Searle (1968).
6

46 Friederike Moltmann
subsentential occurrences of expressions in a particular meaningful config-
uration. Natural language in fact reflects products of rhetic acts as plurali-
ties, not as single entities, as we will see in the next section.
How can that-clauses express types of rhetic acts? A  suggestion is that
such an interpretation results when that-clauses are not fully interpreted, but
instead the process of the semantic composition of the sentence is ‘frozen’ at
a given point – that point depending, of course, on the speaker’s intentions.
This means that the composition of the products of a rhetic act may very
well lead to an illocutionary product. Force-neutrality pertains to rhetic acts
only insofar as they are not fully composed illocutionary products.
On this view, rhetic acts will be fundamentally different from acts of
conveying a proposition, which Searle (1968) had proposed should take the
place of rhetic acts. Rhetic acts are well reflected in natural language, as we
will see in the next section; propositional acts hardly are.

3.3 Explicit Reference to Products of Rhetic Acts

Rhetic acts are explicitly reflected in one particular kind of expression in
English, namely, plural NPs such as a few words, those words, and the words
‘I forgive you’ – ‘words-NPs’ for short. Words-NPs go with locutionary rather
than illocutionary verbs, and what they stand for shows the characteristic
properties of products of rhetic acts.11
In addition to that-clauses and direct quotes, the locutionary verbs say
and write also take such NPs as complements:
(28) John said a few words / those words / the words ‘I forgive you.’
Here words is used not as the plural of singular word, but as a plurale tan-
tum, standing not for pluralities of individual expressions, but for plurali-
ties of words with specific meanings in a meaningful configuration; that is,
it stands for structured pluralities of meaningful elements.12
In some languages, the equivalent of words in that use differs mor-
phologically from the ordinary plural of word. In German, Woerter is the

11
Some observations about words-NPs have been made in chapter 4 of Moltmann (2013), where they
have been taken as an indication for clausal complements of attitude verbs having the status of plu-
ral terms (for ordered pluralities) and as such as support for the Neo-Russellian Multiple Relations
Analysis pursued in that chapter – mistakenly, as I now think.
12
Note that words on that use is still semantically plural, accepting numeral adjectives and plural
quantifiers such as a few and supporting plural anaphora:
(i) a. those three words, ‘I love you’
b. He said a few words, and then he said them again.
47

On the Semantics of Saying and Quoting 47

ordinary plural of word and Worte the plural referring to words used with
a specific meaning in a particular meaningful configuration (Moltmann
2013: 155).13 Words when standing for structured pluralities of meaningful
elements occurs in fact also in the title of Austin’s (1975) book, How to Do
Things with Words, where words would translate in German as Worte (liter-
ally ‘Wie man Dinge mit Worten macht’), not Woerter (‘Wie man Dinge
mit Woertern macht’).
Words-NPs are force-neutral and can relate not only to a declarative sen-
tence, but also to an interrogative and an imperative one, as below:
(29) ‘Leave the house!’ / ‘When should I leave?’, John himself has said
precisely those words.
Force-neutrality is of course a mark of products of rhetic acts.
Related to force-neutrality is the observation that words-NPs do not go
well with predicates of truth or satisfaction:
(30) a. ??? Mary’s words, the words ‘The world will end tomorrow,’ are not
true.
b. ??? The words ‘Finish the paper by midnight!’ cannot be complied
with.
c. ??? John fulfilled / broke the words ‘I will help you!’.
This matches the observation that thought, which stands for products of cog-
nitive rhetic acts, is not really good with truth predicates either – in ordinary
speakers’ intuitions:
(30) c. ?? John’s thought that the world will end tomorrow is not true.
The products of rhetic acts are not bearers of satisfaction conditions, at least
not in a non-derivative fashion.
Words-NPs when referring to the content of the acts described by the
embedding verbs are possible only as complements of locutionary verbs such
as say and write, not illocutionary verbs (Grimshaw 2015). Thus the relevant
reading is unavailable below:

13
Worte rather than Woerter is used also for naming the Ten Commandments. The Ten Commandments
translates either as die Zehn Gebote, literally ‘the ten commandments’, or (less commonly) as die
Zehn Worte, literally ‘the ten words’. In the latter case, Worte is a superplural, standing for a second-
level ordered plurality that consists of ten ordered pluralities of meaningful elements. Instead of
die Zehn Worte, also das Zehnwort is used for the Ten Commandments, where -wort appears as a
mass noun. As a mass noun, it still stands for the same thing as Worte, for an ordered second-level
plurality of ordered pluralities of meaningful elements, just as Wort in das Wort Gottes, ‘the word
of God’.
8

48 Friederike Moltmann
(31) a. ??? John asserted a few words.
b. ??? John promised the words ‘I am always ready to help.’
c. ??? John demanded / asked a few words.
The reason is that illocutionary verbs involve full illocutionary acts, not just
locutionary or rhetic acts, and words-NPs can stand only for products of
rhetic acts and not for products of illocutionary acts.14
Rather than characterizing the product of the act described by a locutionary
verb, words-NPs may also provide the argument of other sorts of predicates:
(32) a. John repeated / understood / interpreted / read those words.
b. John believed / remembered / feared those words.
Here the words-NPs refer to the product of a contextually salient locution-
ary act; they thus have the same sort of function as that-clause comple-
ments of response-stance verb, standing for an entity that is independent
of the event argument of the predicate.15
Not all locutionary verbs take words-NPs. Think does not and say to
oneself does not really either:
(33) ??? John thought those words.
Words-NPs can stand only for physically realized products, not just for
products of cognitive acts.16
Words-NPs thus are an overt reflection in natural language of Austin’s
notion of a rhetic act (referring to the product of such an act). Natural
language, by contrast, hardly displays (nontechnical) terms for abstract
propositions, or, for that matter, for Searle’s (1968) notion of a proposi-
tional act as an abstraction from an illocutionary act.
Illocutionary verbs generally do not take words-NPs. However, some of
them, for example, demand and ask, take direct quotes:
(34) a. ??? John demanded / asked a few words.
b. John demanded ‘Help her!’
c. John asked ‘When did you help her?’
14
Note that words-NPs may stand for products of rhetic acts while acting as nominalizing quantifiers,
rather than providing arguments of the predicate.
15
With those verbs they allow for partially, unlike with say:
(i) a. ??? John partially said those words.
b. John partially repeated / understood / interpreted / read those words.
16
This is also apparent from the way the more neutral verb add is understood. While add can describe
a mental act in (ia), this is not possible in (ib), where it can only describe a discourse-related act:
(i) a. John added a few more expectations to his general hope.
b. ??? John added a few words to his thinking.
49

On the Semantics of Saying and Quoting 49

This indicates that direct quotes can serve to characterize the locutionary
part of an illocutionary act, whereas words-NPs can only stand for a rhetic
product on its own.
This raises a general issue regarding the ontological status of locution-
ary acts. In what sense are locutionary acts parts of illocutionary acts?
Can locutionary acts be performed without performing illocutionary act?
For Austin, the linguistic acts of the various levels are ordered by the by-
relation or what Goldman (1970) calls the relation of ‘level-generation’.
This is a form of composition of acts, but one in which lower-level acts
could have been performed without performing the higher-level acts.
Thus, if killing the king was done by pulling the trigger, the killing has
the trigger-pulling as a non-temporal part and the pulling of the trigger
could have been done without thereby killing the king. Certainly there
are locutionary acts that are performed without performing illocutionary
acts, say utterances for the purpose of grammatical exercise and entertain-
ing thoughts for mere consideration. There is also an alternative view,
though, according to which a locutionary act is an abstraction from an
illocutionary act (Searle 1968). In that case, a locutionary act would be a
part ontologically dependent on the whole and could not be performed
independently.
Abstraction explains best the possibility in natural language of reference
to illocutionary acts in disregard of the locutionary acts by which they are
performed. An example is reports of sharing with illocutionary acts that
would involve very different physical realizations (phatic acts). Thus, (35a,
b, c) can be true at once:
(35) a. John asserted the same thing as Mary (that Bill won the race).
b. John whispered that Bill won the race.
c. Mary screamed that Bill won the race.
This even holds for reports of sharing with locutionary acts. Thus, (35d)
may be true, while (35b, c) are also true:
(35) d. John said the same thing as Mary.
Let us call the kinds of products reported as shared in (35a) and (35d) ‘thin
illocutionary products’ and ‘thin rhetic products’, respectively. Roughly,
as an abstraction from a ‘full’ illocutionary product, a ‘thin’ illocutionary
product will have only those properties relating to its conditions of satis-
faction and force (or direction of fit), but not properties regarding its phys-
ical realization or choice of names or concepts. A similar condition would
individuate a ‘thin’ rhetic product.
0

50 Friederike Moltmann
Also, when using nominalizations, we seem to have a notion of an
illocutionary product that is at least not dependent on the phatic act by
which it is produced. Thus ‘the very same assertion’ could have been made
in English by using a softer voice or by way of writing. Note that inde-
pendently of speech acts, we have a notion of an act that need not have
the lower-level act that generates it as an essential part. The killing of the
king could have been done by throwing a bomb instead of by pulling the
trigger.

3.4 Complex Verbs of Saying

Besides simple verbs of saying, there are semantically complex verbs of say-
ing. I will just make a few remarks about one such type of verb, manner
of speaking verbs. Manner of speaking verbs behave just like simple verbs
of saying, permitting in general that-clauses, direct quotes of the three
sentence types, but not interrogatives or infinitival clauses representing the
content of what would be an imperative:
(36) a. John whispered / screamed / muttered ‘I will come’ / that he will
come.
b. John whispered / screamed / muttered ‘Who did that?’ / * who did
that.
c. John whispered / screamed / muttered ‘Come!’ / * for Bill to
come.
Moreover, they may take NP-complements, including words-NPs:
(37) John whispered / screamed / muttered something /a few words.
Like simple verbs of saying, verbs of manner of speaking don’t permit shar-
ing with illocutionary verbs:
(38) a. ??? John whispered the same thing that Mary asserted / demanded
/ asked.
b. ??? John screamed the same thing that Mary asserted / demanded /
asked.
This would indicate that the manner of speaking verbs involve locutionary
products just like simple verbs of saying. However, reports of sharing with
a manner of speaking verb and a simple verb of saying are impossible:
(39) a. ??? John said what Mary whispered.
b. ??? John said what he screamed.
51

On the Semantics of Saying and Quoting 51

By contrast, reports of sharing are possible with different manner of speak-
ing verbs:
(40) a. ?(?) John screamed what Mary whispered.
b. ? John shouted what Bill yelled.
This indicates that manner of speaking verbs involve both a rhetic and a
phatic act, whereas simple verbs of saying involve only a rhetic act. The
possibility of reports of sharing with different manner of speaking verbs
moreover suggests that the manner of speaking verbs are underlyingly
composites of a simple verb say and a manner of speaking modifier charac-
terizing the phatic act (Grimshaw 2015).
To summarize, rhetic and locutionary acts play a central role in the
semantics of simple and complex verbs of saying, which in general do not
involve full illocutionary acts in their semantics.

4 The Semantics of Quotation

4.1 Quotation in General

Direct quotes, as described so far, serve to characterize both the product
of a rhetic act and the product of a phatic act, where the rhetic and the
phatic act make up the locutionary act as (part of ) the Davidsonian event
argument of the embedding verb. The purpose of this section is to give
a rough outline of a more general semantic theory of quotation accord-
ing to which quotations convey properties of phatic products. This theory
makes crucial use of Austin’s distinction among levels of linguistic acts and
a novel syntactic view of quotation, which permits treating quotation in a
compositional and strictly semantic fashion. In the context of this chapter,
this theory can be presented only in roughest outline with just a few cen-
tral applications. A full formal development and a more elaborate discus-
sion of the existing literature on quotation has to await another occasion.
Quotation presents major challenges for standard formal semantics.
One such challenge is to deal with quotation within a compositional
semantics since the phenomenon of quotation appears to militate against a
compositional semantic treatment (Pagin and Westerståhl 2010). Another
challenge is to develop a unified account of pure (41a), direct (41b), and
mixed (41c) quotation:17

17
For more on pure and mixed quotation, see Maier (2014a, 2014b).
2

52 Friederike Moltmann
(41) a. John said ‘shhh’.
b. John said ‘I will come.’
c. John said that he ‘resides’ in Paris.
Pure quotation differs from direct quotation in that it is not generally
meaning-conveying and does not make up a sentential complement of an
illocutionary or locutionary verb.
The standard view is that pure quotations are expression-referring
terms, managing, in some way, to refer to the relevant expression type, by
acting as descriptions (Geach 1972) or as names (Reichenbach 1947), or
involving a demonstrative (quotation marks) pointing to a displayed token
(Davidson 1969, 1979; Clark and Gerrig 1990; Cappelen and Lepore 2007;
de Vries 2008), or else by ‘presenting’ it (Washington 1992; Saka 1998).18
Direct quotations on the standard view require a different treatment since
they contribute both a content (a proposition) and a form and thus cannot
just act as expression-referring terms.19 Mixed quotations also contribute a
content and a form, though the latter may either characterize the reported
speech act as in (1c) or some contextually given speech act. The standard
view tends to consider both direct and mixed quotation pragmatic phe-
nomena quite distinct from pure quotation. The standard view generally
admits that quotation of various sorts cannot be treated compositionally
and be based on the interpretation of a formal syntactic structure.
Standard formal semantics has difficulties with quotation because of
one of its fundamental assumptions, namely, that the meanings of sen-
tences are abstract propositions, a view that has recently been challenged
by philosophers such as Jubien (2001), Moltmann (2003, 2013), Soames
(2010), and Hanks (2015), who argue in favor of an act-based notion of
sentence meaning. A central aim of this chapter is to show that quotation
provides an important application of an act-related conception of sentence
meaning according to which sentences function as predicates of various
sorts of objects, including products of illocutionary, locutionary, or cogni-
tive acts. For the treatment of quotation (and of verbs of saying), this has
the crucial advantage that sentences may express both content-related and
form-related properties, to be predicated of products of locutionary and
illocutionary acts.20
18
The view that pure quotations are referential terms can also be found in Recanati (2000).
19
A unified treatment has been pursued though in the Davidsonian tradition by Cappelen and Lepore
(2007).
20
The theory of quotation of Ginzburg and Cooper (2014) can also be considered an application of an
act-related view of meaning to quotation. It shares similarities with the present approach to quota-
tion, for example, by making use of ‘locutionary propositions’ for direct quotation. However, its
53

On the Semantics of Saying and Quoting 53

4.2 Pure and Direct Quotation

The central idea regarding the semantics of quotation is that expressions
can be interpreted not just by assigning them their usual semantic value,
but also as properties of products of lower-level linguistic acts. This shift
in interpretation, moreover, is not arbitrary or ‘pragmatic’, but rather has a
strict syntactic basis. It is based on a lower-level linguistic structure being
part of the syntactic structure that is input to interpretation. I will sketch
this theory first for pure and direct quotation and then indicate how it
extends to mixed quotation.
By way of illustrating the idea, it suffices to first stay with the verb say.
Say takes a pure quotation as complement in (41a), repeated below:
(42) a. John said ‘shhh’.
Pure quotations as in (42a) function as complements since they fill in an
obligatory position and can be replaced by special quantifiers like some-
thing, allowing the inference from (42a) to (42b):
(42) b. John said something.
As with that-clauses and direct quotations, pure quotations as comple-
ments may serve to characterize the product of the described event, though
this must be a phatic or phonetic act, not a rhetic act. In (42a), for exam-
ple, the pure quotation serves to characterize the product of a phonetic act.
A pure quotation as complement of say may also characterize a phatic
act. This even holds for sentential quotations, which may in fact be pure
quotations rather than direct quotations. The difference between senten-
tial direct quotations and pure quotations is particularly clear in German.
A sentence as a pure quotation must appear in NP-position, in the middle
field, whereas as a direct quotation, it must follow the verb:21
(43) a. weil Hans endlich ‘Ich liebe dich’ sagen kann
because John finally ‘I love you’ say can
‘because John can finally say ‘I love you’’
b. weil Hans endlich sagen kann ‘Ich liebe dich’
because John finally say can ‘I love you’
‘because John can finally say ‘I love you’’

empirical motivations and theoretical framework are rather different, and the space of this chapter
does not permit a more detailed discussion.
21
For similar syntactic observations about Dutch, see de Vries (2008).
4

54 Friederike Moltmann
Whereas (43a) can report only a linguistic ability (the ability to pronounce
a particular sentence, say), (43b) can report the readiness to express an
emotional state (or the ability to admit to one).
The semantics of (42a) will then be as below, with the pure quotation
predicated of the product of the phonetic act that is the Davidsonian argu-
ment of the verb:
(42) c. ∃e(say(e, John) & [shhh](product(e)))
The question now is, how are pure quotations able to express properties of
products of phonetic or phatic acts? The proposal is that this is so because
pure quotations involve a lower-level linguistic structure as part of the
syntactic structure of the sentence that is input to interpretation (L(ogical)
F(orm)). More precisely, a pure quotation may involve several lower-level
linguistic structures as part of LF, a syntactic and a phonological structure,
say. For the purpose of this chapter, I will just give some suggestions of
how this idea may be spelled out syntactically. A  full syntactic develop-
ment will need to await another occasion.
First, a few general remarks about the syntax of pure quotation are
needed. Pure quotations can obtain categorial specifications as NPs and fill
in positions requiring an NP, as in (42a), but they may also occur in posi-
tions in which no complements may appear, such as in close appositions as
in (44b) and following verbs that take no complements as in (44b):
(44) a. the word ‘hello’
b. John went ‘Hey, hey, hey’.
I will assume that pure quotations form quotational phrases (QPs). The
syntactic structure of (42a) will then roughly be as below:
(42) d. John [said [[[shhh]]QP]NP]VP
Following Giorgi (2016), one may assume that the head Q of a quotational
phrase reflects a quotational pause.22
The new suggestion, then, is that Q is a special category that will act as
a sort of coordinator, setting up a sort of coordinate structure involving
other syntactic planes in a three-dimensional syntactic structure (Goodall
1987; Moltmann 1992). On standard three-dimensional syntactic theo-
ries of coordination, coordination involves a three-dimensional syntactic

22
Giorgi (2016) makes use of a phrase KP for (direct) quotation where K reflects the comma into-
nation. However, close appositions do not involve a comma intonation, as opposed to ordinary
appositions (the poet, Goethe vs the poet Goethe).
55

On the Semantics of Saying and Quoting 55

structure, so that ordinary grammar applies to the various different planes
representing the different conjuncts or disjuncts. The difference with quo-
tation would be that the lower-level linguistic structures of quotations
are represented in other planes precisely in order to escape application of
rules of ordinary grammar. That is because quotations need not be gram-
matically correct, may come from other languages, and involve structures
below the relevant linguistic level of the rest of the sentence. The struc-
tures in the other planes will all be interpreted as properties of products
of phonetic or phatic acts, and their conjunction (intersection) will make
up the semantic value of the entire quotation. Thus, shhh in (42a) will
have a phonetic structure on a plane different from that of the rest of the
sentence and that structure will be interpreted as a property of products
of phonetic acts.
With this brief sketch of the syntax of pure quotation, let us turn to the
difference between pure and direct quotation. A clause that has the status
of a pure quotation has only lower-level linguistic structures, including
possibly a syntactic structure that is as such not input to semantic inter-
pretation. It thus does not have a syntactic structure on the same plane
as the LF-structure of the sentence. By contrast, a direct quote also has a
syntactic structure that is input to semantic interpretation, which allows
it to express a property of products of locutionary acts as well. That struc-
ture is the structure of a main clause, not an embedded clause, which
explains the ‘shifted’ interpretation of indexicals in direct quotes (Giorgi
2016, Grimshaw, ms):

(47) a. Johni said ‘Ii will come’.

b. * Johni said ‘Hei will come!’

Direct quotation exhibits selectivity, which means that not everything in

the quotation marks matters for characterizing the phatic act in question,
but only whatever features the speakers intends to matter. This means that
direct quotation may involve just partial lower-level linguistic structures.
As with pure quotation, the level of structure(s) that is to play a character-
izing role also depends on the speaker’s intentions (in fact more so than
with pure quotation).
Direct quotation will express a conjunction of two properties: a prop-
erty of rhetic products and a property of phatic or phonetic products. The
property of rhetic products is based on the ordinary syntactic structure
of the complement clause; the property of phatic products is based on a
lower-level linguistic structure on a different plane.
6

56 Friederike Moltmann
This account of direct and pure quotation considers quotation a seman-
tic phenomenon, based on syntactic structure. It is compositional because
of the particular way sentential meaning is conceived. The account differs
fundamentally from current approaches to quotation where the utterance
of the quotation, the token, matters for what the quotation contributes
to the meaning of the sentence, both in the tradition of Davidson (1969,
1979; Cappelen and Lepore 2007) and within the more recent identity
theory of quotation (Washington 1992; Saka 1998). On the present theory,
the semantic contribution of quotation is based on structure, and the quo-
tational structure is interpreted as a property of tokens.
The present approach does not take pure quotations to necessarily form
referential terms, unlike the standard view. Pure quotations may but need
not act as referential terms, and in fact, their primary use ought to be pred-
icative rather than referential since pure quotations express properties (of
tokens). As a matter of fact, quotations may occur as syntactic predicates,
namely in as-phrases with the verbs translate and pronounce:
(48) a. She translated red as ‘rouge’.
b. She pronounced ‘red’ as ‘rett’
As requires predicative, rather than referential complements (John as a
father, Mary treated John as a brother), and thus the pure quotations in
(48a, b) must be predicative.23 The current approaches to quotation do
not acknowledge a predicative function of quotation, whereas the pres-
ent view accommodates the predicative function of quotation in (48a, b)
straightforwardly. In (48a), the property expressed by rouge, a property of
phatic products, is predicated of ‘the translation’, the product of the act
of translating, and in (48b) the property expressed by ‘rett’, a property of
phonetic products, is predicated of the ‘pronunciation’, the product of the
act of pronouncing.
Note that as-phrases can also act as adnominal modifiers of the corre-
sponding product nominalizations:
(49) a. the translation of ‘red’ as ‘rouge’
b. the pronunciation of ‘red’ as ‘rett’

23
By contrast, the direct object position of translate and pronounce is not predicational, but referential,
allowing substitution by an explicit expression-referring term and allowing for the ‘ordinary’ (non-
nominalizing) pronoun it:
(i) a. She pronounced / translated the word ‘red’ as ‘rouge’.
b. She had never pronounced / translated it before.
57

On the Semantics of Saying and Quoting 57

Pure quotations of course have also syntactic functions other than that
of a predicate.24
These include other non-referential roles, for example, in close apposi-
tions. Close apposition also illustrates well the various ways in which pure
quotations can act:
(50) a. the morpheme ‘un’
b. the sentence ‘it is raining’
c. the concept ‘horse’
d. the phoneme ‘a’
e. the sound ‘shhh’
The non-referential status of the quotation is indicated by the impossibil-
ity of replacing the quotation by an explicit referential term (*the word the
word ‘maison’, * the concept what ‘horse’ expresses). In close appositions, the
pure quotation is not referential on its own, rather the sortal head noun of
the close apposition structure has a reifying function, or so I have argued
(Moltmann 2013, chapter 6). That is, the sortal head noun involves as part
of its semantics a mapping of the semantic contribution of the pure quo-
tation onto a corresponding object, that is, it involves reifying the literal
interpretation of the pure quotation with its lower-level linguistic struc-
ture, for example, in (50b) a property of tokens and in (50c) a conceptual
meaning.
Some occurrences of pure quotations may be referential in the sense
of involving an implicit close-apposition structure containing an unpro-
nounced sortal noun. This is arguably the case for pure quotations in sub-
ject position, as below, where a pure quotation is replaceable by an overt
close apposition of a suitable sort:
(51) a. ‘Mary’ is disyllabic.
b. The name ‘Mary’ is disyllabic.

24
Pure quotations may occur in other predicative contexts, in particular as predicates in small clause
complements of verbs of calling:
(i) a. John called Mary ‘Marie’.
b. * John called Mary the name ‘Marie’.
As Matushansky (2008) argues, ‘Marie’ in (ia) syntactically has predicative status, in some languages
even showing predicative marking. On the present view, pure quotations can play a predicative
semantic role, since they express properties of phatic products. However, it is less straightforward
to consider names in contexts of calling pure quotations in a predicational function. In that role,
they would not be predicated of phatic products, but rather of individuals being named (though of
course, it is easy to formulate a suitable property of individuals on the basis of a property of phatic
products).
8

58 Friederike Moltmann
By contrast, pure quotations in object position tend to be non-referential,
not permitting replacement by an overt close apposition, even if the position
permits NPs syntactically. Besides say, the verb mean is of that sort. Mean
permits instead of a pure quotation an overt close apposition in subject posi-
tion but not in object position, where only a special quantifier is permitted:
(52) a. ‘Red’ means ‘red’.
b. The word ‘red’ means ‘red’.
c. ???? ‘Red’ means the concept red.
d. ‘Red’ means something.
Let me summarize the central idea of this treatment of pure and direct
quotation. Both pure and direct quotation involve linguistic structures
below the level of LF whose interpretation consists in properties of prod-
ucts of phatic or phonetic acts (or perhaps concept conveying acts). While
this is just what those structures should stand for at the level of grammar
to which they properly belong, when part of LF, those interpretations will
serve as semantic contributions of pure and direct quotes to the composi-
tion of the overall meaning of the sentence.

4.3 Mixed Quotation

The syntactic and semantic account of pure and direct quotation that the
previous section has outlined naturally extends to mixed quotation. In
addition to its normal semantic value, a mixed quotation conveys a prop-
erty characterizing the product of a phatic act.25 In a sentence embedded
under a verb of saying, this act may be the act described by the embedding
verb, as in (41b), repeated below:
(53) a. John said that he ‘resides’ in Paris.
But the act may also be a contextually given phatic act, as below:
(53) b. John ‘resides’ in this neighborhood.
In (53a, b) the quotation may characterize part of John’s utterance, specify-
ing his choice of words in a statement of where he lives.
Like a direct quotation, a sentence containing a mixed quotation has
two meanings: its ordinary meaning, a property of attitudinal objects spec-
ifying their satisfaction conditions, and a property of products of utterance
acts (phonetic or phatic acts). The difference is that with direct quota-
tion, the utterance property is expressed by the entire embedded sentence,
25
For the idea of a ‘dual semantics’ of direct and mixed quotation, see also Potts (2007).
59

On the Semantics of Saying and Quoting 59

whereas in the case of mixed quotation, it is expressed by a subsentential
part. Moreover, with mixed quotation, the utterance property may serve
to characterize a contextually given utterance part, rather than the product
of the act described by the embedding verb of saying.
The syntactic structure involved in mixed quotation, that is, the basis
on which part of a sentence can express a property of part of an utterance,
will be similar to that of direct quotation: the quoted expression will have
an additional, lower-level linguistic structure at a different plane (or several
such structures, a phonological and a morphological structure, say). Again,
this structure may be partial or underspecified, but in any case, it will be
interpreted ‘literally’ as a property of products of phonetic or phatic acts.
Sentences with mixed quotations will have the same sort of composi-
tional semantics as direct quotations. This semantics consists in interpret-
ing an LF-structure containing an additional partial lower-level linguistic
structure, as a property partially specifying the form of an utterance (a
phonetic or phatic product). But in mixed quotation, this property may
be predicated of a contextually given utterance, and thus mixed quotation
involves a pragmatic element not present in direct quotation.

5 Conclusion and Outlook

Austin’s notion of a locutionary act and in particular that of a rhetic act
has long been considered controversial and unmotivated. This chapter has
proposed a way of understanding the notion of a rhetic act within an act-
related semantics and has argued that it plays a central role in the seman-
tics of verbs of saying and of quotation and moreover that it is overtly
reflected in natural language, for example, in English words-NPs.
In addition, by making use of Austin’s distinction among levels of lin-
guistic acts, this chapter has outlined a novel, unified, and compositional
semantics of quotation in its various forms. This semantics went along
with the syntactic view that quotational structures involve the presence of
a lower-level linguistic structure within the syntactic structure that is input
to interpretation. The chapter could only give a general outline of this view,
though, and the proposals regarding the syntax and compositional seman-
tics of quotation must await full theoretic and formal development.26

26
This chapter has greatly benefited from the audiences at the following workshops: “Quotation:
Perspectives from Philosophy and Linguistics,” Bochum, September 2012; “NYU Philosophy of
Language Workshop,” New York University, May 2012; “Workshop on Quotation,” New York
University, April 2014; “Sentences and Clausal Complements,” IHPST, Paris, September 2015. The
chapter has moreover benefited from conversations with Alec Marantz, Jane Grimshaw, Richard
Kayne, and Juergen Moltmann, and comments by Savas L. Tsohatzidis.
0

Ch apter  3

On the Representation of Form and Function

Imperative Sentences
Robert Fiengo

1 Frege’s Turnstile and Illocutionary Force Indicators

There is a tradition in linguistic research according to which a sentence
and the speech-act performed when using it are together awarded a single
representation. I believe that this practice has unfortunate consequences.
I will here argue instead for an Austinian approach: if a sentence’s struc-
ture is sharply distinguished from its use, and the representation of a
sentence’s structure makes no mention of its use, those consequences can
be avoided, and we can come to understand why certain sentences, in
virtue of their overt structures, make them better suited for some pur-
poses than for others. To illustrate the advantages of this approach, I will
examine sentences cast in the Imperative and Indicative moods and show
how their syntactic differences make them suitable for the performance
of distinct kinds of illocutionary acts. While the interactions of mood
and use won’t be fully explored, some portions of this large terrain will
become understandable, and my goal is, in any case, primarily meth-
odological. I hope to convince the reader that the relationship between
form and function has been handled badly, and that a better alternative
is available.
The practice I wish to criticize has its source in a program whose goals
were in many ways far removed from goals pursued today. The story
is worth telling, if only briefly; it begins with Frege’s invention of the
‘turnstile’. Frege (1879: 52–53) lays out the syntax of an ideal concept-
language, a language that would include all and only that which is nec-
essary for correct inference. In the syntax of this language, he wishes
to represent both thoughts and judgments. The Inhaltsstrich (content
stroke) ‘—’ may be prefixed to ‘A’ to yield ‘— A’ only if ‘A’ has what
Frege calls beurtheilbarer Inhalt (judgeable content). He paraphrases ‘—
A’ as ‘the thought that A’. The Urtheilsstrich (judgment stroke) ‘|’ may

60
61

On the Representation of Form and Function 61

then be prefixed to ‘— A’ to yield ‘|— A’: the judgment that A. As Frege
(1880/81: 11) puts it, the judgment stroke ‘converts the content of possible
judgment into a judgment’. The complex symbol ‘|—’ has come to be
called the ‘turnstile’.
With the development of the sense–reference distinction in the early
1890s, some of Frege’s earlier views are transformed, including his treat-
ment of content and judgment strokes. Frege (1893:  215)  notes that he
had previously ‘combined under the expression “judgeable content” ’ what
he has at this point ‘learnt to distinguish as truth-value and thought’.
Considering ‘Inhaltsstrich’ now to be an inappropriate term, he renames
the content stroke the ‘horizontal’. In Frege (1891: 141–2), ‘— x’ is intro-
duced as a function-expression, whose value is the True when ‘x’ is a name
of the True, and the False otherwise. As for the judgment stroke, Frege
(1891: 142) remarks:
[The] separation of the act [of expressing a judgment] from the subject-
matter of judgment seems to me indispensable; for otherwise we could not
express a mere supposition—the putting of a case without a simultaneous
judgment as to its arising or not. We thus need a special sign in order to be
able to assert [behaupten] something. To this end I make use of a vertical
stroke at the left of the horizontal, so that, e.g., by writing
|— 2 + 3 = 5
we assert [behaupten] that 2 + 3 = 5. Thus here we are not just writing down
a truth-value, as in
2 + 3 = 5,
but also at the same time saying [sagt] that it is true.1

Here Frege decisively separates the act of judging from the subject-matter
of judgment. To write down ‘|— A’ is to express the judgment that A is true.
The contrast between the horizontal and the judgment stroke now becomes
apparent. While the horizontal can be used to construct a functional expres-
sion, the judgment stroke cannot; as Frege remarks in a footnote:
The judgment stroke cannot be used to construct a functional expres-
sion; for it does not serve, in conjunction with other signs, to designate
[Bezeichnung] an object. ‘|— 2 + 3 = 5’ does not designate [bezeichnet] any-
thing, it asserts [behauptet] something.

1
This passage is loose. To write down ‘2 + 3 = 5’ is to write down not a truth-value, but a name of a
truth-value. Also, can one ‘write down an equation’ and at the same time ‘say that it is true’ without
falling into a use-mention confusion?
2

62 Robert Fiengo
While sentences designate truth-values, assertions do not designate any-
thing. ‘|— A’ represents the assertion that A is true, but it designates neither
the True nor the assertion that A is true nor anything else.
In these passages, how are we to understand ‘asserting that A’? In the
first passage, Frege says that for an agent to write down a judgment-for-
mula is to assert, while in the second, a judgment-formula itself asserts.
And how are we to understand ‘supposing that A’? Is to write down ‘— A’
to suppose that A, or does the formula ‘— A’ itself suppose that A?2 Does
‘— A’ represent the act of supposing or does it represent, as in Frege (1879),
the possible content of supposing? In Frege (1892), some clarification is pro-
vided: thoughts and truth-values, no longer confounded, are argued to be,
respectively, the senses and referents of sentences. Thoughts remain objec-
tive: Frege (1892: 62fn) says that ‘By a thought I understand not the subjec-
tive performance of thinking, but its objective content, which is capable of
being the common property of several speakers.’ But with the separation
of sense and reference, there is room for a new, more active conception of
judgment. Frege (1892: 65) now proposes that:
Judgments can be regarded as advances from a thought to a truth value.
Naturally this cannot be a definition. Judgment is something quite peculiar
and incomparable.
In a footnote he continues: ‘[a] judgment, for me, is not the mere compre-
hension of a thought, but the admission of its truth.’ Judgments, in this
view, are movements from thoughts to truth-values.
Perhaps not surprisingly, there are difficulties. First, the relationship
between what a judgment takes as its goal and what a true sentence refers
to is unsettled. Truth appears twice in this account – once as the goal of a
judgment and once as the referent of a sentence – and it is not obvious that
one and the same object can play both of these roles.3 Perhaps Frege senses
this difficulty, since having made this proposal he immediately suggests a
fairly radical alternative:
One might also say that judgments are distinctions of parts within truth
values.4 Such distinction occurs by a return to the thought. To every sense
belonging to a truth value there would correspond its own manner of analy-
sis. However, I have here used the word ‘part’ in a special sense. I have in fact

2
These two conceptions are consistent; we may commonly say either that an agent says something or
that the sentence that he says says something. Cf. Fiengo (to appear) for discussion of this alternation.
3
Cf. Furth’s introduction to Frege (1967), where a similar point is made.
4
Here Frege is addressing an alternative to the consequence pointed out in Frege (1892: 65) that, if the
reference of a sentence is a truth-value, ‘in the reference of a sentence all that is specific is eliminated.’
63

On the Representation of Form and Function 63

transferred the relation between the parts and the whole of the sentence to
its reference, by calling the reference of a word part of the reference of the
sentence, if the word itself is a part of the sentence.
Word-senses in thoughts determine parts of truth-values. Judgments are
movements from thoughts (and their parts) to truth-values (and their
parts). Though the consequences of this alternative proposal for the notion
of truth are undeveloped, it does address the problem noted, since truth –
though no longer metaphysically simple – does appear in one role only.
To recap: in 1879, Frege started from the assumption that representa-
tions of judgments contain representations of thoughts:  he represented
judgments by placing judgment strokes in front of representations of
thoughts. In the early 1890s, with the advent of the sense–reference dis-
tinction, he arrived at a view according to which judgments are activi-
ties: in consequence, ‘|— A’ came to represent a complex movement from
parts of the thought that A to parts of the True.
To illustrate, suppose, as Frege (1892) would, that the two sentences
Cicero was an orator and Tully was an orator have distinct senses but the
same referent, the True. The names ‘Cicero’ and ‘Tully’ have distinct
senses, but refer to the same part of the True, and the predicates, being the
same, do also. The judgment-formulae ‘|— Cicero was an orator’ and ‘|—
Tully was an orator’ will, consequently, represent distinct judgments, since
although these judgments have the same goal, they have distinct sources.
But none of this is explicitly represented:  an explicit representation of a
judgment would incorporate a representation of a thought, a representa-
tion of the requisite parts of the True, and a representation of movement
from the thought to them. Frege’s representations lack these things; he does
not provide explicit representations either of thoughts or of parts within
truth-values. But had he succeeded in doing these things, the judgment
stroke and the horizontal, by his own lights, would have thereby been ren-
dered superfluous.5 For if the thought expressed by Cicero is an orator could
be explicitly represented, there would be no need for the horizontal, and
if movement from this thought to the requisite parts of the True could be
explicitly represented, there would be no need for the judgment stroke.
Generally speaking, in a fully explicit representation, there is no need to
include in the representation any indication of what the representation is a
representation of, and, since an ideal concept-language should include only
what is necessary, the horizontal and the judgment stroke would have to be

5
Wittgenstein (1922, 1953) and Anscombe (1959) also hold the turnstile to be superfluous, but for
reasons quite different from those suggested here.
4

64 Robert Fiengo
eliminated. To be sure, the turnstile is necessary when thoughts and truth-
values are not distinguished and representation is inexplicit, but once they
are distinguished, and the activity of judging can be represented explic-
itly, it becomes superfluous. Frege never confused content and judgment.
Having arrived at the view that judgment is movement from thought to
truth-value, a consistent development of his thought would have elimi-
nated the turnstile.
Having never been rendered superfluous, the turnstile survives today
as an indicator not of judgment or assertion as Frege conceived of them,
but of something rather different: the illocutionary speech-act of asserting.
Consider in this connection the views of Searle (1969: 30–31):
We can distinguish two (not necessarily separate) elements in the syntactical
structure of the sentence, which we might call the propositional indicator
and the illocutionary force indicator. . . . Illocutionary force indicators in
English include at least:  word-order, stress, intonation contour, punctua-
tion, the mood of the verb, and the so-called performative verbs.
The general form of (very many kinds of ) illocutionary act is F(p) where the
variable “F ” takes illocutionary force indicating devices as values, and “p”
takes expressions for propositions [as values].
As instances of the form F(p), Searle gives:
|— (p) for assertions ? (p) for yes-no questions
! (p) for requests Pr (p) for promises
W (p) for warnings
What do these representations represent? Searle repeatedly emphasizes their
linguistic nature. Scope of negation is defined on them: he suggests that
the difference between I do not promise to come and I promise not to come
should be represented as the difference between ‘∼Pr (p)’ and ‘Pr (∼ p)’.6
The variables ‘F ’ and the lower-case ‘p’ take linguistic expressions as values,
and these linguistic expressions concatenate in the normal way. Simple
subject-predicate propositions have internal syntax: Searle (1969: 32) intro-
duces the general form F (R P): ‘ “R” for the referring expression and the
capital “P” for the predicating expression’. The clear impression is given

6
What illocutionary act is performed in saying I do not promise to come? Searle suggests that the speaker
is refusing to make a promise, but I doubt this is right. I hereby refuse to promise to come, an explicit
way of refusing to promise, has a very different force from I do not promise to come. The correct
answer, I believe, is that I do not promise to come is simply assertive, the perlocutionary effect being to
inform the addressee that the speaker might not come. On this view, it is comparable to He does not
promise to come, and it should be represented in Searle’s system as: |— (∼I promise to come).
65

On the Representation of Form and Function 65

that these representations represent propositions. Nevertheless, a page
later, Searle (1969: 33) says:
Having divided up (a large number of types of ) illocutionary acts into ele-
ments represented by the letters in the notation “F (R P)”, we can then offer
separate analyses of illocutionary force (F), referring (R), and predicating (P).
In this passage, Searle has ‘R ’ and ‘P ’ taking activities as values, not lin-
guistic expressions, and these activities are parts of illocutionary acts. He
apparently wants representations of the form ‘F(R P)’ to be linguistic rep-
resentations (‘expressions for propositions’) concatenated with illocution-
ary force indicators and to be representations of illocutionary acts. Since
no representation can be both, this creates a muddle. It makes sense to
say that a negative operator takes scope over the nodes dominated by the
node it is in construction with. It does not make sense to say that a neg-
ative operator takes scope over an activity. While the activity of referring
can be part of a speech-act, it cannot be part of a sentence, or part of a
proposition. Searle’s representations are unstable, teetering between form
and function.
Frege never confounded form and function,7 but Searle does do this,
as have many generative linguists. The goal of linguists is not to create
an ideal language for reasoning, nor to provide a philosophical account
of illocutionary acts,8 but to provide an empirical analysis of natural lan-
guage. In pursuit of this goal, linguists have appealed to illocutionary
force indicators. The ‘Q-morpheme’ of Katz and Postal (1964), the ‘per-
formative analysis’ of Ross (1970), and the Force Phrases of Rizzi (1997)
are prominent instances. These mechanisms are technically distinct from
each other  – the differences reflecting wider changes in the field across
time – but each mechanism embodies the form/function confusion I wish
to criticize. In one respect, criticizing them is difficult, however, since
the linguistics literature seldom states explicitly what linguistic repre-
sentations are representations of. I have been told in conversations over
the years that the representations linguists propose represent, or do not
represent, the structures of sentence-types, utterances, sentences in the
‘I-language’, sentences in the ‘E-language’, and events occurring in the
‘mind/brain’  – a wide range. If linguistic representations are to include
illocutionary force indicators, most of these answers are implausible. If

7
The fact that the completion of his program would have rendered the turnstile superfluous is a sure
sign that he was not confusing form with function.
8
Austin’s goal was not to provide a philosophical account of speech-acts; it was to explain how we do
things with words.
6

66 Robert Fiengo
linguistic representations are to be representations of sentence-types, they
certainly should not contain illocutionary force indicators, since sentence-
types don’t have illocutionary force:  they are used to perform activities
that have illocutionary force. Similarly, I find it difficult to imagine that
‘sentences in the I-language’, or ‘events occurring in the mind/brain’ – if
I understand these terms at all – are the kinds of things that could have
illocutionary force; it is utterances that have illocutionary force, not the
hands backstage that aid in the performance. Oddly, the idea that lin-
guistic representations represent utterances is by no means a widespread
view among generative linguists, despite the fact that utterances do have
illocutionary force.
It is fair to say that many linguists now mix form and function with-
out comment. There has been a broad retreat from the methodological
position that syntax is ‘autonomous’: that the syntax of a language should
be defined and justified without appeal to meaning or use.9 On the view
that syntax is autonomous, the objects of syntactic representations pose
no mystery: they represent the syntactic structures of sentence-types and
contain no illocutionary force indicators at all. Function finds its place
elsewhere: in an Austinian theory of the use of sentence-types.
The practice of including illocutionary force indicators in linguistic
representations forgets that, whatever sentence we choose to consider, it
can be used to perform a variety of illocutionary acts, and that, whatever
illocutionary act we choose to consider, various sentences can be used to
perform it. The sentence I’ll be back can be used to promise, threaten,
warn, predict, etc., without structural change, and many structurally dis-
tinct sentences can be used to promise that one will be back.10 There is, as
it happens, neither a function from form to function, nor a function from
function to form. While there are some obvious exceptions – ritualized
formulas come to mind – very generally the relation between form and
function is many-many. If the presence of an illocutionary force indicator
at the beginning of a sentence-type is to indicate that it can only be used to
perform that act, the practice is unjustifiable. Some sentences do seem to
wear their functions on their sleeves – such as those cast in the Imperative

9
Among many sources, cf. Chomsky (1965). There was, and is, another understanding of the term
‘autonomy of syntax’ according to which syntax is respectable, while semantics and use are bunk.
I am using the term methodologically and nonjudgmentally.
10
Austin (1975) suggests that there might be a thousand distinct speech-acts, so, to achieve descrip-
tive adequacy, the syntactic practice I am criticizing should posit that many distinct Force Phrases.
Regrettably, nothing in Minimalism would prevent this.
67

On the Representation of Form and Function 67

mood – but I will now argue that even these sentences contain no illocu-
tionary force indicators.

2 Imperative Sentence-Types and Influential

Illocutionary Acts
Let’s first identify the sentence-types and the illocutionary acts under con-
sideration. Grammar helps in individuating sentence-types; the contrast
between Indicative sentences such as You are quiet and Imperative sen-
tences such as Be quiet! and You be quiet! is plain. The task of individuat-
ing speech-act types is more difficult. There is no lack of data: we can, for
example, learn by observing our own behavior since at least sometimes,
when we speak, we know what we are doing. But also a general theory
of action, even if informally stated, might help in individuating speech-
act-types. If we could gain a perspective from which the nature of human
action was clear to us, we might be able to see that speaking is, at least in
some respects, not special. The five distinct families of speech-act suggested
at the end of Austin (1975) reflect, at least in part, distinctions that apply
across actions generally, and he makes more specific proposals, as when he
compares the case of a speaker’s presupposing the existence of something
that doesn’t exist to the case of a person’s giving something away that he
doesn’t own. But action-theory has not yet systematized human activity in
such a way that its instances can be easily categorized, so its contribution
to this task is limited.
The use of grammatical criteria to individuate sentence-types, though
often straightforward, does, nevertheless, present dangers, due to the seduc-
tive nature of grammatical terminology. The term ‘imperative’ has seduced
many. As a grammatical term, it is the name of a grammatical mood – in
current English characterized by the absence of morphological inflection –
but the common meaning of the word suggests a certain kind of illocu-
tionary act. It is easy to fall into the error of thinking either that all and
only sentences cast in the Imperative mood are used to perform imperative
speech-acts or that all and only imperative speech-acts may be performed
using sentences cast in the Imperative mood. If ‘imperative speech-acts’ are
orders of some sort – as the common sense of the word would suggest –
neither claim is true. Not all uses of sentences cast in the Imperative mood are
commonly speaking imperative, and not all sentences used imperatively –
used to order – are cast in the Imperative mood. To avoid this termino-
logical tangle, I capitalize the name of the grammatical mood and leave
8

68 Robert Fiengo
the common word uncapitalized. In this connection, the importance of
distinguishing types and tokens can be felt. Speakers use sentence-types,
but utter sentence-tokens, and both sentence-types and sentence-tokens
may be cast in the Imperative mood. But while some sentence-types are
cast in the Imperative mood, no sentence-type of any sort is imperative,
since – as previously mentioned – no sentence-type bears any illocutionary
force at all.11 Some sentence-tokens are cast in the Imperative mood, but
not all of these are uttered with imperative force, and not all imperative
sentence-tokens are cast in the Imperative mood.
The family of speech-acts under consideration is what Austin (1975) calls
the ‘exercitives’  – those that involve the ‘exercising of powers, rights, or
influence’. As his book unfolds, Austin turns from providing lists of per-
formative verbs to categorizing illocutionary forces, but the former task
serves the latter well, since, by and large, a performative verb occurring in
an explicit performative sentence names the illocutionary act that the sen-
tence is being used to perform. Different performative verbs might name
the same illocutionary act, and there may be illocutionary acts that no
performative verb is the name of, but a list of performative verbs can go a
long way toward identifying and then categorizing the illocutionary acts.
Austin provides the following list of exercitive verbs, which, for conve-
nience, I have alphabetized:
Advise, announce, annul, appoint, beg, bequeath, choose, claim, command,
countermand, dedicate, declare closed, declare open, degrade, demote,
direct, dismiss, enact, entreat, excommunicate, fine, give, grant, levy, name,
nominate, order, pardon, plead, pray, press, proclaim, quash, recommend,
repeal, reprieve, resign, sentence, urge, veto, vote for, warn
The verbs that name exercisings of powers or rights can be separated
from the verbs that name exercisings of influence. Powers are governed
by the laws of a structured organization such as a branch of the military,
a corporation, or a religious hierarchy. Some of these laws imply a differ-
ence in rank between speaker and hearer, others do not. Adjusting Austin’s
terms, let us call the former ‘privileges’ and the others ‘rights’: some pow-
ers are privileges, others are rights. On the side of privileges, examples
include: demoting a subordinate, pardoning a criminal, and fining a jay-
walker. In performing such speech-acts, speakers exercise powers over
others. On the side of rights, there is nominating, announcing, and
bequeathing, whose performance implies no difference in rank.

11
Also, of course, the types of sentences that are used imperatively are not imperative.
69

On the Representation of Form and Function 69

When a speaker exercises a power of either sort, he is often not address-
ing any specific person; he is merely making a move in a game defined
by a hierarchical organization. These acts tend to be perfunctory rather
than collaborative; they tend to be isolated public pronouncements, not
turns in an ongoing conversation. With the addition of ‘dictate’ – which
Austin omitted – the verbs from Austin’s list that name exercisings of pow-
ers (privileges or rights) are the following:
Announce, annul, appoint, bequeath, choose, claim, command, counter-
mand, dedicate, declare (open or closed), decree, degrade, demote, dictate,
dismiss, enact, excommunicate, fine, give, grant, levy, name, nominate, par-
don, proclaim, quash, repeal, reprieve, resign, sentence, veto, vote for
Remaining are the verbs that name the various exercises of influence. In
this category, Austin’s list is also incomplete. It should include the follow-
ing verbs; those taken over from Austin’s list are italicized:
Advise, admonish, advocate, appeal to, ask, beg, beseech, challenge, coax,
counsel, demand, direct, encourage, enjoin, entreat, exhort, forbid, implore,
incite, induce, importune, insist, instruct, order, plead, prohibit, pray, press,
propose, recommend, request, require, suggest, tell, urge, warn
These verbs name speech-acts in the performance of which a speaker is try-
ing to exercise influence in a normal conversational setting.
As with the powers, there is a separation within the influential speech-
acts. Some influential acts imply a distinction not in rank, commonly
speaking, but in standing between the speaker and the hearer, while other
influential acts do not. Questions of standing are societal in nature and
highly specific to topic. Some questions of standing are settled on the spot;
familial standing, on the other hand, is fixed. Encouraging and proposing
are influential acts that don’t imply difference in standing, but ordering and
begging do. If Jones tells Smith to do something, Smith might complain
that Jones can’t tell him what to do, and he might justify this by pointing to
his lack of experience, his lack of expertise, his motives, etc. He might also
advert to his lack of familial standing by saying: You’re not my father! If Jones
begs Smith to do something, Smith might complain that Jones shouldn’t
demean himself to him in that way. These differences in standing are not
organizationally sanctioned in the way that differences in rank are, but they
are very real nevertheless. Thus we have two intersecting dimensions, which
mark (i) the distinction between exercising power and exercising influence,
and (ii) whether the performance of the act implies difference in either rank
or standing or not. Difference in rank is set by law; difference in standing
varies with the many dimensions that divide individuals themselves.
0

70 Robert Fiengo

Exercise of powers Exercise of influence

Implying In rank (privileges): Appoint, call to In standing: Admonish, appeal

difference order, command, decree, degrade, to, beg, beseech, command,
demote, dictate, dismiss, enact, demand, direct, enjoin, entreat,
excommunicate, fine, levy, order forbid, implore, importune,
(in a restaurant), pardon, proclaim, instruct, order, plead, prohibit,
quash, repeal, reprieve, sentence, pray, require, tell
veto
Not implying In rank (rights): Announce, bequeath, In standing: Advise, advocate,
difference choose, claim, dedicate, give, ask, challenge, coax, counsel,
grant, name, nominate, object, encourage, exhort, incite,
resign, second a motion, stipulate, insist, induce, press, propose,
veto, vote recommend, request, suggest,
urge, warn

There are many small worries that arise with these categorial assign-
ments, of course. Commanding, so-called, certainly can be the exercise
of privilege, underwritten by military rank, but commanding can also be
merely influential. The verb ‘command’ names both kinds of speech-act.
In this it contrasts with decreeing, which is the exercise of privilege only.
Whether vetoing is a privilege or a right depends on the system in force.
Can a beginner advise an expert on his specialty, or is this always a misplay?
There are many questions such as these, but some generalizations can nev-
ertheless be drawn concerning the sentence-types used for each category
of illocutionary act. First, Imperative sentence-types are used to exercise
influence in all of its forms. They are used for both begging and ordering –
where difference in standing between speaker and hearer is opposite – and
they are used both to encourage and to request – where difference in stand-
ing cuts no figure. We might beg Give me a chance!, or we might give the
order Give me your homework!, or we might encourage by saying Give it
another try!, or we might request Give me the salt! Secondly, the exercise of
power is generally not pursued by using Imperative sentence-types. Here
linguistic practice is varied: in some cases, the Indicative is used (to dismiss
a person: I never want to see you again!), in other cases, an explicit perfor-
mative form is used (I (hereby) fine you ten dollars), and in still other cases –
particularly in the case of exercising privileges – fixed forms of expression
are used (Order in the court!). Thirdly, the use of Imperative sentence-types
to exercise influence contrasts with the use of Indicative sentence-types to
exercise influence. Either in saying Peel me a grape, or by saying I like my
grapes peeled, a speaker may impel someone to peel him a grape. In the
former case, the request is illocutionary; in the latter case, the request is a
71

On the Representation of Form and Function 71

perlocutionary effect of an assertion. These observations will be fleshed out
more fully in the next two sections.

3 Austin’s Theory of Truth and Imperative

Influential Utterances
Let’s first address the question why the Imperative mood serves the exer-
cise of influence better than the exercise of power. A speaker, in using a
sentence to exercise either power or influence, does not aim at truth. In
that respect, exercitive utterances and statements contrast. Nevertheless,
exercitive utterances, like statements, contain descriptive elements, and if
they are to be effective, those descriptive elements must be clear. While
exercitive utterances do not aspire to truth, they must aspire to clarity, as
statements and many other speech-acts must. While we cannot assess the
distinction between exercitive utterances and statements by asking under
what conditions we say that each is true, we can ask under what conditions
we say that each is clear.
We can model our answer to this question on Austin’s (1950:  116;
1979: 122) answer to the question, When do we say that a statement is true?
A statement is said to be true when the historic state of affairs to which it
is correlated by the demonstrative conventions (the one to which it ‘refers’)
is of a type with which the sentence used in making it is correlated by the
descriptive conventions.
Statements are utterances, as Austin uses the term, and as I am using the
term here. Speakers use sentence-types to make statements and to perform
other speech-acts. Demonstrative conventions correlate each statement
with an historic state of affairs. That state of affairs is a token of a type of
states of affairs. If the demonstrated state of affairs is a token of the type
of state of affairs described by the sentence-type the speaker uses, we say
that the statement is true. The descriptive and demonstrative conventions
are complex, and they interact. In some simple cases, descriptive conven-
tions relate subjects of sentence-types to individual-types and predicates
of sentence-types to activity-types. Demonstrative conventions relate sub-
jects of sentence-tokens to individuals and predicates of sentence-tokens
to activity-tokens.
When do we say that a statement is clear? We say this when it suf-
fers from no lack of clarity. Lack of clarity could present itself at several
junctures – the statement might be mispronounced, or spoken too softly,
or uttered irrelevantly, but these lacks of clarity are not revealing here.
2

72 Robert Fiengo
There are two cases of importance, which tie in with Austin’s account of
truth. First, it might be unclear which token state of affairs the statement
refers to: this unclearness would derive from a failure of the demonstrative
conventions. The complaint would be that it is not clear which state of
affairs the speaker is talking about. Or, secondly, it might be unclear which
type the demonstrated state-of-affairs token should be taken as belonging
to. In this case, the unclearness of the statement would derive from the
descriptive conventions: the complaint would be that it is not clear what
the speaker is saying about, or how he is characterizing, the token state of
affairs in question. So lack of clarity might derive from defects in either the
demonstrative or the descriptive conventions; if a statement suffers from
none of the kinds of unclearness mentioned, we say that the statement
is clear.
Which kinds of unclearness may arise in exercitive utterances? It helps to
consider a particular case. Suppose someone says Go out and prove you’re bet-
ter than that! The demand gives pause, since there are a great many things
that a person might do to prove himself, and the speaker has set no limits.
The addressee might wonder what constitutes proof, what sort of activity the
speaker has in mind, and how good his activity must be to be better enough.
In these respects, we say, the speaker has not been clear. But where does the
unclearness lie? Modeling our answer on the case of unclear statements, we
can see that the fault lies in the descriptive conventions. The sentence-type is
descriptively correlated with a type of activity whose tokens vary widely and
whose limits are not clear-cut. Given this freedom, and the consequences
some choices might lead to, the hearer would want to choose carefully. He
would probably wish to produce the least difficult token of the activity-type
‘proving oneself’ that is acceptable.
But not only does the unclearness not lie in the demonstrative conven-
tions: it cannot lie in them since the utterance is not correlated with any
token state of affairs at all. If some token activity were demonstrated, the
addressee would not have the freedom of choice that he has. The speaker
has chosen to use a sentence-type which is correlated with an activity-
type, but whose utterance is not correlated with an activity-token, includ-
ing, in particular, a token of the activity-type described. This suggests a
generalization concerning the use of Imperative sentence-types to perform
influential speech-acts. A speaker may assume that, if he uses a sentence
(descriptively) correlated with an activity-type, but whose utterance is not
(demonstratively) correlated with a token of it, the addressee will take it
that he is being told to perform a token of the activity-type the speaker
described. One way to influence someone to do something is to describe
73

On the Representation of Form and Function 73

the kind of thing one wants done while not referring to any token of it.
Imperative sentences are, with respect to Austin’s account of truth, incom-
plete; the hearer’s task is to complete them.12
In the utterance we have been discussing – Go out and prove you’re
better than that! – the verbs ‘go’ and ‘prove’ are cast in the Imperative
mood. As it happens, that fact alone serves to explain why this utterance
is not correlated by demonstrative convention with any activity token. As
a matter of grammar, Imperative sentences lack Tense, and, also as a mat-
ter of grammar, Tenses serve to demonstratively correlate utterances with
activity-tokens. In this respect, the Imperative mood contrasts with the
Indicative. Indicative sentence-tokens do contain grammatical Tense, and
because they do, they are correlated by demonstrative convention with
token states of affairs. In general, a token Tense-morpheme demonstrates
the time of an activity-token, taken relative to the time of the utterance.
Tense is to be distinguished from Aspect. Aspects characterize an activity
as ongoing, or as having come to a conclusion, or as instantaneous. They
give the ‘temporal profile’ of an activity: they do not locate an activity with
respect to the time of utterance.13 The complexity of this topic must be
conceded: languages vary as regards grammatical Tense and Aspect. Some
languages have a Future Tense; English does not. Some languages have sev-
eral Tenses; English has only two: the so-called Present and Past Tenses.14 As
usual, grammatical terminology varies and can give rise to confusion. The
Greek Aorist, for example, is categorized in classical grammars as a Tense;
but as I am using the term here, its distinctive contribution is Aspectual.
Putting complications of this sort aside, however, we may derive a tenta-
tive general conclusion. It is not that Imperative sentence-types are suited
for the performance of influential speech-acts because they have some-
thing that other sentence-types lack (such as an illocutionary force indica-
tor): rather, they are suited for the performance of influential speech-acts
because, as a matter of grammar, they lack something that Indicative sen-
tences have: grammatical Tense.
Utterances cast in the Imperative do not refer to token activities, but
this does not prevent their containing expressions that refer to times

12
Frege (1918: 329) allows that commands have senses, but denies that those senses are thoughts, since
truth cannot arise for them. He does not say that commands are in any way incomplete. I do.
13
Aspects do occur in Imperative sentences: Be sleeping when the clock strikes twelve and Have finished
by the time I get back are examples.
14
There is no Future Tense morpheme in English. To speak of the future, the Present Tense is generally
used: in She is leaving tomorrow, She leaves tomorrow, and She will leave tomorrow, we have occur-
rences of the Present Tense morpheme demonstratively correlated with a time following the time of
utterance. The modal ‘will’ is a Present Tense form; the Past Tense form is ‘would’.
4

74 Robert Fiengo
and places. These expressions serve to inform a hearer when and where
he is to perform a token of the activity-type described; they do not place
a particular activity-token in time or space. The word-token ‘out’ in the
sentence-token Go out and prove you’re better than that! is correlated by a
demonstrative convention with a place that is ‘out’ from the place occu-
pied by the addressee. What is not demonstrated is any particular place, or
any particular act of proof. The specific absence of reference to an activity-
token is the grammatical peculiarity of sentences cast in the Imperative
mood. That absence explains why, in using an Imperative sentence-type,
a speaker can influence a hearer to produce a token of the activity-type
described.

4 Exercitive Perlocutionary Effects of Indicative Utterances

Let’s now turn to the perlocutionary effects of assertion. I  have isolated
one tactic a speaker can pursue to get a hearer to do something. He can
use a Tenseless Imperative sentence-type to perform an influential speech-
act. Being Tenseless, his utterance will fail to refer to an activity-token,
and the hearer will conclude that he is being asked to produce an activity
of the type described. There is another tactic available to a speaker:  by
describing a desirable state of affairs using a Tensed indicative sentence-
type a speaker may impel the addressee to make it so, and by describing an
undesirable future state of affairs he may impel the addressee to prevent it.
Analogously, by describing a present desirable state of affairs, he may impel
a hearer to preserve it or, if the present state of affairs is not desirable, to
change it. ‘Your wish is my command’ goes the saying, but in practice it is
more accurate to say: ‘You describe it; I make it so (or not so).’
The Imperative and Indicative tactics contrast in several respects. First,
and most fundamentally, in uttering an Imperative sentence-token, a
speaker performs an influential speech-act. But in uttering an Indicative
sentence-token (such as It’s cold in here), a speaker performs not an influen-
tial speech-act but a statement. It is only by uttering the sentence-token –
and only in certain circumstances – that a speaker may thereby impel a
hearer to do something. It is the perlocutionary effect that is exercitive,
not the illocutionary act performed. The distinctions among the exercitive
illocutionary speech-acts are also found in the perlocutionary effects of
statements.
Secondly, when using the Indicative tactic, a present or future state of
affairs is described, but no particular activity need be described or referred
to. Unlike the case of Imperative sentences, where the type of the activity
75

On the Representation of Form and Function 75

to be performed is described, the use of Indicative sentences for exercitive
effect may leave the nature of the activity to be performed completely open.
Ordering in a restaurant, one might say Bring me the trout, or I’ll have the
trout. In saying the first, the speaker orders the addressee to bring her trout.
In saying the second, the speaker states how things will be, but whether the
addressee or another waiter is to bring the trout is left open. By saying It’s
cold in here, a speaker may impel a hearer to act, but whether he is to raise
the thermostat, close the window, or provide a sweater is left open. The
use of the Indicative for exercitive effect is appropriate when it is under-
stood that the speaker does not care how the end is to be achieved, details
of execution being left to the hearer. Also, the Indicative tactic assumes
compliance: a safe choice if one is addressing one’s minions. The speaker
of the Indicative You will come here right this second! assumes that he will
be obeyed: indeed, he describes a future in which he has been obeyed. The
perlocutionary effect is to demand. The speaker of the Imperative Come
here right this second! assumes that he has the status to make the demand
but does not assume that he will be obeyed. The illocutionary act is a
demand.
Thirdly, when a judge says There will be order in this court, those in
court will have to do, or cease to do, various things, but none of them is
told what to do. The perlocutionary effect is an exercise of privilege rather
than influence. An influential utterance – such as Sit down and be quiet! –
would be inappropriate: the judge should not stoop to engage the court in
conversation: she should state how things must be. The Indicative tactic
is usually appropriate for the perlocutionary exercise of powers, since the
exercise of power is seldom a conversational move.
Fourthly, the perlocutionary effect of uttering an Indicative statement
and the illocutionary effect of uttering an Imperative sentence may be the
same. In saying Close the window, one may achieve the same effect as by
saying It’s cold in here. All of the exercitive effects may be achieved by saying
an Indicative sentence, but in saying an Imperative sentence only influence
is exercised.
Lastly, there is an important point about negation, which I  can only
treat briefly here. Pursuing the Indicative tactic, a speaker sometimes has
the freedom to use either a positive or a negative sentence. There are prac-
tical limits to this: saying that you will have the trout is quicker than listing
all of the things that you won’t have. But the perlocutionary effect of saying
I’m drowning! and of saying I can’t swim! may be the same. In Indicatives
such as the last, the role of the negative is unremarkable: a present state of
affairs in described in the negative, and the perlocutionary effect is to plea
6

76 Robert Fiengo
for help. But the role of the negative in Imperative utterances is a differ-
ent matter. An utterance of Don’t walk on the grass! prohibits the addressee
from producing a token of the activity-type ‘walk on the grass’ – or, per-
haps, from continuing to do so. The polarity is not between positive and
negative, but between impulsion and prohibition. As it happens, the syn-
tax of the ‘negative imperative’ is varied in the world’s languages. Some lan-
guages exhibit a ‘prohibitive’ morpheme distinct from the ‘negative’ found
in assertions. Other languages use a periphrastic expression for prohibitive
sentences. In the account I have offered here, the point is that both impul-
sive and prohibitive Imperative sentence-tokens fail to refer to activities;
the former impel the hearer to produce an activity and the latter prohibit
the production (or the continued production) of one. In both cases, the
activity in question is described in the positive.15 A  parallel point can be
made about the English word ‘no’. The prohibitive use of ‘no’ occurs in No!
Stop! (as opposed to the impulsive Yes! Keep going!). In assertive utterances,
for example: Did you do it? No, I did not, neither ‘no’ nor ‘not’ is prohibi-
tive; they are both negative. In assertions, ‘not’ expresses a polar opposition
between positive and negative. In exercitive sentences, the polar opposi-
tions expressed include accepting and rejecting, impelling and prohibiting,
and propounding and denying. In the former kind of opposition, matters
of truth are in point, in the latter oppositions, the polarities are illocution-
ary. It is understandable that some languages use distinct words to mark
the different polar oppositions.

5 Conclusions
Austin inaugurated the study of speech-acts, strictly distinguishing them
from the structured sentences used to perform them. Independently, he
offered an account of truth. Both contributions assume a distinction
between function and linguistic form. Here I  have argued that Austin’s
account of truth makes it immediately understandable why Imperative
sentence-types are better suited for the performance of influential exer-
citive speech-acts than Indicative sentence-types are. In Imperatives, the
specific absence of reference to activity-tokens – obvious from their overt
structures – impels the production of activity-tokens. If statements provide
the criterion for completeness, and completeness consists in the presence
of all of the elements identified in Austin’s account of truth, the specific

15
Consider the contrast between Don’t make me happy! and Make me unhappy! Only in the second is
a state described in the negative.
77

On the Representation of Form and Function 77

incompleteness of Imperative sentences makes them suitable for impelling
the production of activities. So Austin’s views on truth and his account
of speech-acts are mutually supportive. In using Indicatives, on the other
hand, speakers may achieve a wide variety of exercitive effects, but these are
perlocutionary, not illocutionary. Speakers exploit the distinction between
the Imperative and Indicative moods when deciding what to say.
In Fiengo (2007), I argued that syntactically distinct English sentence-
types are used to perform distinct kinds of questioning speech-acts.
Deformed sentence-types such as Have you seen Fred? and Who have you
seen? are structurally incomplete, and are used to ask ‘open’ questions. In
the former, the inversion of ‘have’ indicates that the subject and predicate
are held apart; the open question is whether they should be joined. In the
latter, the movement of ‘who’ leaves an empty argument position; the open
question is how it should be filled. In contrast with this, the structurally
complete, undeformed variants You’ve seen Fred? and You’ve seen who? are
used to ask confirmation questions. In the former, the subject and predi-
cate are joined, and the sentence is being used to confirm the truth of the
result, while the latter may serve to confirm the identity of an argument
already referred to.16 There is no need, I argued, to adorn the structures
of sentence-types with covert ‘Q-morphemes’ or Force Phrases:  the for-
mal characteristics relevant to function are overt. I  have made a paral-
lel argument here. Like sentences used to ask open questions, Imperative
sentence-types are incomplete: incompleteness, as in the case of questions,
judged relative to the completeness of the sentence-types used to assert.
It is natural that open questions and influential utterances should both
favor incomplete sentence-types: they are alike in that both demand not
merely a response or an ‘uptake’ of some sort, but the performance of
an activity of a kind determined by the sentence-type used to perform
them. Open questions require the addressee to produce information; exer-
citive speech-acts require the addressee to produce activities. Both kinds of
speech-act require production, a function that incomplete sentence-types
serve well. Such cases challenge the widespread assumption – exemplified
by Searle (1968) among many  – that propositional content is invariant
across the speech-acts. This assumption may be the simplest one, but the
cost of this simplicity is to ignore grammatical form. Would it also be sim-
pler to ignore the distinction between form and function?

16
Frege (1918: 329) holds that ‘wh’-questions do not have complete thoughts as senses but that ‘yes-
no’-questions do. In Fiengo (2007), I support the first claim and deny the second.
8

78 Robert Fiengo
The using of a tool and its physical structure are not the same. The
using of a tool is an activity; its structure is not. Being asked to describe
the structure of a hand-tool – a wrench, for example – we would likely
provide its physical dimensions, along with a metallurgical description,
including such things as tensile strength, torque value, and so forth. We do
not include the fact that it is used to tighten nuts because that fact is not
part of its physical structure. If asked to describe the using of the wrench,
however, we reply that, given that the wrench has the physical properties
that it does – which we might then proceed to trot out – we can, in moving
it in a certain way, tighten nuts of a particular size and shape. The using of
a wrench is not part of its structure, though, given its structure, a wrench
does lend itself to certain uses and not to others. Similarly, the using of a
sentence is not part of its structure, though, given its structure, a particular
sentence may lend itself to certain uses and not to others. The distinction
between a tool and its use is obvious, yet, unfortunately, when linguists
and philosophers address language, they tend to ignore it. There is, to be
sure, a relation between form and function in language; an autonomous
syntax of sentence-types and an Austinian account of their uses together
reveal its nature.
79

Ch apter  4

Uptake in Action
Maximilian de Gaynesford

1 The Austinian Analysis

In an influential series of articles, Rae Langton has argued that some speak-
ers are rendered incapable of performing the illocutionary acts they try to
perform and are thus, in a particular way, silenced.1 I shall first meet recent
strong objections to the general analysis underlying this argument, then
raise a significant new problem, revise the analysis accordingly, and finally
demonstrate its usefulness.
Langton explains her thesis by appeal to J. L. Austin, who claimed that
if a speaker is to perform an illocutionary act, she must secure the ‘uptake’
of her audience, where this means grasping the illocutionary force of her
utterance (Austin 1975: 116–117). In Austin’s own favoured example, to
warn an audience requires getting it to recognise one is issuing a warning.
Suppose that circumstances conspire against taking a speaker or her utter-
ances seriously. One manifestation of this may be that she fails to secure her
audience’s uptake. Suppose she tries to warn an audience, saying, ‘There is
a wolf coming!’, but they think she is merely joking or seeking attention.
Because they do not take her utterance seriously, they may not understand
that she is indeed issuing a warning. If Austin is right and uptake is indeed
a necessary condition on performing illocutionary acts, then this speaker
is unable to perform – better, perhaps, rendered incapable of performing –
the illocutionary act she tries to perform. And given this configuration of
circumstances, we may thus regard her as silenced.2

1
Originally appearing from 1990 onwards, and developing ideas of Catharine MacKinnon (1987:
163–197), the relevant papers have since been revised and collected in Langton 2009 (see in particular
essays 1–8). Jennifer Hornsby formulated arguments to the same conclusion at the same time, but
by revising Austin’s account of illocutionary acts and doing so, at least initially, without appeal to his
notion of ‘uptake’ (see in particular her 1994), so her versions are not the focus here.
2
That this need not be a bad thing is under-recognised. Where peoples’ attempts at illocutionary acts
would, if successful, subordinate others, there is reason to cultivate an environment in which those
acts could not successfully be performed (see de Gaynesford 2009a). This might be particularly

79
0

80 Maximilian de Gaynesford
Because this one particular analysis of one particular sort of silencing
turns on appeal to Austin’s claim about uptake, we may call it the ‘Austinian
Analysis’ (AA).3 Its components are these:  there is a particular form of
silencing in which:  (a)  speakers are taken as non-serious by their audi-
ence, and thus (b) fail to secure their audience’s uptake, where (c) uptake
is a necessary condition on performing illocutionary acts, so that (d) they
are unable to perform the illocutionary acts they try to perform, and thus
(e) should be considered as silenced.
It is rare and exciting that so seemingly diminutive an aspect of philoso-
phy of language (i.e. (c)) should grow into so general an analysis, with a
conclusion (i.e. (e)) that has such deep implications for our lives, across so
broad a range of institutions: social, legal, political. But AA has recently
come under strong attack.4 Miranda Fricker has argued that it is less empir-
ically likely than her alternative analyses of silencing because it requires the
erosion of a speaker’s human and epistemic status. (Fricker 2007, ch. 6.,
esp. pp. 137–142). Ishani Maitra has claimed that it does not offer an analy-
sis of illocutionary silencing because, for all the analysis shows, it is actu-
ally perlocutionary acts that disadvantaged speakers are unable to perform
(Maitra 2009). And Nancy Bauer has argued that it misses the point; to
focus on failure of uptake is to ignore the real issue, which is the absence of
human exchange (Bauer 2015, chs. 5–6, esp. pp. 83, 99).
These objections fail in their own aims, I shall argue (Section 2). But
they do support what we have independent reason to perform: a revision
of AA, by correcting the claim about uptake (Section 3). And the upshot
is positive, because revised AA gives fresh insight into the phenomena of
silencing, a claim I shall demonstrate for a particularly notorious and com-
plex case: Austin’s remarks on poetry (Section 4).

2 Objections and Replies

For reasons that will become clear, it is necessary to describe each of the
three objections first, before responding to them.

relevant to situations involving hate speech, though discussions of the issue (e.g. Waldron 2012) tend
to ignore the possibility.
3
The label is doubly apt if Martin Gustafsson is right, that debate over this issue stands out as one
where ‘participants actually read and discuss what Austin says’ (2011: 18).
4
Most objections to Langton focus on the application of this analysis – to the pornography debate
(e.g. Saul 2006: 229–248; Mikkola 2008: 316–320), but my focus is on the analysis itself. Many
other objections to Langton focus on a different argument altogether: that pornography subordinates
women (see my 2009a, which sharply distinguishes the two; also see Maitra 2012: 95–118). These too
I set aside to remain focused on the silencing issue.
81

Uptake in Action 81
Miranda Fricker argues that AA requires a conception of silencing that is
less ‘empirically likely’ than the alternative she favours (Fricker 2007: 141). The
objection seems mild enough; it does not insist or require that AA be some-
how contradictory or reliant on false premises. But set against Fricker’s careful
delineation of alternative conceptions of silencing, it is particularly telling
(2007, ch. 6). Her contention is that, on AA, the silencing in question would
be ‘purely communicative’, a failure of that ‘reciprocity’ between speaker and
audience which communication calls for and which is required for the audi-
ence’s uptake (2007: 140).5 On her alternative ‘epistemic model’, however, this
silencing would be based on a failure of credibility (2007: 140). In effect, and
with reference to the basic components of AA, this allows Fricker to move
from (a) to (e) whilst bypassing (b)–(d). In cases where speakers are not taken
seriously by their audience (i.e. (a)), what accounts for the fact that they are
silenced (i.e. (e)) is that they are (wrongfully) excluded from the community
of knowing, trusted informants, either because they are simply not asked for
information or because they are treated as mere sources of information (as
something un-minded might be treated).
Fricker then compares the two conceptions in terms of what each
requires ‘before the silencing effect kicks in’ (2007:  142). AA requires a
failure of reciprocity, which would be prior to and more basic than what
her conception requires: a failure of credibility. Indeed, so Fricker thinks,
AA would involve a considerable erosion of the speaker’s ‘human status’
(2007: 142), whereas her conception would involve a much less extreme
(but still potentially wrongful) depletion of confidence in the speaker. It
is the difference between denying the speaker epistemic status altogether,
and undervaluing that status. And because undervaluing (depletion of
confidence) is ‘the more empirically likely possibility’, in Fricker’s view
(2007: 141), she advances her conception as preferable.
Ishani Maitra argues that AA requires what it cannot provide: ‘a crite-
rion that successfully distinguishes illocutionary from perlocutionary acts’
(Maitra 2009: 318). Without such a criterion, she thinks, AA cannot assert
that it is a specifically illocutionary act that a speaker is unable to perform
in the silencing in question. And she contends that we lack such a crite-
rion; neither Austin nor his successors have provided one.6 In effect, and

5
Fricker appeals explicitly to the semi-technical notion of ‘reciprocity’ (2007: 140), introduced and
developed by Hornsby as a way of distinguishing the illocutionary from the locutionary and perlo-
cutionary (see in particular Hornsby 1994). There is perhaps a connection with Simone de Beauvoir’s
The Second Sex, in whose arguments the notion of reciprocity plays a key positive role.
6
Austin himself was unhappy with the criteria he provides (1975: 109–132). Hornsby supplies her own
criteria, by appeal to the notion of reciprocity (1994: 192–195). But Maitra denies that reciprocity
does the work needed (2009: 309–338).
2

82 Maximilian de Gaynesford
with reference to AA, this allows Maitra to accept (a)  and perhaps (b),
but to deny (c) and what follows. For all AA can show, she thinks, what
speakers who are not taken seriously by their audience (i.e. (a)) are unable
to perform is perlocutionary acts. This might or might not be a significant
harm, but it would not be describable as silencing.
So consider examples of lesser and greater harms.7 Suppose I say, ‘There
is a wasp by your ear’, and you grasp that I am issuing a warning, but
decide not to do any of the things that this warning was meant to encour-
age you to do, because you think me timid. You do not take me seriously
and thus block my attempted perlocutionary acts, and that may do me
mild harm – ignoring my warning is a way of showing that you hold me in
some mild disrespect – but we would not describe this as a situation where
you silence me. Or consider a more serious harm: suppose I say, ‘I do not
want to go on this dangerous fairground ride with you’, and you grasp that
I am issuing a refusal but decide not to do any of the things this refusal
was meant to direct you to do, to leave me alone, and instead bundle me
into the ride regardless. Here there is a significant harm, involving perhaps
some physical and much psychological damage. But again, we would not
describe your behaviour towards me – in not taking me seriously and thus
blocking my attempted perlocutionary acts – as a form of silencing.
In short, if AA cannot distinguish an appeal to the illocutionary from an
appeal to the perlocutionary, it fails to discriminate and understand this (or
indeed any) form of silencing. So Maitra offers an alternative conception: the
silencing in question is a matter of communicative disablement, which she
elucidates by appeal to a Gricean communication intention theory, thus
bypassing Austin’s appeal to illocutionary acts.
Nancy Bauer argues that AA requires the audience’s uptake to play a
stronger role than it could, and a more significant role than it should.8
She thinks its role would have to be very strong because AA regards secur-
ing the audience’s uptake as not only necessary for the speaker, if she is to
perform an illocutionary act, but also sufficient for her performing that
act (Bauer 2015: 99).9 And she thinks its role would have to be very sig-
nificant because AA identifies what goes wrong in these kinds of silencing
case with the failure to secure the audience’s uptake. Bauer then contests

7
These are my examples, designed to make Maitra’s claims more salient.
8
Bauer has several other arguments in play: some of them are about the issue of subordination rather
than silencing; others are essentially about scholarship (whether AA captures what Austin says or
means to commit himself to). I set these to one side here, so as to concentrate on the silencing issue.
9
I shall not pursue the issue here, but this is not a defensible version of the uptake claim, and I see no
grounds for believing Austin held it.
83

Uptake in Action 83
both claims. In effect, and with reference to AA, this enables her to accept
(a) and perhaps (b) but to regard (b) as essentially irrelevant to (e) and to
reject (c)  altogether. Because AA gives too strong and significant a role
to uptake, it fails to discriminate and understand the kind of silencing at
issue, misidentifies what really calls for our attention, and hence fails to be
an effective analysis.
Bauer argues that uptake cannot play so strong a role because ‘sometimes
“uptake” is beside the point when it comes to the question of whether a
certain act has been performed’ (2015: 99). She offers this example: suppose
a woman orders a cup of coffee in a coffee shop and the waiter brings her a
dry-measure cup of beans. It would not be ‘illuminating’, Bauer urges, to
say that the woman failed to achieve uptake. And Bauer argues that uptake
cannot play so significant a role because what goes wrong in the silencing
cases at issue must actually be much deeper: it is ‘the fact that there is not a
human exchange going on here’ which we need to appreciate, not whether
or not there has been a failure of uptake (2015: 83).
These three objections to AA are problematic but also insightful. I shall
focus immediately on the problems, because we ought first to ask whether
these objections oblige us to reject AA.
These objections are fundamentally at odds with each other. This is sig-
nificant for various reasons, but particularly this: that if we ought to reject
AA, it is not because they operate collectively to show this. If Fricker is
right, then AA represents too communicative an analysis of the silencing
in question, whereas if Maitra is right, then AA is not communicative
enough. If Bauer is right, then AA adopts too anodyne an attitude to the
silencing in question, thus misidentifying what is actually at stake (human
exchange) with something superficial and insignificant, whereas if Fricker
is right, then AA adopts too hyperbolic an attitude to the silencing in ques-
tion, thus misidentifying what is at stake with something too deep and
extreme (human status). If Bauer is right, we can reject AA without appeal
to an alternative, whereas if Fricker and Maitra are right, rejection requires
an alternative. Moreover, these objections disagree fundamentally about
what undermines AA. For Maitra, it is its appeal to illocution. For Bauer,
it is its appeal to uptake. For Fricker, it is its appeal to reciprocity. So we
cannot endorse any two of these objections together. If any of them show
we ought to reject AA, it must do so individually and independently, and
that is how we shall consider them.
Fricker’s objection is that AA requires erosion of human and epistemic
status, but about this she seems mistaken. AA might require such erosion
if it portrayed the silencing effect as ‘kicking in’ before the question of
4

84 Maximilian de Gaynesford
the speaker’s epistemic status could even arise for an audience, in such a
way as to preclude its attributing such status to the speaker. And that is
evidently what Fricker has in mind (Fricker 2007: 140–142). But in fact AA
is predicated on the audience’s attributing epistemic status to the speaker,
and in a particularly rich sense. What is potentially silencing about the
situation in which the woman says, ‘there is a wolf coming’, for example,
is that her audience might be persuaded not to take her utterances seri-
ously because they think she is merely joking or seeking attention. But
this is not a situation that erodes the human status of the speaker, still
less one that undermines her status as a knower. If anything, the silencing
occurs because the audience treats the speaker as too knowing. They do not
take her utterance seriously because they believe it is not meant seriously,
and they do not believe it is meant seriously because they take her to be
playing epistemic games with them. They suppose that she knows that
they know that she knows that her utterances, which would in ordinary
contexts count as performing warning actions, are in this context not to
be taken as performing warning actions.10 To appreciate (a), that her utter-
ance is not serious, the speaker and her audience have to be reciprocally ‘in
the know’. So AA works by, not against, attributing significant epistemic
status to the speaker. This is a crucial feature of AA, and one that Fricker
overlooks, undermining her objection.
Maitra’s objection is that AA is unable to distinguish successfully in every
case between illocutionary and perlocutionary acts. But it seems wrong to
insist that AA needs a criterion that will do precisely that. Doubtless there
are border issues and grey areas. But if uptake is necessary for illocutionary
acts (and Maitra does not deny this), then all we have to show is that there
is no uptake in order to conclude that – whatever else is happening – no
illocutionary act can be taking place. And that is sufficient for AA to reach
its conclusion, that the speaker is silenced. (A rough analogy: if I know
hydrogen is necessary for water and that there is no hydrogen present, then
I may know there is no water present – regardless of the fact that I may
be quite unable to distinguish successfully in every case between hydrogen
and oxygen.) Doubtless the border issues and grey areas mean that it is
sometimes difficult or impossible to determine whether the speaker suffers

10
In applying AA to the pornography case, advocates similarly identify what is silencing with the
audience’s attribution of rich epistemic status to speakers. The reason why some men are said to
assume that some women do not seriously mean ‘no’ when they say ‘no’ in response to a sexual
invitation is that, encouraged by pornography, they assume a particular kind of knowingness on the
part of the woman: that she knows that they know that she knows she is only saying this as a fleeting
pretence of modesty, or as a flirtatious device to excite desire.
85

Uptake in Action 85
perlocutionary disablement as well. And that may be a cause of further
concern (of further injustice, perhaps). But AA has nothing to say about
this further issue, and it need not: being able to point out illocutionary
disablement is quite enough to identify a particular form of silencing.
Bauer’s objection is that uptake is often beside the point and that what
goes wrong in silencing cases lies much deeper. But it seems wrong to
think that these claims count against AA. What AA offers is a way of dis-
tinguishing a particular sort of silencing, not an explanation of whatever
background conditions make it possible, let  alone an account of what
makes it bad (harmful, wrong) tout court. Bauer thinks it is not ‘illuminat-
ing’ to point out that the coffee-ordering speaker fails to achieve uptake.
But whether something is illuminating depends on what one needs to
shed light on. If one wants to know what distinguishes the kind of silenc-
ing suffered by this speaker from the kind suffered by that speaker, then
uptake-failure is not ‘beside the point’ but illuminating: it tells us that this
speaker, unlike that speaker, is silenced in that she is rendered unable to
perform the illocutionary acts she is trying to perform. Bauer thinks ‘what
goes wrong’ in silencing cases is the absence of human exchange. But that
phrase is ambiguous between, roughly, what proximally causes the harm
and what fundamentally characterises the harm. Bauer is interested in the
latter, but AA claims only to speak to the former, so these positions are
perfectly consistent with each other. What proximally causes the harm is
what distinguishes the form of silencing at issue: that the speaker fails to
secure audience uptake. What fundamentally characterises the harm lies
deeper: it may be, as Bauer insists it is, the absence of elements crucial to
human exchange. And the advocate of AA is both free and able to endorse
this.11
So these objections do not oblige us to reject AA. But they do focus the
attention on what is unclear about the analysis. Fricker and Maitra are
right to prompt closer examination of what AA offers: is it meant to be a
purely communicative conception of silencing or not? And Bauer is right
to prompt closer examination of what AA requires: is it true or not that
uptake is necessary for performing the acts in question? In pursuing these
issues, we shall revise AA and modify its conclusion, but strengthen the
argument in its support.

11
Though this further question does get us into complex and fascinating issues, as we have found in
pursuing Fricker’s objection: i.e. in at least one sense of ‘human exchange’, this kind of silencing is
usually predicated on a peculiarly rich form of it. I don’t think this need ultimately count against
Bauer’s worry that there may nevertheless be something crucial missing here, but it does oblige us
to be more precise in specifying what that something is.
6

86 Maximilian de Gaynesford

3 Revising the Austinian Analysis

A purely communicative conception is attractive for two reasons. First,
it is consistent with a common view about Austin’s uptake claim: that it
operates as a requirement on communication (e.g. Bach and Harnish 1979:
13).12 And second, as a requirement on communication, the uptake claim
is hard to resist. Herbert Clark is evidently frustrated that anyone might
try (Clark 1996: 137–139; see also Clark 1982; 2004), and we can appreciate
why. For ‘communication’ is a certain sort of success term, one that marks
a joint achievement. Just to count as communicative, an act would require
something of both speaker and hearer. And, plausibly, it would require at
least this of the hearer: that they recognise the force of the speaker’s utter-
ance (Searle 1969: 47; Strawson 1964). Unless you understood that I was
issuing a warning, I could not be said to have communicated a warning
to you. That is an essential part of the joint achievement which commu-
nication represents.13 So if we agree with Fricker that AA offers a purely
communicative conception of silencing, we could resolve the tensions that
prompt Bauer’s demand for a closer examination of Austin’s uptake claim:
that claim will now seem neither too strong nor too significant.
But just as there are two good reasons to adopt the purely communica-
tive conception, there are two good reasons to resist it; and they are if any-
thing stronger. One reason has to do with AA itself: that it would not then
be what it claims to be: a conception of silencing. For the fact that I am
rendered unable to communicate does not show, of itself, that I have been
silenced. Suppose I issue a warning, but you fail to take me seriously and
thus fail to grasp that it is indeed a warning. Then you will not act in the
appropriate way, and we may all suffer. I have been tragically ineffectual,
no doubt, but there is nothing to show I have been silenced. If I was able to
issue the warning, I was able to perform the speech act I tried to perform.
The fact that I was unable to get you to appreciate the speech acts I do per-
form is another matter. A purely communicative conception would have
much to say about how a speaker would be ineffectual before an audience

12
They represent Austin’s view on uptake as follows:  ‘successful communication in performing an
illocutionary act consists in uptake, that is, in the hearer identifying the illocutionary act being
performed’ (1979: 130). But Austin did not think success consisted in uptake; he thought success
required it. There is a danger here and elsewhere of misinterpreting the uptake claim as a sufficiency
claim (evident for example in Bauer’s criticism of Hornsby; Bauer 2015: 101), one that Austin did
not make and was right to avoid, even for the specific case of communication.
13
This view, which underlies the treatment of communicative action in Habermas (1984: 273–337), also
underlies the views of those Habermas is directly representing: not so much Austin as Searle (1969).
Butler is critical in ways that she recognises are directly relevant to Langton (Butler 1997: 86–88).
87

Uptake in Action 87
that fails to take them seriously. But it would be an analysis of a speaker’s
impotence, not of their silencing.14
A second reason to resist a purely communicative conception has to do
with what is at stake in AA. What matters most to Langton is freedom of
speech (Langton 2009: 25–63; esp. 60–63).15 But the fact that I am ren-
dered unable to communicate does not show, of itself, that I have been
denied free speech. That freedom may give me the right to express myself
in certain ways. But to communicate requires that I have an audience, that
they listen to me or read me carefully, that they exert themselves to under-
stand me, and so on. And freedom of speech does not give me a right to
these things.
These particular problems disappear if we adopt the alternative concep-
tion instead, conceiving of AA as a straightforward illocutionary analysis
of silencing. On this conception, what fails in situations where speakers
are not taken seriously is that they fail to perform their illocutionary acts.
That they are thus unable to communicate follows, but is not the root of
the problem. So this conception can genuinely claim to be an analysis of
silencing; for someone disabled in this way is indeed silenced. And this
conception puts free speech at stake; for it is the right to express oneself in
certain ways that is being undermined. If what AA shows is that speakers
are rendered unable to perform the illocutionary acts they try to perform,
and we agree with Langton that ‘free speech is a good thing because it ena-
bles people to act, enables people to do things with words: argue, protest,
question, answer’ (Langton 2009: 61), then we can appeal to AA to justify
curtailing activities that render speakers unable to perform such acts. Or at
least we can neutralise the standard liberal counterargument, that curtail-
ing such activities would undermine freedom of speech.
But there is a significant drawback to the straightforward illocution-
ary conception. It pays a steep price for the license to count as a genuine
conception of silencing and to put free speech at stake. The advocate must
defend a correspondingly stronger version of Austin’s uptake claim. On
this version, audience uptake is necessary simply to perform illocutionary

14
Consider the woman who says, ‘There is a wolf coming.’ It is one thing to say that, not being taken
seriously, she is unable to communicate that she is issuing a warning. But the claim stimulating our
inquiry has been much stronger: that, not being taken seriously, she is incapable of even issuing that
warning. Someone in that condition may justly be described as ‘silenced’, despite being able to say
what she means, gathering an audience together, and having effects on that audience.
15
My focus here remains ‘what should we regard AA as offering?’ Langton herself may have moved
to a communicative, possibly purely communicative, conception, though her formulations are
guarded (e.g. ‘what is hoped for is a certain capacity to perform communicative illocutions’ Langton
2009: 73).
8

88 Maximilian de Gaynesford
acts, not simply to communicate them.16 I could not be said to warn unless
an audience recognised the force of my utterance, i.e. unless it under-
stood that I was indeed issuing a warning. And this reawakens the most
significant of Bauer’s concerns: that AA relies on too strong a claim about
uptake. For some agree with William Alston, who vigorously rejected the
uptake claim, urging that no illocutionary act needs uptake to be per-
formed (Alston 2000: 24; 67).17
Alston’s argument is based on appeal to two examples: ‘telling you that
the dean is coming to dinner’ and ‘asking you to bring me a towel’. I can
succeed at the corresponding illocutionary acts, he insists, whether or
not you even heard me, let alone understood the force of my utterances
(2000: 24). But even if he is right about this, two problem cases are evi-
dently not sufficient to show that no illocutionary act requires uptake.
And it is not clear that Alston is right about this, given what is peculiar to
his cases: that they are essentially addressee-involving (‘telling-you that p’,
‘asking-you to V’). If we contrast these cases with another pair – merely
saying that the dean is coming to dinner; merely asking for a towel – it
becomes clear why we might well regard Alston’s utterances as failing in
the absence of the addressee’s uptake. I could have said that the dean is
coming to dinner, asked for a towel, whether or not you even heard me;
but I would not then have told you that the dean is coming to dinner,
asked you for a towel. What is peculiar about Alston’s cases is that the
individuation of the corresponding illocutionary acts – and hence their
criteria of success – seems dependent on the active involvement of the
person addressed.18 Telling you could not be the particular illocutionary
act it is unless securing the addressee’s uptake were indeed required for its
performance. On being told that you had not heard, or had not under-
stood, there would be something bizarre about my continuing to insist,
‘Well, I did tell you!’ Retreating somewhat would be the natural step on
discovering how things stand: ‘Well, I did try to tell you.’ And this adver-
tises one’s sense that, not having secured your understanding, I have not
actually told you. The same goes for asking you.

16
Langton has sometimes expressed her view in ways that are consistent with this interpretation; for
example, in a chapter written jointly with Jennifer Hornsby: ‘Uptake consists in the speaker’s being
taken to be performing the very illocutionary act which, in being so taken, she (the speaker) is
performing’ (Langton 2009: 78).
17
See also Bird (2002: 1–15) and Jacobson (1995: 64–79). For contrary views, see Warnock (1989: 127)
Forguson (1973: 160–185) and Graham (1977: 91).
18
Not because of some general communication requirement. For discussion of related issues about the
second person, see Kukla and Lance (2009: 153–177).
89

Uptake in Action 89
But there are simpler cases that get around this problem and to which
an opponent of the uptake claim, like Alston, could just as easily appeal. If
I utter a sentence like ‘I describe myself as a socialist’, I need have no audi-
ence to perform the illocutionary act it names, let alone one that under-
stands the force with which I utter it. The same holds for sentences like ‘I
accept that I am partly responsible’ or ‘I consent to being taxed’ or ‘I blame
capitalism for this mess’ or ‘I curse this government and all who work for
it.’19 I am not claiming that such sentences are immune to performance
failure – i.e. that in uttering them, one could not fail to perform the illocu-
tionary act they name, no matter what the circumstances. But given that
they require no audience, it seems wholly implausible to suggest that this
might be the reason for failure on any occasion: that no audience recog-
nises the force of the utterance.
Austin himself was guarded about the uptake claim, and we can now
appreciate why he was right to be. He says that the claim is ‘generally’ in
operation and that, without uptake, the illocutionary act would ‘not have
been happily, successfully performed’ (Austin 1975: 116).20 ‘Generally’ might
mean ‘universally’, but there is at least equal reason to think Austin simply
meant ‘commonly’. And the second phrase strongly indicates that he had
not made up his mind on the issue. For Austin uses ‘not happily performed’
as a term of art, to denote acts that are open to criticism but are nevertheless
performed (his general name for such compromised successes is ‘abuses’;
Austin 1975, ch. 2),21 whereas his ‘not successfully performed’ pulls in the
opposite direction: that the acts in question are not performed at all (his
general name for such unequivocal failures is ‘misfires’; Austin 1975, ch. 3).22

19
I focus on explicit cases for simplicity, but the same is true of cases where the force is implicit, e.g.
replacing the first example with ‘I am a socialist.’ Complications I set aside include those raised by
Millikan (1998), to which Strawson (1998) replies.
20
Austin’s guarded phrasing is usually ignored, and he tends to be assigned a very strong version of
the uptake claim, in accord with Strawson’s interpretation of him (Strawson 1964: 449–454): that
no illocutionary act is performed without uptake; that the audience whose uptake is secured must
be the ‘intended’ audience, the audience ‘addressed’; that understanding the force of a locution is
a matter of understanding a feature of the actual use of a sentence on a particular occasion; and
that uptake must be achieved by the speaker’s knowing and intentional involvement. It is surely
not coincidental that ascribing Austin these strong claims makes it easier to weld his account of
illocutionary acts to a Gricean account of non-natural meaning (see Bach and Harnish, passim);
the possibility of such attunement is something Maitra seems to overlook in contrasting Austinian
versions of AA with her proposed Gricean version (Maitra 2009: 309–338).
21
So Alston could agree: if I ask you to bring me a towel and you do not grasp that I am asking you
for anything, there is evidently something ‘unhappy’ about the act I have nevertheless performed;
you will not do for me what performing the act tried to get you to do.
22
Alston could not agree, even if Austin just meant ‘commonly’, for his position is that the perfor-
mance of illocutionary acts is never dependent on uptake.
0

90 Maximilian de Gaynesford
My own view is that we have yet to appreciate the complexities here.23
Austin’s successors lost his guardedness and have tended to think that the
uptake claim is either just obviously true for all illocutionary acts, or just as
obviously false of any of them. The fact that the debate survives on a tiny
diet of examples – a feature bequeathed by Austin – helps explain this.24
We need to open our inquiry to the full range of illocutionary acts. It
would be wrong to anticipate the full results of such an investigation, but
even an initial survey makes it overwhelmingly plausible that the uptake
claim is not general. Some illocutionary acts require the active participa-
tion of an audience and some do not.
We have already seen examples of those that do not. In examples of
those that do, it may be participation of the addressee that is crucial to
performance of the illocutionary act. ‘I entreat you to think again’, or ‘I
thank you for your gift’ are examples. Plausibly, I will not have entreated
or thanked at all in uttering these sentences, even if we have an audience
who grasps what I am about, unless you, my addressee, have grasped that I
am entreating and thanking. Suppose you do not hear me, but we are over-
heard. It seems reasonable to reflect as follows in such circumstances: ‘I did
say “I entreat you to think again,” with the intention of thereby entreating
N. N. to reconsider, but I wonder whether I actually did entreat N. N.,
for it seems that although many heard me, she did not; or if she did hear
me, that she did not understand my words; or if she did understand my
words, that she nevertheless, and for whatever reason, did not grasp that I
was entreating her.’25
These cases contrast with others where the uptake claim holds, because
participation of an audience is required, but that audience need only
take the form of a witness rather than an addressee. For example, if I say
‘I concede defeat in this election’ while addressing you (my opponent), it
is plausible to require that someone recognise I am indeed conceding, for
performance of the act named, but that ‘someone’ need not be you, deaf-
ened as you may be by the adulation of your supporters; a witness will do.
Again, if I say ‘I adjourn this meeting’, I need an audience to recognise the
force of my utterance. But if those I am addressing are still too heatedly

23
In the next paragraphs, I deepen one aspect of my (2011) analysis of the uptake claim and set other
aspects aside.
24
Thus Austin discusses warning alone (1975: 116–117), Strawson adds only bequeathing (1964: 448)
and Alston restricts himself to ‘telling you that’ and ‘asking you to’ (2000: 24).
25
Promising fits this category; its ‘relational structure’, as David Owens puts it, requires that the
addressee accept the speaker’s promise (Owens 2012:  219), which entails, minimally, that the
addressee understand the force of the speaker’s utterance.
91

Uptake in Action 91
in debate even to hear me, I may nevertheless succeed in adjourning the
debate if the secretary witnessing and recording proceedings understands
that this is what I have done in my utterance. (We can imagine circum-
stances in which it is important to know whether an act occurred during
the meeting or after it was adjourned, where the weight placed on this wit-
ness’s testimony is testimony to the truth of this conclusion.)26
This distinction between uptake-dependent and uptake-free illocution-
ary acts cuts across the standard ways of classifying such acts. Appealing
just to the cases I have discussed, the class of Expositives (Assertives) con-
tains the uptake-dependent ‘concede’ and the uptake-free ‘describe’;
Commissives contains the dependent ‘bet’ and the free ‘accept’; Directives
contains the dependent ‘entreat you’ and the free ‘consent’; Exercitives
(Declaratives) contains the dependent ‘adjourn’ and the free ‘curse’; and
Behabitives (Expressives) contains the dependent ‘thank you’ and the free
‘blame’.
The distinction also cuts across standard attempts to characterise illocu-
tionary acts in terms of their special intentional properties, as in the theory
of Bach and Harnish (1979) which develops ideas of Grice (1989: 213–223).
They regard describing, accepting and consenting as requiring uptake by
virtue of being ‘communicative’ illocutionary acts (1979: 13, 17, 70, 151).
But utterances naming such acts may be uptake-free, as we have seen.
Conversely, Bach and Harnish regard conceding, betting and adjourning
as ‘conventional’ illocutionary acts, where the uptake claim is not meant
to apply, since they are not ‘communicative’ acts (1979: 108–19). But utter-
ances naming such acts may be uptake-dependent, as we have seen.
And finally, the distinction cuts across standard attempts to character-
ise illocutionary acts in terms of their special normative properties, as in
the theory of Kukla and Lance (2009) which develops ideas of Brandom
(1994: 3–66). They would regard ‘entreat you’ and ‘thank you’ as requir-
ing uptake by virtue of being essentially second-personal illocutionary acts
(2009: 153–177), and this seems right, for the independent reasons that
I gave in response to Alston’s cases: that they require the active partici-
pation of an addressee. But conceding and adjourning are also uptake-
dependent, as we have seen, and yet for reasons that Kukla and Lance
do not admit, since conceding and adjourning do not require the active

26
‘I bet you all the money I have that Labour will lose’ is, arguably, different again. Undoubtedly
performance of the illocutionary act named requires the uptake of an audience, but what is required
may well be the uptake of an addressee (unlike my examples of ‘concede’ and ‘adjourn’) and the
uptake of a witness (unlike my examples of ‘entreat’ and ‘thank’).
2

92 Maximilian de Gaynesford
participation of an addressee (a witness suffices), and hence there is noth-
ing essentially second personal about these acts.27
In summary, we face a choice. It will help specify it to recall the basic fea-
tures of AA: that there is a particular form of silencing in which (a) speakers
are taken as non-serious by their audience, and thus (b) fail to secure their
audience’s uptake, where (c) uptake is a necessary condition on performing
illocutionary acts, so that (d) they are unable to perform the illocutionary
acts they try to perform, and thus (e) should be considered as silenced.
One option is to regard AA as offering a purely communicative analysis,
where we revise (c) to (c*) uptake is a necessary condition on performing
communicative acts, and (d) to (d*) they are unable to perform the com-
municative acts they try to perform. Now (c*) may be overwhelmingly
plausible, but (e) no longer follows. This is no longer an analysis of any
sort of silencing. And we can no longer appeal to this analysis as part of a
free speech argument.
The alternative option is to retain AA as a straightforward illocutionary
analysis. (e) follows from (a)–(d); this is indeed an analysis of a particular
sort of silencing. And (e) is a sufficiently strong conclusion; we can appeal
to AA as part of an effective free speech argument, e.g. to curtail activi-
ties that are responsible for situation (a). But there is a problem with this
option: (c) and (d) are false.
There is a revised form of this illocutionary option, based around the
closest true claims to (c) and (d), namely (c**) uptake is a necessary condi-
tion on performing some (‘uptake-dependent’) illocutionary acts; and (d**)
speakers are unable to perform the uptake-dependent illocutionary acts they
try to perform. But this version of the argument delivers something much
more modest in relation to speakers who are taken as non-serious by their
audience: i.e. (e**) they should be considered as silenced with respect to the
uptake-dependent illocutionary acts they try to perform.
This revised form of the illocutionary option – call it revised AA – is
certainly an analysis of a particular sort of silencing, so it is preferable to
the purely communicative alternative. Perhaps (e**) is not strong enough

27
To say this, we need not deny that such speech acts ‘essentially place agents in normative relation-
ships structured by the claims we make upon one another’ (Kukla and Lance 2009: 177). It is just
that there need be nothing essentially second personal about an illocutionary act for it to place
agents in this way. Kukla and Lance argue that, without invoking an addressee (what they call speak-
ing to another as ‘you’) or securing their active participation, my attempts to perform illocutionary
acts would treat others as mere ‘normative-status-trading-engines’ (2009: 177). But conceding and
adjourning are counterexamples: I can perform them without speaking to another as ‘you’, but also
without denying the agential status of those concerned – because I depend on the agential participa-
tion of a (third person) witness to my act.
93

Uptake in Action 93
to form part of an effective free speech argument, e.g. to curtail activ-
ities that are responsible for situation (a). But that is a substantial fur-
ther issue, for another occasion. What I  shall briefly demonstrate here,
in the final section, is that revised AA can still shed light on difficult and
complex cases.

4 Uptake and Poetry
Austin is notorious, at least in literary circles, for encouraging us to treat
speakers of poetic utterances – ‘poets’ for short28 – as non-serious.29 Frege
and others had said similar things.30 But Austin was particularly thorough,
seemingly intent on leaving no possible respect in which an audience
might take any poetic utterance as serious.
Poetry is a ‘use of language’ which is ‘ “not serious” ’(Austin 1975: 104);
in poetry, ‘language is … used not seriously’ (Austin 1975: 22); if an utter-
ance occurs in a poem, it ‘figures in a context not wholly “serious” ’ (Austin
1963: 24); in poetry, ‘the words’ are not ‘spoken “seriously” ’ (Austin 1975: 9);
in poetry, ‘the words’ are not spoken ‘so as to be taken “seriously” ’ (Austin
1975:  9); if we ‘issue an utterance of any kind whatsoever’ in ‘writing a
poem’, ‘it would not be seriously meant’ (Austin 1979: 241); if we ‘issue an
utterance of any kind whatsoever’ in ‘writing a poem’, ‘we shall not be able
to say that we seriously performed the act concerned’ (Austin 1979: 241); ‘If
the poet says “Go and catch a falling star” or whatever it may be, he doesn’t
seriously issue an order’ (Austin 1979: 241).
Literary critics sometimes claim that such attitudes are rife, at least
amongst analytic philosophers, and that Austin is partly to blame. We
might deny this, but it would also be interesting to ask what would follow
if they were right. Suppose then that (a) of AA applies: poets are taken as
non-serious by their audience, partly because of Austin’s remarks. Should
we conclude from this that poets are silenced, and that Austin is partly
to blame?
Consider cases in which it is clear that a speaker is attempting to per-
form an illocutionary act in their poetic utterance, and it is clear what act

28
The abbreviation is merely for convenience, and it is safe since nothing here turns on the point, but
of course poets are rarely to be taken straightforwardly as the speakers of their poetic utterances.
29
Austin’s remarks are complex, and the responses to them by philosophers and literary critics even
more so; I discuss them fully elsewhere (see de Gaynesford 2009b; 2011b; 2013; 2017). The points
I make here, in applying revised AA to these remarks, are original to this chapter.
30
Frege (1918) claimed that poetry is not fit for many of the tasks of language, like assertion, on the
grounds that ‘the necessary seriousness is lacking’ (der dazu nötige Ernst fehlt).
4

94 Maximilian de Gaynesford
they are trying to perform, because they make it explicit. Chaucer tries to
dedicate his poem Troilus and Criseyde in saying
O moral Gower, this book I direct
To thee and to thee, philosophical Strode. (Chaucer 2008: 585)
Shakespeare’s speaker tries to concede a point in Sonnet 130 in saying
I grant I never saw a goddess go;
My mistress when she walks treads on the ground. (Shakespeare
2002: 641)
Geoffrey Hill’s speaker tries to ask a question in his The Triumph of Love
in saying
So—Croker, MacSikker, O’Shem—I ask you:
what are poems for? (Hill 1998: 82)
Revised AA tells us that, since we are supposing (a) applies, these speak-
ers will fail to secure their audience’s uptake (i.e. (b)). But it also tells us
that only some illocutionary acts are dependent on securing that uptake;
namely, those that are ‘uptake-dependent’ (i.e. (c**)). And this illuminates
the situation by distinguishing between our cases.
Thus Chaucer’s ‘I direct’ (i.e. dedicate) may be regarded as uptake-
dependent. But this is not because dedicating requires the active partic-
ipation of an addressee (one can dedicate a book to a dead person). It
is because dedicating is a particular kind of act, one whose performance
plausibly requires a witness. Geoffrey Hill’s ‘I ask you’ may also be regarded
as requiring uptake, but in this case because it requires the active participa-
tion of an addressee. This is for reasons that we have discussed earlier in
this chapter, which make ask you an essentially second-personal illocution-
ary act. Shakespeare’s ‘I grant’ (i.e. I concede), on the other hand, is neither
witness- nor addressee-dependent but uptake-free. It is an illocutionary
act that one may perform without any audience at all, let alone an audi-
ence that understands the force of one’s utterance. Alone of these exam-
ples, then, this illocutionary act is performed regardless of Austin’s remarks
and their effects.
In short, there is a category of cases where (for the best of reasons per-
haps) Austin’s remarks nevertheless threaten to silence poets – some poets,
and on some occasions. And revised AA gives us the means with which to
appreciate this fact with clarity and precision.
There is one final related matter on which revised AA sheds light.
Suppose we focus not on poetic utterance but on Austin’s utterances about
poetry: should we take him as non-serious? There would certainly be good
95

Uptake in Action 95
grounds, given the notable levity that characterises his remarks on poetry
(see de Gaynesford 2009b; 2011b). Suppose then that Austin is in fact ser-
ious but his audience takes him not to be, so that (a) of AA applies. Should
we conclude from this that Austin is silenced in his remarks on poetry?
Revised AA tells us that Austin would then fail to secure his audience’s
uptake. But his remarks on poetry are uttered with the force of asser-
tion, which is not uptake-dependent. One may successfully assert what
one likes, without an audience, let alone their understanding that one is
asserting anything. So Austin is not silenced even if we regard him as non-
serious in his remarks on poetry. He performs these illocutionary acts,
regardless of our attitudes. And it is revised AA which explains why.31

31
I am particularly grateful to Savas Tsohatzidis for encouraging me to be more explicit about the
implications of revising the uptake claim. I thank audiences at the SOPHA conference in Montreal
(June 2015) and at Reading for helpful criticism.
6

Ch apter  5

Performativity and the “True/False Fetish”

Savas L. Tsohatzidis

If you hurl a tomato at a political meeting (or bawl “I protest” when

someone else does – if that is performing an action) the consequence
will probably be to make others aware that you object . . . : but this
will not make either the throw or the shout true or false (though they
may be, even deliberately, misleading).
Austin, How to Do Things with Words

1 Introduction
J. L. Austin took it to be a defining feature of the sorts of utterances that
he came to call “explicit performative utterances” that they are neither true
nor false, despite being utterances of declarative sentences, which are tra-
ditionally regarded as paradigms of truth-evaluability; and he did not feel
compelled to give arguments for his non-truth-evaluability thesis, point-
ing out that he considers it too obviously true to require defense. Wedded
as they have been to a truth-conditional conception of linguistic con-
tent, subsequent philosophers of language have refused to accept Austin’s
non-truth-evaluability thesis  – not surprisingly, since maintaining both
it and the truth-conditional conception of content would force them to
conclude that explicit performative utterances have no content at all. And
since those philosophers, along with everyone else, do acknowledge both
the contentfulness of explicit performatives and the significance of the
phenomenon of performativity to which Austin was the first to pay sys-
tematic attention, they have sought to explain what is special about explicit
performatives by devising accounts of them that not only do not incorpo-
rate Austin’s non-truth-evaluability thesis, but positively require precisely
what Austin was ruling out – accounts, that is, according to which what is
distinctive about explicit performatives cannot be understood unless they
are taken to be bearers of a truth value.

96
97

Performativity and the “True/False Fetish” 97

The belief that such accounts are reliable, together with the absence of
arguments on Austin’s part, appears to be the main reason why Austin’s
thesis that explicit performatives are truth-valueless “is denied by almost
everyone nowadays,” as Hornsby (2006:  904)  notes in an overview of
post-Austinian work on performativity. But that denial would be apt for
reconsideration, if it could be shown that anti-Austinian explanations of
explicit performativity face insuperable problems that are due precisely
to their assumption that explicit performatives are truth-evaluable. This
chapter proposes to make a contribution toward such reconsideration.
After a brief reminder of some characteristics of explicit performatives that
are acknowledged by all parties to the dispute, I  argue that attempts to
justify the denial of Austin’s non-truth-evaluability thesis by producing
explanations of performativity that essentially depend on the hypothesis
that explicit performatives are truth-evaluable cannot succeed for at least
two types of reason:  on the one hand, because utterances that, on the
proposed explanations, should be explicit performative ones turn out not
to be explicit performative ones; on the other hand, because utterances
that, on the proposed explanations, should not be explicit performative
ones turn out to be explicit performative ones. Since the source of these
explanatory failures turns out to be none other than the adoption of the
hypothesis that explicit performatives are truth-evaluable, I  suggest that
they strongly undermine the anti-Austinian view and vindicate Austin’s
thesis, in favor of which I then sketch an independent argument based on
the behavior of explicit performatives in deductive inferential contexts.
My conclusion is that the Austinian thesis can by no means be regarded as
having been superseded, and that Austin’s opponents might even have to
seriously consider adopting it if some of their own broader interests were
to be safeguarded.

2 Performativity and the Anti-Austinian View

The sorts of utterances to which Austin was focusing attention when he
came to use the label “explicit performative utterances,” and which are
commonly discussed under that label today, are utterances of grammati-
cally declarative sentences each of which is such that (a) its speaker, refer-
ring therein to himself/herself in the first-person singular, predicates of
himself/herself an illocutionary act named by a simple-present-tense,
active main verb (and its object, if it has one), and (b) its issuance in the
right circumstances constitutes the performance, by the speaker, of the act
8

98 Savas L. Tsohatzidis
that he/she thereby predicates of himself/herself. Thus, an utterance, in the
right circumstances, of the sentence “I deny that arithmetic is complete”
can constitute a speaker’s denial that arithmetic is complete, and so can
be an explicit performative utterance; whereas an utterance of the sen-
tence “I prove that arithmetic is incomplete” can under no circumstances
constitute a speaker’s proof that arithmetic is incomplete, and so cannot
be an explicit performative utterance. Similarly, an utterance, in the right
circumstances, of the sentence “I recommend the Hammerklavier sonata”
can constitute a speaker’s recommendation, to some hearer or hearers, and
for some purpose or purposes, of the Hammerklavier sonata, and so can
be an explicit performative utterance; whereas an utterance of the sentence
“I perform the Hammerklavier sonata” can under no circumstances con-
stitute anyone’s performance of the Hammerklavier sonata, and, therefore,
cannot be an explicit performative utterance.
Two constraints on explicit performativity, both of them noted by
Austin, are generally acknowledged and would be worth keeping in mind
in the present context. The first is that a declarative sentence otherwise
conforming to the type of declarative sentence just described is not, in
general, performatively usable – i.e., usable in such a way that its produc-
tion by its speaker can constitute the performance, by that speaker, of
the act named by its active main verb – if its active main verb occurs in a
grammatical person other than the first person or in a grammatical tense
other than the (simple) present tense: Although “I request your support”
can constitute my request of your support and “I offer you my car” can
constitute my offer to you of my car (and so, can be explicit performative
utterances), neither “I requested your support” nor “I offered you my car”
can constitute my request of your support or my offer to you of my car,
and so cannot be explicit performative utterances. Furthermore, although
my saying to you “I request your support” can constitute my request of
your support, and my saying to you “I offer you my car” can constitute my
offer to you of my car, it is not the case that your saying to me “You request
my support” can constitute my request of your support, or that your say-
ing to me “You offer me your car” can constitute my offer to you of my
car, which means that neither of these latter utterances can be an explicit
performative utterance.
The second constraint on explicit performativity is that, even when a sen-
tence fully conforms, as regards the grammatical features of its main verb,
to the type of declarative sentence described previously, it is, in general,
only some, and not all, interpretations of those grammatical features that
are compatible with the sentence’s explicit performative use. In particular,
99

Performativity and the “True/False Fetish” 99

if the present tense of the sentence’s main verb is interpreted in a way that
allows utterance-time and reference-time to diverge, the sentence cannot
normally be used performatively. Thus, “I promise never to lie again” can
constitute my promise never to lie again (and so, can be an explicit per-
formative utterance), if the present tense of its main verb is understood as
referring strictly to the time of speaking (as it would normally be, if the
utterance was, for example, my response to your utterance of the impera-
tive, “Promise me never to lie again!”). But it cannot constitute my prom-
ise never to lie again, if, in uttering it, I am simply explaining to you what
I  do whenever I  am caught lying (as in, “Each time I  am caught lying,
I promise never to lie again”); or if, in uttering it, I am simply describing
to you the commitments I have undertaken in a letter that I have posted
earlier today (as in, “In the final paragraph of the letter I have posted earl-
ier today, I promise never to lie again”); or if, in uttering it, I am simply
rehearsing what I plan to do in an upcoming important meeting (“Here is
my plan for tonight’s meeting. First, I thank them for accepting to see us.
Second, I apologize for having lied yesterday. Third, I promise never to lie
again.”) It is, in short, only on specific interpretations of its grammatical
features – and, in particular, on the (relatively uncommon) interpretation
of the present tense as referring strictly to the time of speaking – that an
utterance of a declarative sentence of the sort described previously can nor-
mally be an explicit performative utterance.
Austin called “explicit performative verbs,” or simply “performative
verbs,” the verbs that, under the constraints just noted, can be used as the
main verbs of explicit performative utterances. And, in moving from the
exposition of his doctrine of explicit performative utterances to the expo-
sition of his doctrine of illocutionary acts, he suggested (1975:  149–150)
that, if one wants to obtain a minimally comprehensive list of the types of
illocutionary act that are commonly recognized in a language, there is no
better method than to compile (as he himself had set out to do for that
purpose) a list of the explicit performative verbs of that language – that
is, a list of those verbs that can be used as the main verbs of explicit per-
formative utterances whose issuance in appropriate circumstances would
constitute the acts that the verbs name.1

1
In expounding his doctrine of illocutionary acts, Austin occasionally used the term “performative”
not merely as a shorthand for “explicit performative,” but also in a different, extended sense, in
which it refers to any utterance accomplishing some speech act or other, whether or not the utterance
names (in the way his “explicit performative utterances” do) the act it accomplishes. Coupled with
Austin’s thesis that every natural language utterance normally accomplishes some speech act or other,
this terminological choice has the unfortunate consequence that every normal natural language
0

100 Savas L. Tsohatzidis

Apart from their use in providing him with what he regarded as an
appropriate point of entry into his nascent theory of illocutionary acts,
the main use that Austin has made of explicit performative utterances
(and the one that is our primary concern here) was a polemical one. For
Austin took it to be a defining feature of explicit performative utterances
not only (i)  that their speakers, by issuing them in appropriate circum-
stances, accomplish the illocutionary acts that they thereby predicate of
themselves, but also (ii) that, in predicating those acts of themselves, they
do not produce, nor do they intend to produce, a truth-evaluable represen-
tation of themselves as accomplishing the acts in question (and so cannot
be supposed to be accomplishing them by way of describing themselves
as accomplishing them, or by way of stating that they are accomplishing
them, or by way of declaring that they are accomplishing them, or by any
other way that would require of them to be producing, in accomplish-
ing them, a truth-evaluable representation of themselves as accomplishing
them). However, explicit performative utterances are utterances of gram-
matically declarative sentences; and since, on the view that was prevalent
in the philosophy of Austin’s time, utterances of grammatically declarative
sentences (if such sentences did not belong to those that were condemned
as nonsensical for positivistic reasons) were assumed to be, and to be
intended to be, truth-evaluable representations of reality, the existence of
explicit performative utterances constituted, for Austin, a distinctive and
decisive type of evidence against the aforementioned assumption, which
he dubbed “the descriptive fallacy” and against which he has found addi-
tional occasions to position himself in his work.2

utterance is “performative” in the extended sense, and thus deprives the notion of performativity in
the extended sense of any clearly distinctive theoretical role. In this chapter, I use “performative” only
as a shorthand for Austin’s “explicit performative” – that is, only by reference to utterances that name
(in their first-person, present-tense active main verbs) the illocutionary acts they accomplish – and
not in Austin’s inflationary extended sense; similarly for “performativity.”
It seems that Austin became aware of the problem created by his use of “performative” in the
extended sense when it was too late for him to correct it: the editors of How to Do Things with Words
tell us in their Appendix that, and at the point in Austin’s lecture notes where the transition to the
doctrine of illocutionary acts (and to the inflationary use of “performative”) is effected, there is “a
marginal note dated 1958” in which Austin writes, “All this isn’t clear!” and asks rhetorically, “Won’t
all utterances be performative?” (Austin 1975: 167). Unlike Austin, some of his commentators, and all
of his popularizers, appear never to suspect that the use of “performative” in the extended sense risks
trivializing the notion of performativity; what is worse, they sometimes advance arguments where
conclusions about performativity in the strict sense (that is, explicit performativity) are fallaciously
drawn from premises concerning performativity in the extended sense, or conversely.
2
Austin’s non-truth-evaluability thesis regarding explicit performatives is asserted in all three of
his extended treatments of them that have followed his brief, incidental, discussion of the topic
in his 1946 article “Other Minds”: the 1955 Harvard lectures posthumously published as How to
Do Things with Words (Austin 1975), the 1956 BBC talk posthumously published as “Performative
101

Performativity and the “True/False Fetish” 101

Post-Austinian philosophers of language are, if anything, even more
unwilling than some of Austin’s own contemporaries might have been
to accept Austin’s claim that explicit performative utterances are truth-
valueless, since most of them have come to subscribe to truth-conditional
theories of linguistic content; if Austin were right, therefore, their accounts
of linguistic content should be acknowledged to have the undesirable con-
sequence that they constrain them to claim, about certain obviously con-
tentful utterances of natural language sentences, that they have no content
at all. On the other hand, post-Austinian philosophers of language would
have even less an excuse than Austin’s contemporaries might have to claim
that the truth-evaluability, and hence (for them) the contentfulness, of
explicit performatives simply follows from their declarative grammati-
cal form, since that claim, besides begging the question against Austin,
would beg the question against grammatical theory itself: whatever else a
grammar of a natural language is, it is nowadays widely agreed that it is
not something that is capable of delivering verdicts as to which linguistic
objects are truth-evaluable and which aren’t. It is therefore not surprising
that those post-Austinian philosophers of language who have attempted to
seriously address the issue raised by Austin should have sought to resist his
non-truth-evaluability claim about explicit performatives in a way that, at
least in appearance, neither begs the question against Austin nor makes
gratuitous assumptions about imaginary pronouncements of grammati-
cal theory. And the way that most of them have found to be adequate to
that task can best be represented as a kind of abductive argument in favor
of the denial of Austin’s claim – an argument, specifically, that purports
to justify the hypothesis that explicit performatives are truth-evaluable by
claiming that, if one assumes, contra Austin, that they are truth-evaluable,

Utterances” (Austin 1979: 233–252), and the 1958 Royaumont Abbey talk posthumously published as
“Performatif-Constatif ” and translated as “Performative-Constative” (Austin 1963). On all three of
these occasions, the non-truth-evaluability of explicit performatives is presented as evidence against
what Austin calls the “descriptive fallacy,” a view that he had already targeted under that name in
“Other Minds” (1946: 174; 1979: 103). The non-truth-evaluability of explicit performatives is also
asserted, and presented as evidence against the “descriptive fallacy,” in Austin’s 1950 article “Truth,”
which mentions other kinds of utterances of declarative sentences besides explicit performatives that,
in Austin’s view, are not truth-evaluable despite their declarative grammatical form (1950: 125–127;
1979: 130–132).
It may be worth noting that Austin’s teacher, H. A. Prichard, in a paper on promising written,
according to its first editor, circa 1940, and published posthumously in 1949 (see now Prichard
2002: 257–265), explicitly associates what Austin was later to call explicit performativity with non-
truth-evaluability. Commenting on the promise made in uttering “I promise not to reduce the rates,”
Prichard writes that “while everyone would allow that a promise may be made either in good or in bad
faith, no one would allow that it could be either true or false,” and adds that “promising resembles ask-
ing a question or issuing an order in that it consists not in making a statement” (Prichard 2002: 258).
2

102 Savas L. Tsohatzidis

then one can explain the most distinctive feature of explicit performatives
that both Austin and everyone else acknowledges, namely, that their speak-
ers can accomplish, in uttering them, the acts that they thereby name.
According to the proposed explanation, an explicit performative utterance
can accomplish, by being issued, the act that it names because it is a truth-
evaluable utterance of a declarative sentence which, unlike truth-evaluable
utterances of other types of declarative sentences, has the special property
that its truth-condition is such that it can be satisfied, thus rendering the
utterance true, by the utterance’s own issuance. (Thus, what accounts for
the fact that an utterance like “I deny that arithmetic is complete” can be
an explicit performative utterance – that is, can constitute a denial that
arithmetic is complete – whereas an utterance like “I prove that arithme-
tic is incomplete” cannot be an explicit performative utterance – that is,
cannot constitute a proof that arithmetic is incomplete – is that, although
they both are truth-evaluable utterances of declarative sentences, the for-
mer, but not the latter, has the special property that its truth condition is
such that it can be satisfied, thus making the utterance true, by the utter-
ance’s own issuance.) And, according to the abductive argument that I am
here reconstructing, it is precisely because it affords this explanation of
what explicit performativity really is that the hypothesis that explicit per-
formatives are truth-evaluable is justified, and can be upheld in opposition
to Austin’s thesis that they aren’t.
In a section of his Philosophical Analysis in the Twentieth Century enti-
tled “A Lesson about Explicitly Performative Sentences,” Scott Soames
expresses this currently widespread anti-Austinian understanding of
explicit performativity in these terms:
[T]he explicitly performative sentence I promise to return the book has a
straightforwardly descriptive meaning that represents the world as being a
certain way, and so imposes truth conditions on it. What makes this sen-
tence special, and performative, is that one can bring it about that these
truth conditions are satisfied (i.e., one can bring it about that the proposi-
tion expressed by the sentence is true) simply by uttering the sentence in the
right circumstances. (Soames 2003: 127)
Soames mentions Lemmon (1962) and Lewis (1979) as exponents of ver-
sions of this view, but a far more influential twentieth-century philoso-
pher, Quine, could have been mentioned as well. In his contribution to
a symposium on Austin held in 1965 and first published in its entirety in
1969, Quine acknowledges that Austin’s discussion of explicit performa-
tives utterances in How to Do Things with Words was directed against what
Austin there calls  – identifying it as one of the principal targets of his
103

Performativity and the “True/False Fetish” 103

work – the “true/false fetish” (Austin 1975: 151), but claims that explicit per-
formative utterances do not justify “Austin’s animus against the true/false
fetish,” since what is special about an explicit performative utterance – in
Quine’s view – is not, as Austin thought, that it lacks a truth condition,
but rather that its truth condition is such that it can be satisfied by the
utterance’s own issuance, thus rendering the utterance one that is true:
“I bid you good morning” is true of us on a given occasion if and only if,
on that occasion, I bid you good morning. A performative is a notable sort
of statement, I  grant; it makes itself true; but then it is true. (Quine in
Urmson, Quine, and Hampshire 1969: 90)
Variously embellished, the view that Quine here expresses underlies almost
all post-Austinian accounts of explicit performatives, including the two
currently most popular (and otherwise antagonistic) ones, due respectively
to Kent Bach (1975) and to John Searle (1989). According to Bach, explicit
performative utterances are “statements” that “are true in virtue of being
made” (Bach 1975: 230); and, according to Searle, “performative utterances
are those in which saying something makes it true” (Searle 1998b: 115).
I will call the kind of view just outlined “the anti-Austinian view,” and
it will be the object of critical attention in what follows. It should be noted
that although all proponents of the anti-Austinian view agree that explicit
performative utterances are truth-evaluable utterances of declarative sen-
tences with the special property that their truth conditions are such that
they can be satisfied, thus making the utterances true, by the utterances’
own issuance, they need not agree, and in fact they rarely do agree, on
the question as to why explicit performatives utterances happen to have
this alleged special property (whereas truth-evaluable utterances of other
types of declarative sentences happen to lack it). Since examining, on a
case-by-case basis, the conflicting answers to that question given by vari-
ous anti-Austinians would be relevant only if one had no doubt about the
correctness of the basic anti-Austinian claim to which they all subscribe,
and since it is the correctness of that basic claim that I doubt, I will have
nothing further to say about internal disputes within the anti-Austinian
camp in this essay.

3 Against the Anti-Austinian View, Part I

Suppose, as the anti-Austinian view has it, that the explicit performativity
of an utterance (that is, the fact that it can constitute the act that it names)
is due to its having a truth condition that is such that it can be satisfied by
4

104 Savas L. Tsohatzidis

the utterance’s own issuance, which can thus make the utterance one that
is true. It follows from this that any utterance that has the same truth condi-
tion as the truth condition that, by hypothesis, a given explicit performa-
tive utterance has will also be capable of constituting, through its issuance,
the act that it names. In other words, if performativity is the result of utter-
ing a sentence that has a certain kind of truth condition, then utterances of
any sentences that share an explicit performative’s truth condition should
also share its performativity.
Call this consequence of the anti-Austinian view Thesis T. Among the
ways of seeing that Thesis T is false, and that, therefore, the anti-Austinian
view cannot be right, a simple one consists in assuming the nowadays
standard, Kaplanian view of the logic of indexicals, and in considering the
effect of replacing, in certain contexts, the first-personal indexical subject
term of an explicit performative utterance with a co-referential second-
personal indexical subject term. The relevant part of the standard view
of indexicals is simply the thesis that singular indexicals are directly ref-
erential expressions, and can be expressed as follows: given a declarative
sentence containing a singular indexical, the indexical’s contribution to the
truth condition that, relative to a context of use, the declarative sentence
has is just the referent of the indexical in that context of use, rather than
any “way of presenting” or any “way of fixing” that referent in that context
of use (notice that the standard view does not deny that different “ways
of presenting” or different “ways of fixing” their referents may be associ-
ated with distinct singular indexicals as parts of their distinct linguistic
meanings; what it asserts is that it is only their referents, and not those
“ways of presenting” or “ways of fixing” their referents, that the indexicals
contribute to the context-relative truth conditions of the sentences where
they occur; cf. Kaplan 1989a, 1989b). It follows from this that if the only
difference between two truth-evaluable declarative sentences is that they
contain distinct singular indexicals that are co-referential relative to a con-
text of use, then the two sentences have exactly the same truth condition
relative to that context of use. For example, relative to a context in which
John produces a token of (1a) while Maria, addressing John, produces a
token of (1b),
(1a) I am Italian.
(1b) You are Italian.
(1a) and (1b) have exactly the same truth condition, namely, the condition
that John is Italian; and relative to a context in which Maria produces a
token of (1a) while John, addressing Maria, produces a token of (1b), (1a)
105

Performativity and the “True/False Fetish” 105

and (1b) have exactly the same truth condition, namely, the condition that
Maria is Italian.
But now, if the standard logic of indexicals is taken to apply to explicit
performative utterances, as it must be taken to apply if explicit performa-
tives are assumed to be truth-evaluable, then the anti-Austinian view is
constrained to derive the obviously false conclusion that a host of utter-
ances that are clearly not explicit performatives ones are in fact explicit
performatives ones, since those utterances will have exactly the same truth
conditions as the explicit performative ones, and so their issuance will sat-
isfy those conditions just as much as the issuance of the explicit performa-
tive ones does. Consider a context in which John produces a token of (2a)
while Maria, addressing John, produces a token of (2b), and in which the
present-tense markers of both John’s and Maria’s tokens have the same
present time reference:
(2a) I deny that arithmetic is complete.
(2b) You deny that arithmetic is complete.
Assume – as it can certainly be the case – that John’s token is an explicit
performative one (that is, in producing it, John accomplishes the act,
which he names, of denying that arithmetic is complete). Then, given
the standard logic of indexicals, the anti-Austinian view entails the obvi-
ous falsehood that not only John’s but also Maria’s token can constitute
John’s denial that arithmetic is complete. For, on the assumption that
John’s explicit performative token is truth-evaluable, the standard logic
of indexicals will entail that John’s and Maria’s tokens have exactly the
same truth condition – namely, the condition that John denies that arith-
metic is complete. And on the further assumption that a token’s explicit
performativity is due to its having a truth condition that is such that it
can be satisfied by the token’s issuance, it will follow that the issuance of
either one of the two tokens can satisfy their identical truth condition,
and can thus be an explicit performative token – it will follow, in other
words, not only that John’s token of (2a) can constitute his denial that
arithmetic is complete, but also that Maria’s token of (2b) can constitute
John’s denial that arithmetic is complete. Since, however, it is obvious
that, although John’s token of (2a) can constitute John’s denial that arith-
metic is complete, Maria’s token of (2b) cannot constitute John’s (or, for
that matter, anyone else’s) denial that arithmetic is complete, the conclu-
sion must be that the anti-Austinian view cannot explain the explicit
performativity of (2a) without falsely attributing explicit performativity
to (2b).
6

106 Savas L. Tsohatzidis

Since this point is obviously generalizable to indefinitely many cases,
the explanatory failure of the anti-Austinian view that it signals is a
failure on a massive scale. Take any pair of utterances of present-tense,
declarative sentences of the forms “I Φ” and “You Φ,” naming an illocu-
tionary act of Φ-ing, of which the first is an explicit performative utter-
ance (that is, constitutes the act of Φ-ing that it names) while the second
is not an explicit performative utterance (that is, does not constitute the
act of Φ-ing that it names), and in which the present-tense forms have
the same present time reference and the first- and second-person singu-
lar indexicals are co-referential. Then, on the anti-Austinian assumption
that the utterance of the form “I Φ” is truth-evaluable, the standard
logic of indexicals will entail that “I Φ” and “You Φ” have exactly the
same truth condition. And this, together with the further anti-Austinian
assumption that the performativity of an utterance is due to its having a
truth condition that is such that it can be satisfied by the utterance’s issu-
ance, will entail the obvious falsehood that “You Φ” can be an explicit
performative utterance just as much as “I Φ” can – in other words, will
entail that the issuance of “You Φ” can constitute the act of Φ-ing that
it names just as much as the issuance of “I Φ” can constitute the act of
Φ-ing that it names.
In order to avoid this explanatory collapse while remaining true to his or
her anti-Austinianism, the anti-Austinian would have to defend the extra-
ordinary claim that an explicit performative utterance of the form “I Φ,”
even though it is, by anti-Austinian lights, truth-evaluable, cannot have the
same truth condition as a corresponding non-performative utterance of the
form “You Φ,” when the present-tense forms of both utterances have the
same present time reference, and their first- and second-person indexicals
are co-referential. But it is very hard to see how this claim could be credibly
defended, since its defense would require revisionary assumptions, which
hardly anyone would be willing to accept, about either the interpretation
of indexical expressions or the interpretation of performative verbs. If the
claim’s defense were meant to rely on a revisionary assumption about the
interpretation of indexical expressions (rather than of performative verbs),
the claim would be rejected since the requisite assumption would have to
be nothing less than the assumption that it is not the case that, if the mem-
bers of a set of declarative sentences differ only in that they contain distinct
singular indexicals that are co-referential relative to a context of use, then
all members of the set have exactly the same truth condition relative to that
context of use. But taking this not to be the case entails taking it to be false
that, for example, “I am fifty years old,” said by me to you today, has the
same truth condition as “You are fifty years old,” said by you to me today.
107

Performativity and the “True/False Fetish” 107

And any theory of indexicality that takes that to be false would certainly
be rejected.
If, on the other hand, the defense of the anti-Austinian claim were to
rely on a revisionary assumption about the interpretation of performative
verbs (rather than of indexical expressions), then it would be rejected too,
since the requisite assumption about the interpretation of performative
verbs would have to be that every apparently unambiguous performative
verb is ambiguous between at least two necessarily divergent senses, in
the first of which the verb has only first-person present-tense occurrences
and in the second of which it has only non-first-person, non-present-tense
occurrences (on that account, that is, the reason why, for example, “I deny
that arithmetic is complete,” said by me to you, and “You deny that arith-
metic is complete,” said by you to me, would, allegedly, have necessarily
disjoint truth conditions, would be that the verb “deny,” as used in the first
utterance, is not the same verb as, and has necessarily a different meaning
from, the verb “deny” as used in the second utterance). But the assump-
tion that every apparently unambiguous performative verb is ambiguous
in this highly peculiar way, besides being one that every anti-Austinian is
on record as emphatically denying, is one that would entail that explicit
performative utterances, as normally understood, do not exist at all, and
would, if only for that reason, be rejected. For, on the normal understand-
ing of performativity, an explicit performative utterance of the form “I
Φ” is, among other things, one whose issuance by a speaker would have
satisfied an imperative utterance of the form “Φ!” that would have been
addressed to that speaker (if you were to address to me the imperative
“Promise never to lie again!” my utterance of the performative “I promise
never to lie again” would have satisfied your imperative; if you were to
address to me the imperative “Admit that you are guilty!” my utterance
of the performative “I admit that I am guilty” would have satisfied your
imperative; and so on). On the assumption under consideration, however,
no explicit performative of the form “I Φ” could ever satisfy a correspond-
ing imperative of the form “Φ!” since, despite all appearances, the act that
the utterer of the performative would be naming and performing would,
of necessity, not be the act whose performance the utterer of the imperative
would have solicited. And this is surely a consequence that no account of
performatives (or, for that matter, of imperatives) would tolerate.
Since the anti-Austinian has no credible way of avoiding the explana-
tory collapse engendered by the falsity of Thesis T, the claim from which
Thesis T follows – namely, that the explicit performativity of an utterance
is due to its having a truth condition that is such that it can be satisfied
by the utterance’s own issuance – must be rejected. And, of course, there
8

108 Savas L. Tsohatzidis

are many other kinds of case that lead to the same type of explanatory
collapse. A different, and in a sense even simpler, kind of case becomes
apparent if one considers the effect of replacing, in certain types of con-
text, the indexical subject term of an explicit performative utterance with
a co-referential proper name, and if one assumes (as it is perfectly stand-
ard to assume) that, at least in that sort of position, proper names are as
much directly referential expressions as indexicals are (that is, contribute
only their referents to the truth conditions of the sentences containing
them). Suppose that Steven Stevens is amnesiac about his name, and
that, fully believing that he is someone other than Steven Stevens, he
makes what he takes to be a conjecture about Stevens’s current deeds
by saying,
(3a) Steven Stevens denies/bets/submits that there is life in other planets.
It is clear that Stevens would not have thereby denied/bet/submitted
(unbeknownst to himself, as it were) that there is life in other planets. On
the other hand, if Stevens were to say instead,
(3b) I deny/bet/submit that there is life in other planets.
he would have thereby denied/bet/submitted that there is life in other
planets. The anti-Austinian view, however, cannot acknowledge the expli-
cit performativity of (3b) without falsely attributing explicit performa-
tivity to (3a). For since, if both (3a) and (3b) were truth-evaluable, they
would, in the circumstances, have exactly the same truth condition, the
anti-Austinian view would wrongly entail that the issuance of (3a) can con-
stitute the act of denying/betting/submitting that it names just as much as
the issuance of (3b) can.
Indeed, it is not even necessary to invoke unusual conditions like a
speaker’s ignorance of his or her own name in order to produce cases with
just this effect. Quite ordinary conditions like an addressee’s ignorance of
a speaker’s name would do as well. Suppose that I am Philip Philips and
know full well what my name is, but that you don’t know what my name
is, and I am fully aware of the fact that you don’t know what my name is.
If, in these circumstances, I were to say to you,
(4a) I condemn/applaud your actions.
I would have thereby condemned/applauded your actions. But if, in the
same circumstances, I were to say to you,
(4b) Philip Philips condemns/applauds your actions.
109

Performativity and the “True/False Fetish” 109

I would not have thereby condemned/applauded your actions, no matter
what I might be “privately” taking myself to be doing. However, if both
(4a) and (4b) were truth-evaluable, then, relative to a context in which they
would be uttered by me and addressed to you, they would have exactly
the same truth condition; and this, given the anti-Austinian explanation
of performativity, would inescapably lead to the obviously false conclu-
sion that my utterance of (4b) could constitute my act of condemning/
applauding your actions just as my utterance of (4a) could.
Since the anti-Austinian explanation of performativity was the principal
piece of evidence purportedly supporting the anti-Austinian’s abductive
argument against Austin’s thesis that explicit performatives are truth-
valueless, the conclusion must be that the explanation’s failure removes
the principal reason for upholding the anti-Austinian view and for thereby
doubting Austin’s thesis. Indeed, Austin’s thesis that explicit performatives
are truth-valueless might now appear to offer the simplest explanation of
the fact, revealed in the foregoing discussion of the anti-Austinian argu-
ment, that certain non-first-personal utterances that should behave exactly
as explicit performatives utterances do, if the latter were truth-evaluable,
do not so behave: they do not so behave because they are truth-evaluable
utterances whereas explicit performatives aren’t.

4 Against the Anti-Austinian View, Part II

The argument of the previous section was that anti-Austinian explanations
of performativity cannot be correct since there exist utterances that, on
those explanations, should be explicit performative ones, but nevertheless
turn out not to be explicit performatives ones. The argument of the present
section will be that anti-Austinian explanations cannot be correct for the
complementary reason that there exist utterances that, on those explana-
tions, should not be explicit performative ones, but nevertheless turn out
to be explicit performative ones.
Suppose again, as the anti-Austinian view proposes, that the explicit
performativity of an utterance (that is, the fact that it can constitute the
act that it names) is due to the fact that its truth condition is such that it
can be satisfied, thus making the utterance true, by the utterance’s own
issuance. From this it follows that nothing could be an explicit performa-
tive utterance if, considered as something truth-evaluable, it would entail
something that happens to be false. For, if something that is truth-evaluable
entails something that happens to be false, it cannot itself be true. And if
something cannot be true, it cannot be true by virtue of being issued. So,
0

110 Savas L. Tsohatzidis

if an explicit performative utterance is something truth-evaluable that is
true by virtue of being issued, then nothing can be an explicit performative
utterance if, considered as something truth-evaluable, it entails something
that happens to be false.
Let us call this consequence of the anti-Austinian view Thesis F. There
are several types of counterexample showing that Thesis F is false, and that,
therefore, the anti-Austinian view cannot be right, but I shall here con-
centrate on a particular type of case, which is connected to the important
topic of the role of the concepts of sincerity and insincerity in accounts of
linguistic action. Consider the following utterances,
(5) I sincerely assert that I haven’t been here before.
(6) I sincerely wish you a speedy recovery.
(7) I sincerely apologize for my last remark.
(8) I sincerely promise to pay all my debts to you.
concerning which the following statements indisputably hold. First,
each one of them can constitute, by virtue of being issued, the act (of
asserting, of wishing, of apologizing, and of promising, respectively)
named by its first-person, present-tense main verb – in other words,
each one of them can be an explicit performative utterance. Second,
each one of them can constitute the act named by its first-person, pre-
sent-tense main verb (and so, can be an explicit performative utterance)
even when its utterer does not satisfy the characteristic sincerity expectation
associated with performances of the act in question – that is, even when
its utterer does not believe what he asserts (in the case where the act he
names and performs is an assertion), does not desire or hope what he
is giving a wish for (in the case where the act he names and performs
is the act of giving a wish), does not feel regret over what he apologizes
about (in the case where the act he names and performs is an apology),
and does not intend to do what he promises to do (in the case where
the act he names and performs is a promise). Thus, I can successfully
assert, in uttering (5), that I haven’t been at the place of my utterance at
a time prior to the time of my utterance, even though I do not believe
that this was the case, and indeed even if I know that it was not the case
(I may simply, in making my assertion, be lying to a police officer who
is conducting a murder investigation); similarly, I can successfully, in
uttering (6), give you my wish for your speedy recovery, even though I
don’t really want you to recover soon, and indeed even if I secretly hope
that you will not recover soon; again, I can apologize, in uttering (7), for
a previous remark of mine, even though, far from feeling the slightest
111

Performativity and the “True/False Fetish” 111

regret about having made it, I am secretly delighted about having made
it, noticing that it has had the damaging effect that I had intended; and
of course, I can promise, in uttering (8), to pay all my debts to you,
even though I do not have the intention of paying any of my debts to
you, but am simply trying to gain time in order to better organize my
disappearance.
The phenomenon that these cases exemplify is perfectly general, and
it is one that no adequate account of linguistic action should fail to rec-
ognize:  for any illocutionary act that generates a characteristic sincerity
expectation (in the case of assertions, the expectation that the speaker
believes what she asserts; in the case of apologies, the expectation that
the speaker regrets what she apologizes about; in the case of promises, the
expectation that the speaker intends to realize what she promises; and so
on), a speaker can successfully perform the illocutionary act – by, among
others, explicit performative means, when such are available – without sat-
isfying its associated sincerity expectation. And, of course, serious attempts
at constructing theories of illocutionary acts routinely recognize, under
various terminologies, this fact. For example, Searle and Vanderveken,
who call the characteristic sincerity expectation generated by an illocut-
ionary act the “sincerity condition” on that act, rely, in their Foundations
of Illocutionary Logic, on a fundamental distinction between, on the one
hand, “success conditions” on illocutionary acts (that is, conditions whose
non-satisfaction entails the nonperformance of the acts) and, on the other
hand, “non-defectiveness conditions” on illocutionary acts (that is, condi-
tions whose non-satisfaction does not entail the nonperformance of the
acts), and clearly acknowledge that all so-called sincerity conditions on
illocutionary acts are merely conditions on illocutionary non-defectiveness
and not conditions on illocutionary success (Searle and Vanderveken
1985:  chapter 1). And although Searle and Vanderveken mistakenly assert
(Searle and Vanderveken 1985: 13) that Austin “fails” to draw such a dis-
tinction, the truth is that Austin makes an exactly parallel distinction, and
an exactly parallel acknowledgment, in his long discussion, in How to Do
Things with Words (Austin 1975: Lectures II–IV), of the conditions under
which an explicit performative utterance naming an illocutionary act may
be, in his terms, “infelicitous” with respect to the performance of the act
that it names. For Austin there sharply distinguishes between two funda-
mental types of such conditions (and, correspondingly, two fundamental
types of “infelicity”): on the one hand, those whose non-satisfaction (called
by him a “misfire”) entails the nonperformance of the act named by the
performative verb, and, on the other hand, those whose non-satisfaction
2

112 Savas L. Tsohatzidis

(called by him an “abuse”) does not entail the nonperformance of the act
named by the performative verb; and he explicitly assigns all conditions
having to do with characteristic sincerity expectations associated with per-
formatively usable illocutionary verbs (these are the conditions belonging
to subclass “Γ.1.” in his classification of felicity conditions) to the category
of conditions whose non-satisfaction would merely constitute an illocut-
ionary “abuse” and not an illocutionary “misfire.”
But although it is evidently true and widely recognized that an illo-
cutionary act denoted by a performative verb can be accomplished even
when its agent does not satisfy the characteristic sincerity expectation
associated with that act, the implications that this phenomenon has on
the analysis of utterances such (5)–(8) have not been investigated, and are
especially significant in the present context. For, not only can speakers of
these utterances accomplish the acts they name without satisfying their
associated sincerity expectations, they can accomplish them in that way in
spite of the fact that they present themselves as satisfying those expectations.
And this means that, considered as truth-evaluable utterances, these utter-
ances could not fail to be false when they thus accomplish the acts they
name and accomplish.
The key fact to notice in order to appreciate this point is that, although
the bare statement that an agent performs a certain illocutionary act does
not entail (as just noted) that the agent satisfies the sincerity expectation
associated with that act, the statement that an agent sincerely performs
a certain illocutionary act does entail that the agent satisfies the sincer-
ity expectation associated with that act – that is why, for example, third-
person statements like (9)–(12) are not contradictory, whereas third-person
statements like (13)–(16) are contradictory:
(9) He asserted that he hadn’t been there before, even though he didn’t
believe that this was the case, and indeed knew that it wasn’t the case.
(10) He wished her a speedy recovery without really wanting her to
recover.
(11) He apologized for his remarks, even though he never felt regret about
those remarks.
(12) He promised to pay all his debts to her, without having the slightest
intention of paying any of his debts to her.
(13) He sincerely asserted that he hadn’t been there before, even though
he didn’t believe that this was the case, and indeed knew that it wasn’t
the case.
113

Performativity and the “True/False Fetish” 113

(14) He sincerely wished her a speedy recovery without really wanting her
to recover.
(15) He sincerely apologized for his remarks, even though he never felt
regret about those remarks.
(16) He sincerely promised to pay all his debts to her, without having the
slightest intention of paying any of his debts to her.
Once this point is appreciated, however, it is not difficult to see that the
undeniable fact that speakers of explicit performative utterances such as
(5)–(8) can be performing the illocutionary acts they name without sat-
isfying the sincerity expectations that they therein present themselves as
satisfying is simply inconsistent with the anti-Austinian explanation of
performativity. Suppose that, anxious to avoid the possibility of being
enlisted as a murder suspect, I produce, responding to a relevant inquiry
by a police officer, the following utterance:
(5) I sincerely assert that I haven’t been here before.
My utterance of (5) would certainly constitute the assertion that it names –
namely, the assertion that I haven’t been at the place of my utterance at a
time prior to the time of my utterance – even though I might not, while
uttering it, believe that I  haven’t been at the place of my utterance at a
time prior to the time of my utterance, and might even know that I  have
frequently been there in the past. (Notice that if the police officer later
discovers that I  have frequently been there in the past, he could rightly
accuse me of having asserted, falsely, that I have not been there in the past;
and that I could hardly defend myself against that accusation by claim-
ing that, since I never believed that I hadn’t been there in the past, I have
not made that assertion at all, and so cannot have made it falsely.) On the
anti-Austinian account, however, if I were to utter (5) without believing
that I haven’t been at the place of my utterance at a time prior to the time
of my utterance, I could not thereby be asserting that I haven’t been at the
place of my utterance at a time prior to the time of my utterance, and my
utterance of (5)  could not have been an explicit performative utterance.
For, on the anti-Austinian view, an explicit performative utterance is a
truth-evaluable utterance that is rendered true by being issued. Considered
as a truth-evaluable utterance, however, my utterance of (5) would, in the
envisaged circumstances, be certainly false, since what it would entail  –
namely, that I believe that I haven’t been at the place of my utterance at a
time prior to the time of my utterance – would itself be false. And since
an utterance that is false cannot at the same time be true, let alone true
4

114 Savas L. Tsohatzidis

in virtue of being issued, my utterance of (5) would wrongly be taken, on
the anti-Austinian view, to be an utterance that could not possibly be an
explicit performative utterance.
For exactly analogous reasons, the anti-Austinian view makes obviously
incorrect predictions about a host of relevantly similar cases. For example,
in uttering (6), (7), and (8),
(6) I sincerely wish you a speedy recovery.
(7) I sincerely apologize for my last remark.
(8) I sincerely promise to pay all my debts to you.
I may, respectively, be giving you my wish for your speedy recovery without
really wanting you to recover, be apologizing for a previous remark of mine
without really regretting that remark, and be promising to pay my debts to
you without in the least intending to pay my debts to you. On the anti-
Austinian view, however, none of these things could possibly happen; for,
(6), (7), and (8) cannot, on that view, be used for performing the illocu-
tionary acts they name unless they are truth-evaluable utterances that are
rendered true by being issued; and, since, considered as truth-evaluable
utterances, (6), (7), and (8) would, in the envisaged circumstances, be false
rather than true – given that what they would respectively entail about my
desires, my regrets, and my intentions would be false – it would follow,
contrary to fact, that (6), (7), and (8) could not possibly be, respectively,
my wish for your speedy recovery, my apology for my remark to you, and
my promise to pay my debts to you. Indeed, there is literally no end to the
wrong predictions that the anti-Austinian view would commit its adher-
ent to in this area: for any illocutionary act that is associated with a char-
acteristic sincerity expectation, and that a speaker performs through an
explicit performative utterance in which he or she purports to be satisfying
the act’s associated sincerity expectation without in fact satisfying it, the
anti-Austinian view will wrongly entail that the illocutionary act itself is
not performed at all, and hence that what is in fact an explicit performative
utterance is not an explicit performative utterance.
There is exactly one way in which the anti-Austinian might try to
remain anti-Austinian in view of this new explanatory impasse, but I don’t
believe that anyone, including the anti-Austinian, would, on reflection,
be willing to employ it: it would consist in stipulating that nothing is to
count as an illocutionary act unless its associated sincerity expectation is
in fact satisfied (thus, no one is to count as asserting something unless one
believes what one asserts, no one is to count as apologizing for something
unless one feels regret over what one apologizes about, no one is to count
115

Performativity and the “True/False Fetish” 115

as promising something unless one intends to do what one promises to do,
and so on). The problem with this stipulation, of course, is that it would
oblige the anti-Austinian to define out of existence a great many things
that without any doubt exist  – for example, lies. Suppose that you are
suspected of having committed a crime, and that, in response to a relevant
inquiry, you defend yourself by saying:
(17) I sincerely assert that I am innocent.
Everyone would agree that, in saying this, (a)  you would be asserting
that you are innocent, and (b) you might be lying about your innocence.
According to the stipulation under consideration, however, that would be
quite impossible: if your utterance is an assertion that you are innocent,
then it cannot be a lie about your innocence, since assertions that can be
lies simply do not exist. This is the sort of consequence that, I  suppose,
would force everyone, including the anti-Austinian, to reject the stipula-
tion under consideration as a defensible way of saving the anti-Austinian
view from its new explanatory impasse. And since no other decent way
appears to be available, no salvation appears to be forthcoming.
Since what causes this explanatory impasse is the falsity of Thesis F, the
claim from which that thesis follows – namely, that an explicit performa-
tive utterance is a truth-evaluable utterance that is rendered true by virtue
of being issued  – must be rejected. And if that claim is rejected, it can
of course provide no motivation whatsoever for the anti-Austinian view
that explicit performative are truth-evaluable. On the contrary, it is the
Austinian view that performatives are not truth-evaluable that, if accepted,
would help one understand why there has been an impasse here in the first
place: if explicit performatives are not truth-evaluable, then they do not
literally have entailments, and so the question of what happens to “their”
truth values when “their” entailments turn out to be false does not even
begin to make sense.

5 Against the Anti-Austinian View, Part III

I will now give an independent argument in favor of the Austinian claim,
which, if correct, undermines the anti-Austinian view at an even earlier
juncture than the previous ones, since it focuses just on the thesis that
explicit performatives are truth-evaluable utterances rather than on the
more specific thesis that they are truth-evaluable utterances that are ren-
dered true by being issued. The argument is based on the simple idea that if
explicit performative utterances were truth-evaluable, they should be able,
6

116 Savas L. Tsohatzidis

while retaining their performativity, to participate as premises in deduc-
tive arguments that would be recognized as instances of valid argument
forms. For, since recognizing such an argument as an instance of such
a form requires only that its premises be recognized as truth-evaluable,
rather than as true, nothing more – and, of course, nothing less – than its
truth-evaluability would be required of any explicit performative utterance
in order for it to be able to participate as a premise in a deductive argument
that would be recognized as an instance of a valid argument form. So, an
explicit performative’s inability to participate as a premise in an argument
of that kind would constitute evidence that its performativity is incompat-
ible with its truth-evaluability.
That performativity and truth-evaluability are indeed incompatible
for this reason can be seen by considering (among many other similar
cases) the sharp contrast in perceived validity between the “argument” in
(18*) and the argument in (18),
(18*) If you don’t have a gun, I don’t guarantee your safety.
I guarantee your safety.
Therefore, you have a gun.
(18) If Mary didn’t have a gun, John didn’t guarantee her safety.
John guaranteed Mary’s safety.
Therefore, Mary had a gun.
or the equally sharp contrast in perceived validity between the “argument”
in (19*) and the argument in (19):
(19*) If you don’t like pepper, I  don’t recommend to you the poulet au
poivre.
I recommend to you the poulet au poivre.
Therefore, you like pepper.
(19) If she didn’t like pepper, he didn’t recommend to her the poulet au
poivre.
He did recommend to her the poulet au poivre.
Therefore, she did like pepper.
No one, I suppose, would claim that the “arguments” in (18*) and (19*)
are in any sense valid arguments if their respective second premises are
read as explicit performative utterances (that is, if the second premise of
(18*) is read as being the guarantee that it names, and the second premise
of (19*) is read as being the recommendation that it names). On the other
hand, everyone would accept that the arguments in (18) and (19), where
117

Performativity and the “True/False Fetish” 117

the second premises cannot be read as explicit performative utterances, are
classically valid arguments. But these sharply contrasting judgments about
validity cannot be squared with the hypothesis that the second premises
of (18*) and (19*) are truth-evaluable when they are read as explicit per-
formative utterances. For, if the second premises of (18*) and (19*) were
truth-evaluable when read as explicit performative utterances, it would
be arbitrary in the extreme, and disrespectful of every compositionality
requirement, not to formalize (18*) and (19*) as instances of the same argu-
ment schema that (18) and (19) are instances of. And since that schema is
classically valid, one would be forced by classical logic to take the patently
invalid “arguments” in (18*) and (19*) to be exactly on a par, in respect of
their validity, with the clearly valid ones in (18) and (19). As a result, the only
way to reconcile the common formalizations with the sharply contrasting
judgments about validity would be to reject classical logic – in particular,
to deny that modus tollens is an unrestrictedly reliable inference rule.
And, of course, if modus tollens were to be given up on such grounds,
it should not be surprising that modus ponens would have to be given up
as well. No one, presumably, would accept that the “argument” in (20*) is
a valid argument, if its second premise is read as an explicit performative
utterance (that is, if its second premise is read as constituting the request
that it names):
(20*) If I ask you to have dinner with me, you pretend you are busy.
I ask you to have dinner with me.
Therefore, you pretend you are busy.
On the other hand, everyone would accept that the argument in (20),
where the second premise cannot be read as an explicit performative utter-
ance, is a classically valid argument:
(20) If he asks me to have dinner with him, I pretend I am busy.
He asks me to have dinner with him.
Therefore, I pretend I am busy.
But this sharp contrast in perceived validity cannot be squared with the
hypothesis that the second premise of (20*) is truth-evaluable when read
as an explicit performative utterance; for, if the second premise of (20*)
were truth-evaluable when so read, it would be compositionally irrespon-
sible not to formalize (20*) as an instance of the same argument schema
that (20) is an instance of; and since that schema is classically valid, one
would be forced by classical logic to conclude that the obviously invalid
‘argument’ in (20*) is exactly on a par, in respect of its validity, with the
8

118 Savas L. Tsohatzidis

obviously valid one in (20). Consequently, the only way to reconcile the
common formalization with the sharp contrast in perceived validity would
again be to reject classical logic, concluding this time that modus ponens,
no less than modus tollens, is an unreliable inference rule.
But rejecting classical logic would be quite an exorbitant price to pay in
order to save the hypothesis that explicit performatives are truth-evaluable.
And I  am reasonably confident that those who have been attracted by
that hypothesis would be quite unwilling to pay that price, and not pleas-
antly surprised to learn that the defense of their hypothesis would require
them to become wanton logical revisionists. I therefore conclude that it
would be in their best interests to reject that hypothesis and to accept that
explicit performatives are not truth-evaluable, just as Austin, without offer-
ing arguments, has been urging.

6 Conclusion
Austin, as I  mentioned in the beginning, claimed that his thesis that
explicit performatives are not true or false is too obviously true to require
argument:  “It needs argument,” he wrote (1975:  6), “no more than that
‘damn’ is not true or false.” It is clear with hindsight that, in making that
claim, Austin was greatly overestimating the degree to which subsequent
philosophers would be prepared to accept without argument the view that
explicit performatives are truth-valueless. I  have tried to show, however,
that Austin’s opponents can be given some good reasons for accepting
that view, and that their best argument for not accepting it (namely, that
by rejecting it they could explain how explicit performatives manage to
accomplish the acts they name) turns out to be fatally flawed. If I am right,
it is perhaps time to start suspecting that Austin’s opponents may have
been just too hasty in supposing that they have successfully neutralized the
serious threat that explicit performatives pose to the bundle of prejudices
that Austin was referring to as “the descriptive fallacy.”
119

Ch apter  6

The Vulnerability of Reality

Austin, Normativity, and Excuses
Sandra Laugier

The question of performative utterances seems to go hand in hand with a

reflection on social normativity, especially on the transgression of norms
implied in Austin’s thorough study of the failures (abuses and misfires)
of the performative in How to Do Things with Words (1975). Indeed, the
circumstances surrounding Austin’s discovery and introduction of speech
acts are grounded in thought about law and normativity. It is a well-known
fact that legal philosopher H. L. A. Hart was a member of the seminar,
held just after the war, in which Austin presented the theory of speech
acts. Moreover, Adolf Reinach’s theory of social acts, an early form of the
theory of speech acts, was developed in a legal context. Reinach’s A Priori
Foundations of Civil Law (1913) anticipates Austin’s speech act theory so
closely that some historians of speech act theory have proposed that Austin
was aware, via Ryle, of Reinach’s work.
My foremost aim here is to clarify the connection between speech acts,
normativity, and law. The connection between law and speech acts has
engendered all sorts of difficulties and misunderstandings, especially with
respect to the normativity of these acts and with respect to the existence of
norms or rules grounding them. This large question has often been taken
up in order to make normative use of Austin’s theories, for example by
Searle. For this reason, it is important to see what use can really be made
of Austin’s work in the theory of norms.
I will first approach this issue by examining the specific nature of the
speech act in Austin and contrasting it with its mainstream interpretations.
Reflection on law not only reveals the variety of speech acts beyond the
paradigmatic example of promising, it also shows what is truly at stake in
Austin’s theory. Furthermore, it alerts us to the full radicalism of Austin’s
philosophy of language. Austin’s mainstream interpreters and inheritors,
like François Recanati, in making performatives “social” acts determined
by rules, neglect the properly linguistic dimension of Austin’s discovery.
What is at stake, therefore, is the very idea of performance, in other words,
119
0

120 Sandra Laugier

the idea of agency, independently of any creation of a state of affairs or of
a social reality.
From this point of view, it is important to recall Austin’s permanent
insistence on the risk of failure of the speech act. This dimension of failure
is forgotten in most of the social theories of speech acts (founded on the
positivity and success of the act), but it is essential to law. Starting from
this theory of defeating conditions, we can, through a reconsideration of
Austin and Reinach, return to the question of the reality that speech acts
create, describe, or perceive.
We can then raise the general question of linguistic agency through the
problematic of excuses, and offer a new understanding of the normativity
of the performative, through Austin’s redescription of action by the clas-
sification of excuses and failures. Such an emphasis on failures and practi-
cal errors will lead to a different, resolute approach to the normativity of
language, and to reality itself.

1 Austin and the Reality of Speech Acts

Austin’s theory of speech acts, though well known in its own right, must be
considered in the context of the rest of Austin’s writings, and in particular
the essays “Truth,” “Pretending,” “A Plea for Excuses,” and “How to Talk.”
Austin has not “only” a theory of speech acts, but also a theory of truth,
of meaning, and of what it is to say something. He has a whole theory of
what is said. But the investigation of what is said is inevitably an investiga-
tion of the articulation of speech acts and of states of affairs. The discovery
of performatives puts into question, for all of our utterances, the idea of a
uniform relation between words and the world.
Yet this is not obvious, as demonstrated by mainstream accounts of
Austin’s work, for example, that of Recanati in his classic book Performative
Utterances: “By seriously making an utterance in a communication situa-
tion, the speaker performs a kind of social act, defined by the relation
established by means of the utterance between the speaker and the hearer”
(Recanati 1980:  19). Right away, Recanati situates Austin’s performative
utterance within a problematic that is defined by these terms: 1) a situation
of communication; 2) a social act. It is puzzling how Recanati could, in
this passage, define speech acts in communicative and institutional terms,
which are also, here, ontological terms: a relation is established by means of
the utterance.
The act (the promise) establishes a connection, by virtue of which
there is (in particular) an obligation to perform an action. This action is
121

Austin, Normativity, and Excuses 121

accordingly the end, the completion of the promise, to which the utter-
ance is the means. The act (promising) creates a situation (connection,
state of affairs). But this creation is not the only action involved, since the
content of the promise (the thing promised) is still to be carried out.
Austin’s criticism of the descriptive fallacy is wholly resolute:  he dis-
tances himself from any “ontological” or “representationalist” idea about
the effects of performative utterances – in particular, from the idea that
such acts create social relations or “make themselves true.” To be sure, most
uses of Austin in legal theory make the speech act the creation of a legal
reality. But Austin nowhere speaks of the creation of a reality (nor of the
creation of a situation). Of course one can point out that an act is always
a modification of reality, but for Austin such a remark would be close to
nonsense. Austin never speaks of such an entity as a “state of affairs” (in
contrast not only with Searle, but also with those like Barwise who talk in
terms of “situations”).
It is quite illuminating to confront Austin’s approach with Reinach’s in
order to understand the ontological status of speech acts. Both Austin and
Reinach begin by questioning the idea that the function of language is
essentially descriptive. Both have the idea that there are utterances that do
not reflect (represent) reality. The difference between the two rests, then,
not in the nature of their discovery – the fact that there exist non-descriptive
uses of language – but rather in the term with which they contrast what is
descriptive: Austin calls it “performative” and Reinach “ontological.” Both
emphasize the practical and social character of the speech acts.
Austin and Reinach are very close, at least, in their awareness of the
importance of their discovery. It is remarkable that both Reinach and
Austin think of themselves as having discovered something completely
new and unnoticed. They are considered discoverers, almost in an empir-
ical sense, of a new “phenomenon”: as if it were a matter of uncovering
a phenomenon in nature that was there all along. This mixture of famil-
iarity and strangeness characterizes descriptions of the discovery of per-
formatives, as well as Austin’s descriptions of the phenomena of ordinary
language – a fact well noted by Stanley Cavell (1979:  chapter 4; see also
Laugier 2005, 2013). These things have always been there, right before our
eyes, but we have never paid attention to them.
All of this can strike us as obvious, or as curious, according to the attitude in
which we approach it. It is “obvious” in that it is something which everyone
knows, which everyone has passed by a thousand times, and which one can
now pass by for the thousand and first time. But just as in other cases it hap-
pens that our eyes suddenly open to something which we have long been
2

122 Sandra Laugier

familiar with, and that we really see for the first time in all its proper char-
acter and in its distinctive beauty what we have seen already many times, so
it can happen here too. (Reinach 1913/1983: 8)
Austin begins How to Do Things with Words by isolating a category of utter-
ances, or more specifically a “phenomenon,” that is “obvious” but to which
not enough attention has been paid.
The phenomenon to be discussed is very widespread and obvious, and it
cannot fail to have been already noticed, at least here and there, by others.
Yet I have not found attention paid to it specifically. (Austin 1975: 1)
To say that there are speech acts is not, however, to offer a theory: it is
simply the observation of a phenomenon to which philosophy has not
paid attention, especially the philosophy of language – since the dominant
paradigm in the philosophy of language equates the sense of an utterance
with the representation of a state of affairs. In affirming this neglect, Austin
attacks the representationalism on which all philosophy of language since
Frege had been founded: it is only if a proposition has sense that one can
ask whether it is true or false, i.e., whether it represents a state of affairs
or not.
The classical expression of this thought (and what Austin aims to
criticize) can be found in a standard reading of the Tractatus Logico-
Philosophicus: a thought is a proposition provided with sense, about which
one can only ask whether it is true or false. There can be no ethical propo-
sitions, since they do not state anything. I note this point because Austin
himself discusses ethics and the nonsense of ethical propositions, as well as
the necessity of restoring to their place utterances that philosophers have
said lack sense.
Austin wants to break with the idea, which he calls “the descriptive
fallacy,” that the first function of language is the depiction of states of
affairs. A great number of linguistic expressions are used for purposes other
than to describe reality, and only the dominance of the representational-
ist model obscures this fact. For Austin, utterances do not represent: this
thesis is explicit in his essay on “Truth” (1950; 1979:  117–133), where he
criticizes the Tractatus, but is also present in “Other Minds”:
To suppose that “I know” is a descriptive phrase is only one example of
the descriptive fallacy, so common in philosophy. Even if some language
is now purely descriptive, language was not in origin so, and much of
it is still not so. Utterance of obvious ritual phrases, in the appropriate cir-
cumstances, is not describing the action we are doing, but doing it (“I do”).
(Austin 1946: 174; 1979: 103)
123

Austin, Normativity, and Excuses 123

The performatives that Austin describes in the first lecture of How to Do
Things with Words do not describe facts. Their utterance is the accomplish-
ment of an act. The remarkable thing about performatives is that they are
utterances that are also acts, not utterances that describe something (such
as an empirical state of affairs): “It has come to be commonly held that
many utterances which look like statements are either not intended at all,
or only intended in part, to record or impart straightforward information
about the facts” (Austin 1975: 2).
For Austin as for Reinach, the point is to show that language does things
other than describe, even in sentences that appear grammatically “normal.”
It was for too long the assumption of philosophers that the business of a
“statement” can only be to “describe” some state of affairs, or to “state some
fact,” which it must do either truly or falsely. (Austin 1975: 1)
Through the act of promising, something new enters the world. A claim
arises in the one party and an obligation in the other. What are these curi-
ous entities? They are surely not nothing. But they cannot be brought under
any of the categories which are otherwise familiar. (Reinach 1913/1983: 9)
The speech act that constitutes the promise thus cannot be a description of
a reality, either mental or physical. With Austin as with Reinach, its point
is not to describe any reality (neither an empirical nor a psychological
one).1 This is a particularly important point in the distinction between per-
formative and constative: “In these examples it seems clear that to utter the
sentence (in, of course, the appropriate circumstances) is not to describe
my doing of what I should be said in so uttering to be doing or to state that
I am doing it: it is to do it” (Austin 1975: 6).
A performative can be described, as can the performance accomplished
by saying it, or (because it is not a question of means) in saying it (whence
comes the term “illocutionary”). But this does not show that it is itself a
descriptive utterance.
[The utterance] should not be confused with statements about experiences
which are now taking place or have just taken place. If I say “I am afraid”
or “I do not want to do that,” this is an utterance about experiences which
would have occurred without any such utterance. But a social act, as it is
performed between human beings, is not divided into an independent act
and a statement [Konstatierung] about which it might or might not be made;
it rather forms an inner unity of voluntary act and voluntary utterance. For

1
For an interesting parallel, see also H. A. Prichard’s 1940 essay “The Obligation to Keep a Promise”
(Prichard 2002:  257–265). I  am grateful to Savas Tsohatzidis for having drawn my attention to
this text.
4

124 Sandra Laugier

the inner experience here is not possible without the utterance. And the
utterance for its part is not some optional thing which is added from with-
out, but is in the service of the social act. Of course there can be statements
about social acts that are accidental to them: “I have just given a command.”
(Reinach 1913/1983: 20)
One can note here the linguistic relevance of Reinach’s remark: it nicely
shows the difference and the connection between the performative and
the constative that describes its occurrence. In this way, the discovery
of the performative is, in the first instance, motivated by doubt about the
descriptive paradigm in the philosophy of language. This is clear from
Austin’s initial examples, which are mere actions:
“I name this ship the Queen Elizabeth” . . .
“I give and bequeath my watch to my brother” . . .
“I bet you sixpence it will rain tomorrow” . . . (Austin 1975: 5)
These are examples of utterances that, grammatically, resemble assertions,
but do not “describe,” do not “represent” any fact, and are neither true
nor false, even though they are used perfectly correctly. Their utterance is
equivalent to the accomplishment of an act.
To cast doubt on the descriptive function of language is actually to cast
doubt on the connection between meaning and states of affairs. Austin is
just as skeptical of the “fetishized” dichotomy “true/false” as he is of the
dichotomies “fact/value” and “norm/value.”
(A) The total speech act in the total speech situation is the only actual phe-
nomenon which, in the last resort, we are engaged in elucidating.
(B) Stating, describing, etc., are just two names among a great many
others for illocutionary acts; they have no unique position. (Austin
1975: 148)

2 The False Promise of Emotivism

The point of Austin’s critique of descriptivism appears clearly in his
account of promises, which obviously clashes with the standard accounts.
Promising is, for Austin as for Reinach, the “purest” speech act, the one
that most clearly challenges the descriptive model. According to Reinach,
social acts “are performed in the very act of speaking” (Reinach 1913/
1983: 36). They are in no way a report or expression of a state of affairs.
Here, Reinach makes a critique of Hume’s conception of the promise, for
to say “I promise” is not to describe something that one is in the process
of doing, it is promising.
125

Austin, Normativity, and Excuses 125

Promises are an example of what Austin calls explicit performatives, in
contrast with primary performatives. The former explicitly announce what
they do. This explicit character is necessary, in Austin’s view, to legal usage,
because in questions of law, a less explicit performative could be consid-
ered ambiguous and so vitiated. It would thus be a case of misfire (in his
classification, it is in the category misfires, misexecutions):
In the law . . . this kind of inexplicit performative will normally be brought
under B.1 or B.2 – it is made a rule that to bequeath inexplicitly, for instance,
is either an incorrect or an incomplete performance, but in ordinary life
there is no such rigidity. (Austin 1975: 33)

We do not have on the one hand the utterance, and on the other hand
the act that makes it true. Rather, the two form a “unity.” In other words,
the act is not a supplement (of whatever sort:  social, emotive, assertive,
etc.) to what is said, to a “p” that might be defined by a content, a prop-
osition, or a state of affairs. Reinach saw this point clearly about these
acts, “which do not have in words and the like their accidental, additional
expression, but which are performed in the very act of speaking” (Reinach
1913/1983: 36).
This essentially quite simple observation amounts to a questioning,
avant la lettre, of contemporary analyses of the performative as a proposi-
tion to which a “force” is added, illocutionary or otherwise. The general-
ization of the theory of performatives in the triad locution/perlocution/
illocution is here a source of misunderstandings  – a point to which we
will return. The point for now is that the radicality of the discovery of per-
formatives has not always been perceived since we have been tempted to
reduce performatives to something else (semantic, psychological, or social
conditions), not taking account of the fact that they are indeed acts.
From this point of view, the idea of the act as a “supplement” is insepa-
rable from the descriptive fallacy. If we recognize the nature of the act,
we must admit that it cannot be reduced to an assertion about a state of
affairs, or a report, or an expression or description of a psychological state.
All these linguistic forms are legitimate in their place, but none of them
can create an obligation.
It is interesting to note that Reinach, who first described the properties
of law in terms of the notion of an act, says that “promising is by no means
reducible to making known a decision of will” (1913/1983: 17). Put another
way, an internal resolution cannot bring about an obligation. The very def-
inition of normativity is in question here for Reinach. A descriptive utter-
ance cannot be normative. Informing someone else of one’s resolution,
6

126 Sandra Laugier

however strong it might be, does not amount to a commitment. The “con-
stative” description of an experience does not create an obligation, and it
is in this sense that it is not an act. In this way, the discovery of speech acts
is inseparable from a radicalization of the descriptive/normative distinc-
tion. This is paradoxical inasmuch as the legal uses of the notion of the
speech act tend, on the contrary, to put that distinction in question. The
distinction between descriptive utterances and normative ones is redefined
in terms of acts: there are no linguistic criteria for the normative (a conclu-
sion arrived at in different ways by Reinach, Wittgenstein, and Austin).
The same utterance can be either constative or normative (“The window
is open” can be understood as a description or an order, “I will be there
tonight” as a description or a promise or a threat). Its role cannot be deter-
mined through its structure (semantic or syntactic).
One could consider the utterance in question, following the perverse
use of the notion of force as defined by Frege, on the model F(p):  the
utterance is produced, in one case, with a purely assertive force (on the
model of the assertion sign), in the other case, with an illocutionary force.
But this would be to discount a point reiterated by Reinach and Austin,
namely, that the force and what is said are inseparable. The speech act is
“what is said” taken as a whole (Travis 1981): to understand what a speech
act truly is, one has precisely to understand that it is not distinguished sim-
ply by an “additional” force. That would be nothing but a psychological or
intentional ersatz, as pitiable as banging one’s fist on the table or (to take
an example of Wittgenstein’s) on one’s chest, in an attempt to shore up a
doubtful or insincere assertion.
The attempt to define the act as an additional force is a resurgence of
expressivism and emotivism, and Reinach as much as Austin takes on this
tradition, the rejection of which is necessary to any proper definition of
speech acts. This is the source of Reinach’s forceful critiques of Hume and
his “expressivist” conception of promising: the point is that the promise
is made, not expressed. We have here an early critique of a whole current
(that became highly influential during the twentieth century) of norma-
tivist moral philosophy, which includes expressivism and emotivism: on
this reading, an utterance like an ethical judgement, which Wittgenstein
described as nonsense, can be reintegrated into language by analyzing it
as a descriptive utterance associated with an emotion. Cavell has rightly
mocked this conception, which sees a moral or aesthetic judgment as a
factual proposition associated with an “aah!” of approbation or admiration
(Cavell 1979:  chapters 9 and 10). In light of this, it is interesting to com-
pare the conception of speech acts as it is elaborated by Reinach and Austin
127

Austin, Normativity, and Excuses 127

with the “emotive theory of ethics” that Ogden and Richards proposed in
their well-known The Meaning of Meaning (1923). This theory attempts to
distinguish two opposed functions within language, the symbolic, descrip-
tive function and the emotive function, which is “the use of words to
express emotive attitudes” (Ogden and Richards 1923: 149). Ethical utter-
ances, for them, have an emotive rather than a cognitive function; they
express a sentiment about a state of affairs. To say that X is good is not to
express knowledge about X, but my sentiment or attitude in relation to
X. It seems to me that it is precisely in opposition to this conception (con-
nected not only to the Humean tradition, but also to Russell’s concept of
a propositional attitude) that Austin sets out his notion of the speech act.2
According to some theorists of pragmatics, Austin’s discovery of speech
acts is to be situated within the development of this emotive theory. For
Recanati,3 for example, Ogden and Richards brought about a transition
from logical positivism to ordinary language philosophy. But this claim
is mistaken, and a look at Reinach’s work can help to clarify this point
by showing just how much the notion of the speech act is opposed to
any idea of emotive expression. Moreover, Ogden and Richards’ emotive
theory, like non-cognitivist theories in general, remains propositionalist
and so doubly subject to the descriptive illusion – both in the idea (which
returns in the twentieth century) of a force associated with a proposition,
and in the idea of the utterance as the expression of a desire or an inten-
tion. Again, Reinach was very clear about what he calls “the ‘bond’ result-
ing from promises”:  “We have found in promising an act all its own”
(Reinach 1913/1983: 26).
There has been a remarkable philosophical regression, losing the insights
of Austin and Reinach and returning to the idea, explicit in standard phi-
losophy of language, that the performative is equivalent to “publicly mani-
festing a certain intention,” an act that can then be validated or sanctioned
by social institutions. For Austin, understanding that the performative is
not descriptive is ultimately a question of ethics. We may be tempted, he
argues beginning his first lecture, to say that a performative, for example,
a promise, expresses an intention that would be definable or explicable
outside of the field of language (of semantics itself ). But for Austin, such
an interpretation would be not only mistaken but also “immoral.”
It is but a short step to go on to believe or to assume without realizing
that . . . the outward utterance is a description, true or false, of the

2
Wittgenstein also rejects this tradition, though in a different way. See Cavell 1969, 1979.
3
Recanati 1991.
8

128 Sandra Laugier

occurrence of the inward performance. The classic expression of this idea
is to be found in the Hippolytus (l.612), where Hippolytus says . . . “my
tongue swore to, but my heart (or mind or other backstage artiste) did
not.” Thus “I promise to. . .” obliges me – puts on record my spiritual
assumption of a spiritual shackle. It is gratifying to observe in this very
example how excess of profundity, or rather solemnity, makes for immo-
rality. For one who says “promising is not merely a matter of uttering
words! It is an inward and spiritual act!” is apt to appear as a solid moral-
ist standing out against a generation of superficial theorizers. . . . Yet he
provides Hippolytus with a let-out, the bigamist with an excuse for his “I
do,” and the welsher with a defense for his “I bet.” Accuracy and morality
alike are on the side of the plain saying that our word is our bond. (Austin
1975: 9–10)
What is important in Reinach and Austin is the fact that “our word is our
bond”:  the commitment is made in saying. The matter is not even any
more the de-mentalization and de-intentionalization of the speech act, but
the definition of the speech act, or better, its semantics.

3 Some Misfortunes of the Performative

One of the problems that the theory of the performative gives rise to is
in fact that of generalization, which tends to efface the initial performa-
tive/constative dichotomy. What interests Austin is, as he describes it, “the
total speech act in the total speech situation.” To arrive at this conclu-
sion, Austin uses arguments that are already present in Reinach. There is
no grammatical criterion by which the performative is distinguished; the
same utterance can be performative and constative. He seems, therefore, to
substitute for his first distinction, performative/constative, a second: locu-
tionary/illocutionary. “The locutionary act as much as the illocutionary is
an abstraction only: every genuine speech act is both” (Austin 1975: 146).
There are, then, in every utterance these three dimensions:  locutionary,
perlocutionary, and illocutionary. Or, more precisely, every utterance can
be considered as an act of each of these sorts. It is important to be precise
about this. There is a crude tendency in pragmatics to decompose the act
of speaking into three parts (often reduced to two; no one is interested in
the perlocutionary anymore except Cavell). The locutionary thus becomes
“the content or the proposition” (Recanati recognizes that this is not very
Austinian) and the illocutionary “not the content of the statement, but
what it is as act.” Recanati hence construes the remark of Austin’s cited
earlier, that “every genuine speech act is both,” as: “Every genuine speech
act includes both” (Recanati 1978: 119).
129

Austin, Normativity, and Excuses 129

Austin does not propose the locutionary/illocutionary distinction in
order to replace the performative/constative distinction. The two distinc-
tions are not exactly on the same level. The locutionary does not designate
the propositional dimension of the utterance, but rather the utterance as
seen under the aspect of the locutionary act. The point is not, for Austin,
to distinguish what is said from the fact that it is said, but rather to see
what is said as a whole, the complete act of speaking.
So what Austin affirms is the performative dimension of all language.
This leads to the fraying of the border between performative and consta-
tive, but not to its elimination. The generalization of the theory of per-
formatives is not a way for Austin to efface the performative/constative
distinction through an analysis of all utterances into corresponding “ele-
ments” or dimensions. It is a way to extend to “constative” utterances the
notion of action, and inextricably to extend to “performatives” the notion
of truth: to sketch a logic of performativity.
In order to see the character of Austin’s approach, we have to look out-
side the context of How to Do Things with Words and consider the articles
“Truth” and “A Plea for Excuses.” Austin’s idea again is to extend his ques-
tioning of the “true/false fetish,” not in the way of relativism or pragma-
tism, but by a double movement:  by extending true/false to apparently
performative utterances, and by extending felicity/infelicity to apparently
constative utterances.
That we do speak of a false promise need commit us no more than the fact
that we speak of a false move. “False” is not necessarily used of statements
only. (Austin 1975: 11)

In order to make clear what kind of normativity Austin is after, we must

examine such a conception of the false: the falsity of actions, or, we might
say, the logic of failures and successes. A well-known and central element
in Austin’s theory is the idea of the possible infelicities of a performative,
for which he offers a classification. Among the possible infelicities of a per-
formative, there are two main types: misfire and abuse (Austin 1975: 18).
Austin’s examples of misfire are famous:  I  christen an infant, or a ship,
without being qualified to do so, or in the wrong circumstances, or with
another name than the one intended, or I  christen a penguin. In these
cases, the act, for conventional reasons (of procedure), is null and void; it
has not been performed. His examples of the second category of infelicity,
the category of abuse, are less famous. In these cases, interestingly, the act
is performed, but it is hollow. They form the topic of the fourth lecture of
How to Do Things with Words.
0

130 Sandra Laugier

This interest in infelicities in speech acts is shared again by Austin and
Reinach. It has sometimes been observed that this is the real originality
of Reinach and Austin. Before them, those who thought about social acts
had not made the failure of such acts an explicit theme of discussion. It is
not discussed by Reid, for example, whose theory of social acts (or social
operations) might be mentioned as a distant precursor of our two authors.
The same is true of Hume’s famous discussion of promises. This is not
surprising: as Austin’s remark about Hippolytus shows, a “psychological”
or intentionalist interpretation excludes failure – even if it does not analyze
the promise as an “expression.” Benveniste, as well, was not interested in
failures and wants to exclude them from his theory; Searle and Recanati
are interested only in the conditions of validity of utterances (explained
either in terms of their social or institutional background, or through their
reflexive structure or representational content), and not in failures.
Yet it is crucial to understand, in reading Reinach and Austin, that only
a theory of failure or of infraction can explain the nature of social acts
and the species of practical reality they create. It is precisely the possibility
of failure that defines the speech act as an act, and that places the theory
of speech acts in the context of a theory of action. Moreover, it is this
aspect of the theory, much more than the pragmatic generalization, that
is relevant to law:  by the introduction of a connection between failures
and excuses, Austin introduces excuses in the framework of the theory of
speech acts, and therefore articulates it to a general redefinition of action.
“A Plea for Excuses” is inseparable from Austin’s theory of performatives,
and presents a series of failures and failed speech acts. Here is where the real
issue of generalization becomes clear: Austin aims to show that it is in the
nature of language to be able to fail, and not only by missing its object (rep-
resenting it incorrectly). Falsity is not the only way for language to break
down. It can fail not only by missing reality or truth; it can go wrong, as
Austin says, just like any other human activity. The possibility of failure is
not something that sets speech acts apart from other kinds of social acts;
rather, it is what defines action as such. In Reinach as well, this dimen-
sion of failure is inseparable from the definition of action as performance,
and he also classifies such failures. The failure of a promise is thus called
by Reinach a “pseudo-performance” (Scheinvollzug). Reinach, like Austin,
observes that a false promise (a promise made without the intention of
keeping it) is really a promise. The act is not null and void, but failed. The
theory of infelicity/failure is a way of responding to this problem: a false
promise is a pseudo-performance, that is to say, it tries to pass for the real
thing. Nevertheless, it is an act. Failures are thus pseudo-acts, in the sense in
131

Austin, Normativity, and Excuses 131

which Carnap’s pseudo-statements were nevertheless statements (“pseudo”
indicates here not falsity, but tampering or faking). These statements do not
succeed in saying something and fail to say anything whatever.
The permanent possibility of the failure of the performative marks lan-
guage as a human activity, happy or unhappy. However, and reciprocally,
by his insistence on failures, Austin returns – via a sea change that it would
be an error to neglect – to the question of the act, defined as what can fail,
can go wrong. Here we encounter the importance of the theory of excuses,
which deals with cases where I have “done wrong.” When does one give an
excuse for oneself, or excuse someone else’s behavior?
In general, the situation is one where someone is accused of having done
something, or . . . where someone is said to have done something which is
bad, wrong, inept, unwelcome, or in some other of the numerous possible
ways, untoward. Thereupon he, or someone on his behalf, will try to defend
his conduct or to get him out of it. (Austin 1957: 2; 1979: 176)
Excuses are the exact complement of failures:  it is when something has
been done badly, when the performance has failed, that one has recourse
to an excuse. Austin remarks that a theory of excuses is more useful than
a theory of justification, insofar as failures are more interesting than suc-
cesses. This observation can be extended to the domain of the normative
in general: it is the way we have of explaining or justifying our failures (bad
actions, etc.) that determines a norm’s mode of constraint. For Austin, the
existence of excuses is essential to the nature of human action. The sheer
variety of excuses reveals the impossibility of defining our acts in a general
way, other than in the detail and diversity of our modes of responsibility.
Action is thus defined not positively (Austin nicely shows the difficulty
of producing such a definition), but in terms of vulnerability, the pos-
sibility of transgression, and of failure. Action is precisely that which we
excuse, what we do not exactly do, as Austin puts it in the conclusion of
“Pretending,” where he speaks of a general project of describing the failures
and vulnerability of human agency: “the long-term project of classifying
and clarifying all possible ways and varieties of not exactly doing things,
which has to be carried through if we are ever to understand properly what
doing things is” (Austin 1958: 278; 1979: 271).

4 Extenuating Circumstances
Perhaps unsurprisingly, Austin cites the realm of law as a particular source
of inspiration in his treatment of excuses. His essay contains, as one of its
central examples, a long analysis of a legal case, Regina v.  Finney (Austin
2

132 Sandra Laugier

1957: 21f.; 1979: 195f.), through which he explores the issue of responsibil-
ity. The issue in the case is the determination of an excuse, in other words,
the sense in which what was done was not done voluntarily, deliberately,
etc., but rather inadvertently and without ill intention. The realm of the
excuse is not that of justification (intentions can excuse, but they cannot
justify). Austin remarks pertinently that philosophy is too concerned with
justifications and not enough with excuses, which are a much finer tool of
description on his account.
In two main ways the study of excuses can throw light on these fundamental
matters. First, to examine excuses is to examine cases where there has been
some abnormality or failure: and as so often, the abnormal will throw light
on the normal, will help us to penetrate the blinding veil of ease and obvi-
ousness that hides the mechanisms of the natural successful act. It rapidly
becomes plain that the breakdowns signalized by the various excuses are of
radically different kinds, affecting different parts or stages of the machinery,
which the excuses consequently pick out and sort out for us. Further, it
emerges that not every slip-up occurs in connection with everything that
could be called an “action,” that not every excuse is apt with every verb –
far indeed from it:  and this provides us with one means of introducing
some classification into the vast miscellany of “actions.” (Austin 1957:  6;
1979: 179–180)
Here again we can discern the influence of Hart – especially his analyses of
the “defeasibility” of legal concepts, inherent in the concept of ascriptivity,
a concept whose consequences in fact go far beyond those Hart himself
recognized.
It is tempting to suppose that this coupling of (failed) action and
excuses, which Cavell was the first to remark as characteristic of Austin, is
where Austin connects language and law. It is precisely from this point of
view, in the end still a limited one, that reflection on speech acts is inextri-
cable from reflection on the nature of legal norms. This kind of legal reflec-
tion is not only de-mentalized and de-psychologized, but perhaps even
de-moralized, despite the moralistic appeal of the slogan “Our word is our
bond.” The passage in the first lecture of How to Do Things with Words in
which Austin alludes to that slogan is itself focused on excuses, the (bad)
reasons we give for having not, e.g., kept a promise. From the beginning,
the discovery of speech acts is connected with this problematic of failure
and transgression, which is what constitutes its deepest connection with
the question of law, certainly a deeper one than its connection with the
bare notion of normativity.
The essay “A Plea for Excuses” poses exactly the same problem – the rela-
tion between what is done and what is said – as that of the performative,
133

Austin, Normativity, and Excuses 133

but this time starting from the side of action. Excuses are deeply embed-
ded in the human form of life. Furthermore, Austin says, the question of
excuses can help us in moral philosophy, if it can help us to get an idea of
“what is meant by, and what not, the expression ‘doing an action’ or ‘doing
something’ ” (Austin 1957: 4; 1979: 178). But we do not have such an idea.
Thus we must not forget that in the background of the theory of perfor-
matives there is a real perplexity about what it is to do something (how
to do. . .). In fact, we know nothing about it, and philosophers who have
reflected on this question have let themselves be taken in by the “myth
of the verb,” according to which there is some “thing,” “performing an
action,” which exhibits the essential characteristics of what we range under
the variable “perform an action.”
All “actions” are, as actions (meaning what?), equal, composing a quarrel
with striking a match, winning a war with sneezing: worse still, we assimi-
late them one and all to the supposedly most obvious and easy cases, such as
posting letters or moving fingers, just as we assimilate all “things” to horses
or beds. (Austin 1957: 5; 1979: 179)
For Austin, it is our excuses – what we say when it appears that we have
acted badly (clumsily, inadequately, etc.) – that allow us to gain a better
understanding of what an action is, or rather to begin to classify everything
we bring together under the general term, the “dummy” term, action. The
existence of excuses is in fact essential to the nature of human action  –
they do not somehow only come on the scene later; they are implied in
action itself. Action in this sense, Austin interestingly notes, is something
specifically human: the human vulnerability as defined by the “linguistic
constellation” of excuses.
In this way, Austin presents the complexity of human actions and of
their possible classifications in terms of excuses, which are the particular
objects of his study. We can get a sense of this through such familiar exam-
ples as killing a donkey:  “by accident” or “by mistake”? (Austin 1957:  11;
1979: 185) Austin’s example shows that there are differences between doing
something by mistake and doing it by accident, even though it often seems
that the two expressions are used as synonyms. But he also shows that by
starting from ordinary language – an infinite source of distinctions that
philosophical language has effaced – we can understand something about
the nature and classifications of our actions. Austin points out that we do
not give just any type of excuse for just any type of action. One can excuse
lighting a cigarette or covering one’s books by “the force of habit,” but a
killer cannot excuse his murdering “by the force of habit” (an example of
Austin’s reported by Pitcher). The diversity of excuses shows the diversity
4

134 Sandra Laugier

and variety of actions, and for any given excuse there is a limit to the acts
for which it will be accepted: what Austin calls norms of the unacceptable.
“We may plead that we trod on the snail inadvertently: but not on a baby –
you ought to look where you are putting your great feet” (Austin 1957: 20;
1979: 194). It is excuses – what we say when it appears that we have done
badly (clumsily, inadequately, etc.) – that help us to know what an action
is, to begin to classify and differentiate what we bring together under the
general term action. Excuses are essential to human action – they are in no
way a mere “after-thought.” The variety of excuses shows the impossibility
of defining agency otherwise than through the detail and diversity of our
modes of description and explanation, through our styles of performing
(or not quite performing) our actions.
Here the ontological point of the theory of excuses becomes clear – one
that is quite different from what is sought when we look for some social
entities created by performatives. Also the connection to our initial prob-
lem appears. Austin means to introduce a classification of actions, hence
a description of reality, through excuses: to create differences within real-
ity through immanent differentiation – the degrees and modalities of the
attenuation of agency – articulated by excuses. His colleague Hart described
this point well in his essay “The Ascription of Responsibility and Rights”:4
The sentence “Smith hit her” can be challenged . . . in two distinct ways.
Smith or someone can make a flat denial of the relevant statement of the
physical facts . . . Alternatively, any of a vast array of defenses can be pleaded
by Smith or his friends which, though they do not destroy the charge alto-
gether, soften it, or, as lawyers say, “reduce” it.
Thus, to “he did it” (he hit her), it may be pleaded:
1. “Accidentally” (she got in his way while he was hammering in a nail)
2. “Inadvertently” (in the course of hammering in a nail, not looking at
what he was doing)
3. “By mistake for someone else” (he thought she was May, who had
hit him)
4. “In self-defense” (she was about to hit him with a hammer)
5. “Under great provocation” (she had just thrown the ink over him)
6. “But he was forced to by a bully” (John said he would trash him)
7. “But he is mad, poor man.” (Hart 1949: 190–191)

It is interesting to see here that the classification of excuses is a description

of actions themselves as “missed.” We find the same coupling of failure
4
An essay that greatly influenced Austin, and later Erving Goffman. See Laugier 2009a.
135

Austin, Normativity, and Excuses 135

and excuse, of offense and compensation, in Goffman’s micro-sociological
analyses, notably in his examples of failures and inappropriate behavior –
as if, for Goffman, there were no conditions for successful behavior. To
bring this out we can read the following passage from Regina v. Finney,
as quoted by Austin, in a sociological spirit (or in the spirit of Goffman’s
Asylums. . .), and see how the structural dimension of excuse and failure in
social relations emerges clearly.
The Prisoner was an attendant at a lunatic asylum. Being in charge of
a lunatic, who was bathing, he turned on hot water into the bath, and
thereby scalded him to death. The facts appeared to be truly set forth in
the statement of the prisoner made before the committing magistrate, as
follows: “I had bathed Watkins, and had loosed the bath out. I intended
putting in a clean bath, and asked Watkins if he would get out. At this
time my attention was drawn to the next bath by the new attendant, who
was asking me a question; and my attention was taken from the bath where
Watkins was. I did not intend to turn the hot water, and I made a mis-
take in the tap. I did not know what I  had done until I  heard Thomas
Watkins shout out; and I did not find my mistake out till I saw the steam
from the water.” . . . (It was proved that the lunatic had such possession
of his faculties as would enable him to understand what was said to him,
and to get out of the bath.) . . . Verdict: Not guilty. (Austin 1957: 22–23;
1979: 196–197)

Attention to excuses and to the compensation due to others is the essen-

tial connection between Austin and Goffman, especially in Goffman’s later
major work, Frame Analysis:
When the individual visibly muffs a task, he can, of course, act as if nothing
wrong has occurred, obliging the witnesses to act accordingly. Alternatively,
he can stop his doing in midflow for a moment to offer an excuse or an
apology . . . to provide this sort of remedy for trouble is to demand of the
others involved that they suddenly accept the actor on a different footing,
that of a human being who can make mistakes in his carrying out of a spe-
cialized role. (Goffman 1974: 542–543)

Austin claims, recall, that a theory of excuses is much more useful than a
theory of justification, insofar as failures are more interesting, more dif-
ferentiable, than successes. This observation can be extended to the realm
of the normative in general: it is our way of compensating, explaining, or
justifying our failures (bad or misguided actions, etc.) that determines nor-
mativity’s mode of constraint, and, ultimately, allows us to give it a reality
in a very realistic (Austinian) sense of the word.
6

136 Sandra Laugier

5 Reality as Vulnerability
We shall conclude by trying to define the sense of reality involved here,
in a still sketchy way. “Performatives, if adequate to reality, are felicitous,
if not, then, in specific ways, infelicitous” (Cavell 1994:  81). One of the
goals of ordinary language philosophy is to determine the various ways in
which an utterance can be infelicitous, inadequate to reality, can miss its
goal. Austin, just like Goffman, aims to set out the conditions of felicity
of language as an ordinary practice, to make clear the vulnerability of our
usages, and to specify certain adequate tools of compensation (excuses,
compensations).
What is reality for Austin? The matter is settled, in a way, in Sense and
Sensibilia by Austin treating the question as ludicrous, or, better, spurious
(Austin 1962: 4). Austin still defines it, minimally, by agreement: he sets
aside the difficulty, so often invoked in philosophy, of “arriving at agree-
ment” on an opinion or a theory, in favor of another difficulty, that of
agreeing on a point of departure, on a datum: the agreement on “what we
should say when”: “Here at last we should be able to unfreeze, to loosen up
and get going on agreeing about discoveries, however small, and on agree-
ing about how to reach agreement” (Austin 1957: 9; 1979: 183).
Austin adds that this agreement is an “agreement on the way we deter-
mine a certain fact,” “on a certain way, one way, of describing and grasping
the facts.” The agreement must bear on the forms of description of what
is going on. Ordinary language cannot pretend to be the last word: it is
the first (Austin 1957: 11; 1979: 185). Agreement and discovery are possible
because ordinary language is a collection of differences and “embodies all
the distinctions men have found worth drawing, and the connections they
have found worth marking, in the lifetimes of many generations,” at once
subtler and solider than “any that you and I are likely to think up in our
arm-chairs of an afternoon – the most favored alternative method” (Austin
1957: 8; 1979: 182). It is this capacity to mark and inventory differences that
makes language an instrument of insightful perception: because reality is
made of these details and differences.
Further, the world must exhibit (we must observe) similarities and dissimi-
larities (there could not be the one without the other): if everything were
absolutely indistinguishable from anything else or completely unlike any-
thing else there would be nothing to say. (Austin 1950: 115; 1979: 121)
We can better understand, from this vantage point, the enigmatic pas-
sage in “A Plea for Excuses” where Austin excuses his speaking of linguistic
137

Austin, Normativity, and Excuses 137

phenomenology by appealing to the fact that a sharpened awareness of
words is a sharpening of our perception of realities:
When we examine what we should say when, we are looking again not
merely at words (or “meanings” whatever they may be) but also at the reali-
ties we use the words to talk about: we are using a sharpened awareness of
words to sharpen our perception of, though not as the final arbiter of, the
phenomena. (Austin 1957: 8; 1979: 182)
This theme of seeing differences and similarities (shared of course with
Wittgenstein) is characteristic of ordinary, natural realism (to take up an
expression used by Putnam [1999] with reference to Austin, James, and
Wittgenstein). The distinctions that ordinary language philosophy estab-
lishes, Cavell notes, are natural, real – taken from observation, not fabri-
cated like those of theorists and philosophers of language.
Too obviously, Austin is continuously concerned to draw distinctions, and
the finer the merrier, just as he often explains and justifies what he is doing
by praising the virtues of natural distinctions over homemade ones . . .
1. Part of the effort of any philosopher will consist in showing up differ-
ences, and one of Austin’s most furious perceptions is of the slovenliness, the
grotesque crudity and fatuousness, of the usual distinctions philosophers
have traditionally thrown up. Consequently, one form his investigations
take is that of repudiating the distinctions lying around philosophy –
dispossessing them, as it were, by showing better ones. And better not
merely because finer, but because more solid, having, so to speak, a greater
natural weight; appearing normal, even inevitable when the others are
luridly arbitrary; useful when the other seem twisted; real where the oth-
ers are academic; fruitful when the others stop cold. This is, if you like, a
negative purpose. 2. The positive purpose in Austin’s distinctions resem-
bles the art critic’s purpose in distinguishing works of art, namely, that in
this crosslight the capacities and salience of an individual object in ques-
tion are brought to attention and focus. . . . This is plainly different from
their entrance in, say, philosophers like Russell or Broad or even Moore,
whose distinctions do not serve to compare and to elicit differences but
rather, one could say, to provide labels for differences previously, somehow,
noticed. One sometimes has the feeling that Austin’s differences penetrate
the phenomena they record – a feeling from within which the traditional
philosopher will be the one who seems to be talking about mere words.
(Cavell 1969: 102–103)
The connection, or proximity between language and reality rests, then,
on the inventory of differences, not on some metaphysical “realism.” It
is in this sense that philosophy is fieldwork. Austin’s fundamental intui-
tion thus becomes clear, and is one that is developed more completely by
8

138 Sandra Laugier

Goffman: that language itself is something one has to perceive, as a con-
textualized, framed practice, and that it is as a practice that it “fits” or does
not. “Fit” thus designates for Austin a concept that is neither correspon-
dence nor even correctness; it designates the character that is appropriate,
proper to the utterance in the circumstances: it fits or it does not (Travis
2001). As Austin (1950: 124; 1979: 130) puts it, “The statements fit the facts
always more or less loosely, in different ways on different occasions for dif-
ferent intents and purposes.”
In his later work Wittgenstein also describes this inseparably social and
perceptual level in which agreement (Übereinstimmung) in language  –
human cooperation – is a matter of the mutual adjustment of action and
of felicity, as if objectivity and agreement in language depended on the
sheer contingency of le mot juste (the pitch, as opposed to the philosophi-
cal topos of the kairos): “Just think of the expression, and of the meaning
of the expression ‘the right word’ [das treffende Wort]” (Wittgenstein 1953,
II, xi).
Austin makes the possibility of finding an ordinary adequacy to the
world depend both on agreement and on “mapping the fields of conscious-
ness lit by the occasions of a word” (Cavell 1969: 100). It is founded on
the reality of language as a social activity of exchange with the world: con-
versation/conservation. Ordinary language is a tool, it represents inher-
ited experience and insight – its “acumen has been concentrated primarily
upon the practical business of life” (Austin 1957: 11; 1979: 133). In this way,
it is a tool for drawing distinctions. This brings us back to the classifica-
tion of actions in “A Plea for Excuses,” and to the distinction presented in
“Three Ways of Spilling Ink” between spilling it intentionally, deliberately,
and on purpose – the infinitesimal detail of human action in its capacity
for disaster.
If we reach this agreement, we shall have some data (“experimental” data, in
fact) which we can then go on to explain. Here, the explanation will be an
account of the meanings of these expressions, which we shall hope to reach
by using such methods as those of “Agreement” and “Difference.” (Austin
1966: 429; 1979: 274)
This is more than an incidental remark. Indeed, the work of agreement and
difference is the only way to define “harmony,” that is, the ordinary inti-
macy of words with the world that Cavell calls their “mutual interiority.”
Goffman (1974), explicitly alluding to Austin, connects aptness of percep-
tion of “what is going on,” access to reality (recalling a chapter of William
James’ Principles of Psychology entitled “The Perception of Reality”), and
139

Austin, Normativity, and Excuses 139

felicity of speech. In “Felicity’s Condition,” Goffman unites Austin’s con-
ditions of felicity with the condition of felicity of interaction, adding his
own discovery to Austin’s:
Nor (it should now be plain) was it right to define Felicity’s Condition
restrictively in terms of verbal acts. Speech need not figure even in a
reduced way for Felicity’s Condition to apply: the general constraint that
an utterance must satisfy, namely, that it connect acceptably with what
recipients have in, or can bring to, mind, applies in a manner to non lin-
guistic acts in wordless contexts. These acts, too, insofar as they can be
perceived by individuals in the vicinity, will have to be styled so as to pro-
vide evidence that their doer is engaged in something that perceivers find
understandable, even if they are not favored thereby. This paper has dealt
only with utterances, but some of what it has dealt with about utterances
could almost as well be dealt with about entirely nonlinguistic doings.
(Goffman 1983: 50–51)
Austin and Goffman share a definition of felicity as “normality,” and as
maintenance of the expressive order, the “order” present in the word ordi-
nary. The possibility of conversational failure reflects the vulnerability of
human action, and of the very form of ordinary life.
Through his emphasis on failure Austin shows the vulnerability of ordi-
nary human action, defined, just as the performative utterance, as what
can go wrong. Thus the pragmatic theme is reversed (the title How to Do
Things with Words was chosen by Austin for his William James lectures in
ironic homage to the pragmatist maxim): action is structured and regu-
lated by language’s capacity or tendency to go wrong.
“A wrong construction is put on things,” says Austin in Sense and Sensibilia,
mentioning “misreadings, mishearings, Freudian over-sights, etc.” There
is, again, no clear dichotomy between “things going right and things going
wrong: things may go wrong, as we really all know quite well, in lots of differ-
ent ways” (Austin 1962: 13, emphasis added).
Our lives, as Emerson would say, mix the casual and the casualty.
Things can permanently go wrong. Goffman, even without describing
the failures of interaction, defines the human character of action in the
same way: “Action will be defined analytically. An effort will be made to
uncover where it is to be found and what it implies about those places.
Wheresoever action is found, chance taking is sure to be” (Goffman
1967: 149).
Action signifies (analytically, one might say) that misfortunes lie in wait
for oneself and for others, and that we take risks (threats to oneself or oth-
ers): it means that our actions have circumstances, which make us, and
0

140 Sandra Laugier

reality itself, vulnerable. And this may well appear to be the very social
reality of speech acts.

The point is to see all of the human form of life as vulnerable, defined by a
constellation of possibilities of failure, of ways we have to make amends, of
strategies we can use to forgive or forget, to iron things out, to swallow our
difficult condition as beings of failure and rupture. Thus Goffman, in “On
Cooling the Mark Out” (1952), examines cases where we have to accom-
pany someone in the suffering of a radical social failure. Goffman’s analysis
of interactions gives their full place to disorders, turmoil, embarrassment,
shame, the stage fright of social interactions (Goffman 1967), encroach-
ments, intrusions, offenses, rips in the surface of “normal appearances.”
These things make us feel the fragility of the ordinary as what makes order
and they make us feel our vulnerability in the presence of others.
Whether crucial or picayune, all encounters present occasions when the
individual can become spontaneously involved in the proceedings and
derive from this a firm sense of reality. When an incident occurs and spon-
taneous involvement is threatened, then reality is threatened. . . . The min-
ute social system that is brought into being with each encounter will be
disorganized, and the participants will feel unruled, unreal, and anomic.
(Goffman 1967: 135)
The ordinary is reality as itself vulnerable – to others, to ourselves, and to
our mistakes.
Ordinary language philosophy, and the discovery of speech acts in par-
ticular, has been connected, as we have seen with Austin and Reinach, to
this problematic of failure, of transgression, and of the vulnerability of
social personhood. Goffman pursues a descriptive program of cataloging
differences and our perception of “what is going on,” following Austin’s
insight of the description of reality through the variety of excuses.
Cavell traces this vulnerability to reality back to our expressive body, as
Emerson puts it, “the giant I carry around with me.” Excuses, in their per-
manent recognition of human vulnerability, in the realm of what Austin
calls the “civil,” place the normativity of the ordinary in the realm of the
tragic (Cavell 1994). But Austin, though not as prone to emphasizing trag-
edy as Cavell would like, also envisaged the worst, what he defined as “the
unacceptable,” the inexcusable.
It is characteristic of excuses to be “unacceptable.” Given, I suppose, almost
any excuse, there will be cases of such a kind, or of such gravity, that “we
will not accept” it. It is interesting to detect the standards and codes we thus
invoke. . . . We can plead that we trod on the snail inadvertently: but not on
141

Austin, Normativity, and Excuses 141

a baby – you ought to look where you are putting your great feet. (Austin
1957: 20; 1979: 194)
Comparably, the activity of reparation of offence, for Goffman, is under-
taken in the context of this radical vulnerability, because the ordinary is
as manifest in small, everyday offences (the word of excuse that does not
come, the absence of attention) as much as in tragedy. Hence Goffman,
in Frame Analysis, takes up the theory of perception developed by Austin
in Sense and Sensibilia. The offense may be caused by mere thoughtless-
ness: not being attentive to the situation, to what is “really happening.”
Such a “miss” may be inherent to all social perception.
My perspective is situational, meaning here a concern for what an individ-
ual can be alive to at a particular moment. I assume that when individuals
attend to any current situation, they have the question: “What is it that is
going on here?” (Goffman 1974: 8)
The matter is to see “what is going on,” what the actual situation is.
Indeed, Austin himself discusses the cases where a wrong construction is
put on things: where there is no difference between misperceiving and mis-
behaving, or doing simply wrong. We miss reality, not because it is inac-
cessible, or away, but by lack of care and attention. Cavell says about the
importance of film:
It says that the perception of poetry is open to all, regardless as it were
of birth or talent, as the ability is to hold a camera on a subject, so that a
failure so to perceive, to persist in missing the subject, which may amount
to missing the evanescence of the subject, is ascribable only to ourselves, to
failures of our character; as if to fail to guess the unseen from the seen, to
fail to trace the implications of things – that is to fail the perception that
there is something to be guessed and traced, right or wrong – requires that
we persistently coarsen and stupefy ourselves. (Cavell 1984: 14)
Reality is itself vulnerable  – to failures of language, and of actions
altogether (see Laugier 2013:  chapter 9). A false perception of situations,
defined from the beginning by Austin as a moral matter, is the origin of
practical errors, which come back to put us in the wrong or to embarrass
us. The kind of realism we are looking for here involves, just like the mat-
ter of speech acts, both the linguistic and the practical. For Austin, the one
essential concern is lack of attention, thoughtlessness, inconsiderateness –
through failure at the stage of appreciation of the situation:
So too in real, or rather civilian, life, in moral or practical affairs, we can
know the facts and yet look at them mistakenly or perversely, or not fully
realize or appreciate something, or even be under a total misconception.
2

142 Sandra Laugier

Many expressions of excuse indicate failure at this particularly tricky
stage:  even thoughtlessness, inconsiderateness, lack of imagination are
perhaps less matters of failure in intelligence or planning than might be
supposed, and more matters of failure to appreciate the situation. (Austin
1957: 20; 1979: 193–194)
Goffman’s misframing (1974: 309) echoes Austin’s notion of mispercep-
tion here. His question “What is it that is going on here?” is an echo
of Austin’s “What should we say when . . . ?” Both Austin and Goffman
define a task for ordinary realism:  to describe what is going on, and to
heighten our sense of the vulnerability of “what is going on” (a reformula-
tion in ordinary language of the Wittgensteinian question, “what is the
case”) itself. “Thoughtlessness,” like abuse and infelicity, is a lack of atten-
tion to reality, of awareness of details, as well as of precision of speech or
commitment to promises. “Accuracy and morality alike are on the side of
the plain saying that our word is our bond” (Austin 1975: 10). Reality itself
is vulnerable to our attention and our agreements, to our misperceptions
and carelessness, and this vulnerability should be taken into account in any
definition of “realism.”
143

Ch apter   7

Berkeley and Austin on the Argument

from Illusion
Robert Schwartz

What we have above all to do is . . . rid ourselves of such illusions

as ‘the argument from illusion’ – an ‘argument’ which those (e.g.
Berkeley, Hume, Russell, Ayer) . . . have been most adept at working.
Austin, Sense and Sensibilia

Sense and Sensibilia (Austin 1962, hereafter abbreviated as SS) is perhaps

Austin’s most prominent work, and his attempt to undercut the Argument
from Illusion its central focus. For many, Austin’s attack on the Argument
was devastating and should have laid the issue to rest. Others, like Ayer
(1967), a prime target of Sense and Sensibilia, were unconvinced and
responded. They not only found Austin’s criticisms faulty, but believed
Austin had often misunderstood what the real issues actually were. Still, it
has been widely assumed that Austin won the day. Sense data and the prev-
alent foundationalist epistemologies that relied on them were not tenable.
More recently, there has been renewed interest in the Argument from
Illusion and with it further scrutiny of Austin’s position. The conclusion
many have drawn is that although Austin’s criticisms of sense data theories
hit the mark, deep epistemic and metaphysical issues remain concerning
the “objects of perception” that he either did not see, sought to avoid, or
did address, but unsuccessfully. Although these newer controversies are
not over the positing of sense data, they do depend on claims and assump-
tions central to the Argument from Illusion. In particular, they focus on
the problems and implications that follow from the fact that the very same
qualitative experience can be a veridical reflection of reality, an illusion, or
a hallucination. I will examine what I conceive would be Austin’s response
to this newer problematic at the end of this chapter.
At present, suffice it to say that I agree with critics that Austin did not
specifically address or provide a satisfactory treatment of current worries
over the nature of the objects of perception. But my reasons as to why he
is not doing that are not theirs. On the account I wish to offer of Austin’s
143
4

144 Robert Schwartz

project, his goal is not to solve problems concerning the objects of percep-
tion, but to dissolve them. To explain Austin’s arguments for adopting this
stance, I think it useful to explore Berkeley’s treatment of the Argument
from Illusion, he being another of Austin’s targets in Sense and Sensibilia.1
According to Austin, Berkeley played a major role in giving life to
the Argument from Illusion, and hence for the problems and misguided
philosophical enterprises it spawned.2 Unlike Ayer, however, Berkeley
did not have the opportunity to defend himself. I would like to respond
on Berkeley’s behalf, not by defending him from Austin’s attacks, but by
examining instead where they both agree. For I believe a comparison of
their positions can not only illuminate Austin’s ideas, it can lay to rest a
number of misunderstandings of Berkeley’s work that influence interpreta-
tions of Austin and continue to provide the background and context for
debates in the philosophy of perception.3

1 Illusions
Austin’s charge that Berkeley was responsible for the Argument from
Illusion and much of the mischief it did is very puzzling. For Berkeley
was not only bothered by many of the same worries as Austin, Austin’s
responses are quite in keeping with Berkeley’s own.4 By definition, illusory
experiences are taken to be deceptive. But, Berkeley asks, wherein lies this
supposed deceptiveness? For he agrees with Austin (SS, 11)  that “decep-
tion only makes sense against a background of general non-deception.” The
concept “misperception” presupposes and is meant to contrast with the
idea of “veridical perception.” Berkeley and Austin argue, however, that

1
All works by Berkeley are cited in the text from the nine-volume A. A. Luce and T. E. Jessop edition
(Berkeley 1948–1957), where they appear as follows (abbreviated titles used in the text are given in
parentheses):  An Essay towards a New Theory of Vision (NTV) and Philosophical Commentaries (PC)
in vol. 1; A Treatise Concerning the Principles of Human Knowledge (PHK), Three Dialogues between
Hylas and Philonous (TD) and Siris in vol. 2; Alciphron (Alc) in vol. 3. References in the text are by
work title and Berkeley’s section number for all works except TD (where references are by title, part
number, and page number) and Alc (where references are by title, dialogue number, and Berkeley’s
section number).
2
Austin makes critical remarks about Berkeley in other writings as well, but as in Sense and Sensibilia,
he says that he relies largely on Warnock’s (1953) reading of Berkeley. More recent readings of
Berkeley, which focus on his New Theory of Vision, tell a different story about Berkeley’s project. See
Atherton (1990) and Schwartz (1994).
3
For a recent book claiming that many of the mistakes fueling current controversies can be traced
back to Berkeley, see Searle (2015). See also Travis’ (2015) review of Searle’s book from an Austinian
perspective.
4
I am not claiming, though, that there are not other significant differences between Berkeley and
Austin’s views, or that Austin would see matters exactly as I do.
145

Berkeley and Austin on the Argument from Illusion 145

proponents of the Argument from Illusion have mistaken views about the
nature of illusions and in turn, the nature of perceptual veridicality.
Berkeley, like Austin, denies there are illusions as usually conceived, and
he has good reason for adopting this stance. It is a core claim of his theory
of vision that there is no such thing as the proper or the correct way an
object should look if perception is to be veridical. There are no necessary
correspondences or resemblances between the physical/tangible proper-
ties of objects and their visual appearances.5 He points out, for example,
that a tower far away may appear phenomenally small, but on approach
its appearance occupies more and more of the visual field. In his New
Theory of Vision (NTV, 61), Berkeley asks, “[W]hich of all these various
[visual] extensions is that stated, determinate one that is agreed on for the
common measure of other magnitudes?” He answers, “No reason can be
assigned why we should pitch on one more than another.” None of the
appearances is inherently correct, and therefore, none could be inherently
illusory or deceptive.
Consider, next, Berkeley’s analysis of the classic bent oar illusion.
A physically straight oar part in and part out of water looks bent. Hence,
it is usually assumed that its bent appearance must be deceptive. Berkeley
disagrees. In Three Dialogues (TD, III, 238), he says, “Thus in the case of
the oar, what he [a perceiver] immediately perceives by sight is certainly
crooked. . . . But if he thence conclude, that upon taking the oar out of
water he shall perceive the same crookedness; or that it would affect his
touch, as crooked things are wont to do: in that he is mistaken. In a like
manner, if he shall conclude from what he perceives at one station, that
in case he advances toward the moon or tower, he should still be affected
with like ideas, he is mistaken. But his mistake lies not in what he perceives
immediately and at present . . . but in the wrong judgment he makes con-
cerning the ideas he apprehends to be connected with those immediately
perceived.”
Berkeley’s account of the veridicality of color perception follows suit.
In response to an argument by Hylas, he says, “Upon this supposition
indeed, the objections from the change in color in a pigeon’s neck, or
the appearances of a broken oar in the water, must be allowed to have
weight. But those and the like objections vanish, if we . . . place the reality

5
In New Theory of Vision, Berkeley identifies the physical with the tangible; in Principles of Human
Knowledge, he explains that he did so mainly for rhetorical reasons. For present purposes, I leave this
issue aside, along with debates among Berkeley scholars about whether his position is phenomenal-
ist. I think the label inappropriate and discuss matters here and elsewhere that support this reading,
but I do not argue the point in this chapter.
6

146 Robert Schwartz

of things in ideas, fleeting indeed, and changeable; however not changed
at random, but according to the fixed order of Nature. For herein consists
that constancy and truth of things, . . . and distinguishes that which is
real from the irregular visions of the fancy” (TD, III, 258). According to
Berkeley, there is no uniquely correct way the color of the pigeon’s neck
should look, any more than there is the correct way straight oars should
look. Nonetheless, there are law-like regularities in Nature that link these
patterns of appearances.
The straight appearance of the oar out of water can mislead as to its
appearance in water, just as its appearance in water can mislead with regard
to its appearance when not submerged. The same holds for the color
appearances of the pigeon’s neck. It is a fact about the pigeon’s neck that
its color appears differently under varying viewing conditions. In turn,
there is no single appearance that correctly captures its “true” color, and no
appearance that is inherently deceptive. Its appearances under “ideal” con-
ditions can lead to mistaken expectations about its appearances under non-
ideal viewing conditions and vice versa. Both veridical and non-veridical
appearances are properties of the items we name and take to be oars or
pigeons. That we can be mistaken in what they lead us to expect does not
mean they are mere appearances. They are experiences of the objects we
perceive them to be.
Austin’s take on the issue is much the same. He rejects the conception
of the dichotomy between illusory and veridical perception that gives epis-
temic and metaphysical force to the Argument from Illusion. Similarly, he
traces the supposed problems to philosophers’ misunderstanding of the
nature of both illusory and veridical perception. Austin points out the con-
fusions that result from thinking that “If I gradually approach an object
from a distance I may begin by having a series of perceptions which are
delusive in the sense that the object appears to be smaller than it really is”
(SS, 45). He criticizes Ayer for suggesting that, since “a distant star has an
extension greater than the earth” but appears as “a silvery speck no bigger
than a sixpence,” it follows that the appearance is deceptive or that we are
seeing something other than the star (SS, 92). No single appearance is the
correct way any object’s size should look.
Austin challenges standard treatments of the bent oar/stick case, along
the same line. He asks, “Well now, does the stick look ‘bent’?” And answers,
“we can agree it does, we have no other way of describing it.” But, “What is
wrong, what is even faintly surprising, in the idea of a stick’s looking bent
sometimes? Does anyone suppose that if something is straight, then it jolly
well has to look straight at all times and in all circumstances? Obviously,
147

Berkeley and Austin on the Argument from Illusion 147

no one seriously supposes this. So what mess are we supposed to get into
here, what is the difficulty?” (SS, 29). That a stick looks bent in water does
not indicate that such experiences are non-veridical. Nor does it imply that
we are seeing something other than the stick. To appear “phenomenally
bent” when half submerged is as real a fact about a straight stick as is its
phenomenally straight appearance when not in water, or for that matter its
point-like appearance when viewed end on.
Austin’s treatment of color, too, has a Berkeleian ring. “Suppose . . . there
is a species of fish which looks vividly multi-coloured, slightly glowing per-
haps, at a depth of a thousand feet. I ask you what its real colour is. So you
catch a specimen and lay it out on deck, . . . the light is just about normal
and . . . it looks a muddy sort of greyish white. Well is that its real colour? . . .
In fact, is there any right answer in that case?” (SS, 65–66). “[T]he way
things look is, in general, just as much a fact about the world . . . as the
way things are” (SS, 43). There is, of course, no denying that perception
on occasion can and does mislead. For both Austin and Berkeley, though,
mistakes lie in what we make of appearances, not in the nature or inherent
qualities of appearances per se.6

2 Immediate Perception
It might seem, nevertheless, that Austin still has good reason to be critical
of Berkeley. After all, Berkeley holds that perception of distance, magni-
tude, shape, and color is not immediate, even when perception is veridical.
And it is this claim that Austin argues is responsible for many of the skep-
tical epistemic and metaphysical problems the Argument from Illusion is
thought to pose. For the idea that perception of the world is not immedi-
ate is usually taken to imply that we do not have direct access to Reality.
Our knowledge of the world is filtered through a subjective intermediary
that skeptics maintain precludes having justified beliefs or knowledge of
the world as it really is.
Austin argues that such worries rest on confusions, confusions primar-
ily over the meaning of words. Philosophical uses of the terms “direct”
and “immediate” tend to lack substantive content. The terms, however,
are not meaningless when employed in most everyday contexts. We do
have a satisfactory grasp in ordinary situations of when it is appropri-
ate to distinguish things or events observed directly from those observed

6
For an account along these lines of problems with the common understanding of illusions in phi-
losophy and vision science, see Schwartz (2012).
8

148 Robert Schwartz

indirectly, via intermediaries. For instance, if someone tells us how things
are, or if we see things in mirrors or on TV, our knowledge may be con-
sidered indirect. When so employed, however, the terms “direct” and
“immediate” have no real epistemic or metaphysical significance. By con-
trast, when philosophers move beyond the commonplace and give these
terms such weight, their use no longer has either a practical or theoretical
anchor. According to Austin, the epistemic and metaphysical puzzles phi-
losophers find so haunting continue to resist treatment, because they have
no point to begin with. They rest on accepting a direct/indirect dualism
he finds untenable.
For Austin, the direct/indirect distinction on which the Argument from
Illusion trades is a put up job. It has no explanatory role to play in an
account of perception or perceptual knowledge. Other than in contexts
where it serves to contrast with “indirectly,” the term “direct” has no use.
Thus, Austin argues, “it is essential to realize that here the notion of indi-
rectly wears the trousers” (SS, 15). The direct/indirect dichotomy has a
point in context, but it does not pick out two types of visual experience,
one providing direct access to the world and the other a problematic step
removed. Indeed, it is hard to make sense what this philosophical distinc-
tion can amount to.
Austin has a further reason for rejecting the idea of immediate per-
ception as then conceived by his opponents. He is convinced that foun-
dationalist accounts of knowledge that rest on claims of the certainty
or incorrigibility of reports of immediate experience are not viable. He
argues, in fact, that they cannot possibly have this exalted status. “Any
description of a taste or sound (or colour) . . . involves saying that it is
like one or some that we have experienced before: any descriptive word is
classificatory” and as such is “often uncertain and unreliable” (1946: 163;
1979: 92). All implications that flow from sense experiences are the result
of classification. “Sensa are dumb, and only previous experience enables us
to identify them” (1946: 169; 1979: 97). Sensa are not self-presenting; they
are not given to us classified into types. As Travis (2004: 57) comments, for
Austin “the senses are silent . . . Perceptual experience is not as such either
veridical or delusive. It may mislead, but it does not take representation to
accomplish that.”
It is not surprising, then, that Austin believes that in endorsing the idea
that perception of the world is not direct, Berkeley, like proponents of the
Argument from Illusion, falls into the skeptical traps it sets. And Berkeley
is in no position to step back from his claim that distance, size, shape,
and color perception are non-immediate. The claim is central to both his
149

Berkeley and Austin on the Argument from Illusion 149

theory of vision and epistemology. Berkeley denies, nevertheless, that his
theory of perception entails skeptical conclusions.
As Berkeley employs it, “immediate” is a technical term, not part of
ordinary discourse. He knows, for example, that most people believe that
when they open their eyes they perceive distance immediately, and he sees
no reason to dispute them. Yet, in the technical context intended, Berkeley
is correct in maintaining, “It is . . . agreed by all that distance, of itself
and immediately, cannot be seen” (NTV, 2). Berkeley, like most vision
scientists well into the twentieth century, adopted a two-stage model of
visual processing. Temporally, we first experience sensations, the qualita-
tive properties of a given sense. As a result of associative learning, they
come to trigger perceptions of what to expect. Sensations are experienced
immediately, solely as a matter of the physiological nature of the sense
organ stimulated. That which they bring to mind, perceptions, are thus
experienced mediately.7
In keeping with standard scientific use, for Berkeley, the proper objects
of perception are the qualitative properties of experience that serve to dis-
tinguish the senses. Color and light are the proper objects of sight, smells
of olfaction, sounds of hearing, etc. Sense organs are characterized by the
qualities of the sensations they give rise to. Distinguishing sense modalities
in this way does not require assigning their proper objects special epistemic
status, or assuming that what we experience immediately stands in the way
of perceiving the worldly objects we ordinarily think we see.
Since there is no way things should look, sensations lack assertive con-
tent; they have no inherent truth conditions. Sensations are neither veridi-
cal nor non-veridical “[S]ense, or soul as far forth as sensitive, knoweth
nothing” (Siris, 305). “[B]y sensible things I  mean those only which are
perceived by sense, and that in truth the senses perceive nothing which
they do not perceive immediately: for they make no inferences.” (TD, I,
174–175) In addition, we have neither certain knowledge of what is imme-
diately seen, nor are reports of our sensations generally accurate. “So swift
and sudden and unperceived is the transition from visible to tangible ideas
that we can scarce forbear thinking them equally the immediate object of
vision” (NTV, 145). As a result, “[W]e cannot without great pains cleverley
separate and disentangle in our thoughts the proper objects of sight from

7
See Hatfield and Epstein (1979) and Schwartz (1994) for further explication of this two-stage
model and its centrality in the history of vision science. See the selection of historical papers
on the issues in Part I of Schwartz 2004b, and more modern views on the issues in Part III of
Schwartz 2004b.
0

150 Robert Schwartz

those of touch which are connected with them. This, indeed, in compleat
degree seems scarce possible to perform” (NTV, 159).8
Some confusion in discussions of the issue results from an ambiguity in
the term “object.” The man on the street typically uses the term to refer to
the thing that is perceived, not to the technically speaking proper object of
perception. “Strictly speaking, . . . we do not [immediately] see the same
[proper] object that we feel. . . . But in case every variation was thought suf-
ficient to constitute a new kind or individual, the endless number or con-
fusion would render language impracticable. Therefore . . . men combine
together several ideas, apprehended by divers senses or by the same sense
at different times, or in different circumstances, but observed, however
to have some connexion in Nature, either with respect to co-existence or
succession; all which they refer to one name, and consider one thing” (TD,
III, 246, emphasis added).9
A cherry, for example, is an object/thing that feels, looks, sounds, smells,
and tastes thus and so. That is how it is perceived and characterized by the
mediate ideas associated with those immediately experienced. Still, how
a cherry appears immediately is a real property of the cherry, not a mere
appearance, an alien object standing between reality and us. It is part of
what it is to be a cherry. Of course, when we look at a particular cherry we
are never certain that we are seeing a cherry, or that the perception we have
of all its properties is accurate. This, though, is no cause for skepticism.
As Austin says, “The human intellect and senses are inherently fallible and
delusive, but not inveterately so. . . . It is futile to embark on a ‘theory of
knowledge’, which denies this liability; such theories constantly end up by
admitting the liability after all, and denying the existence of knowledge”
(1946: 169–170; 1979: 98).

3 Existence and the Real

Austin maintains that many skeptical arguments are also due to mistaken
views about the meaning of the terms “existence” and “real.” As in his dis-
cussion of “direct” and “indirect,” Austin argues that when the function of

8
Similar claims are found in Descartes and Locke. And well into the twentieth century, perception
theorists agreed with Helmholtz (1950: 6) that “We are not in the habit of observing our sensations
accurately.” Reports of the character of our sensations are not only corrigible, they are often false.
When Berkeley says that sensations are “perfectly known” (PHK, 87), his point is that there is noth-
ing more to be known about them than what is given by sense. He does not intend to rule out doubt
or mistaken reports.
9
The literature on Berkeley’s use of the term “object” is vast. For a review and response to several of
the more prominent interpretations of the term, see Rickless (2013).
151

Berkeley and Austin on the Argument from Illusion 151

these words is made clear, the epistemic and metaphysical conundrums the
Argument from Illusion supposedly raises vanish. He maintains that the
term “real” is not meaningless and can be employed usefully in everyday
contexts. Austin’s masterstroke is to show that in its ordinary use, “real” is
a contrast word.10
We do, for instance, contrast real cream with cream substitutes,
cardboard props that look like cream, cream hallucinations, and cream
holograms. According to Austin, independent of its occurrence in such
comparative contexts, the label “real” has no point. Each of these “creams”
is real in its way. On the other hand, when used properly as a contrast
word, the term “real” does not signal any metaphysically significant onto-
logical status. It does not serve to pick a particular kind or class of privi-
leged “real” objects. Thus there is no room for questions about whether
perception can put us in touch with the real or whether there is a Reality
perceivers cannot access. Used contrastively, talk of what is or is not real
could not possibly foster such philosophic skepticism.
What’s more, for Austin, the hallowed philosophical distinction between
so-called Realist and anti-Realist positions itself presents a false dichotomy.
People can differ about what things they believe exist  – cats, unicorns,
atoms, rainbows, quarks, Gods, demons, and the man on the moon. And
they can be right about some such claims and wrong about others. These
existence claims are not questions about what is real, but about what sound
inquiry does or does not lead us to posit. The practice of equating material
things with the real is foreign to this discourse. “For ‘material thing’ is not
an expression which the ordinary man would use.” (SS, 7) It is philoso-
phers who use the term “as designating in a general way the class of things
of which the ordinary man both believes and from time to time says that
he perceives particular instances.” (SS, 7) Apart from its use in everyday
contexts, the term has no clear meaning. The ordinary man typically talks
of seeing named things like “chairs, tables, pictures, books, flowers, pens,
cigarettes; the expression ‘material thing’ is not here (or anywhere else. . .)
further defined. But does the ordinary man believe that what he perceives
is (always) something like furniture or like these other ‘familiar objects’ –
moderately-sized specimens or dry goods?” (SS, 7–8). The ordinary man
also believes he sees “rivers, mountains, flames, rainbows, shadows, . . .,
vapours and gases. . . . Are all these ‘material things’?” (SS, 8). “Real” is a

10
See, in particular, chapter  3 of Sense and Sensibilia (SS, 20–31), and the section on “Reality” in
“Other Minds” (1946: 158–161; 1979: 86–89).
2

152 Robert Schwartz

contrast term. To identify the real with material things only breeds confu-
sion and bogus philosophical problems.
A person, of course, may be mistaken when pointing and saying, “That
is a pig,” but the possibility of being mistaken does not show the person is
not justified in asserting that he or she is in the presence of a pig. “We have
a pretty rough idea what pigs look like, what they smell and sound like,
and how they normally behave; and no doubt if something did not look
at all right for a pig, behave as pigs do, or make pig-like noises and smells,
we’d say it wasn’t a pig” (SS, 120–121). “We learn the word ‘pig’, as we learn
the vast majority of words for ordinary things, ostensively – by being told
in the presence of the animal, ‘That is a pig’; and thus, though certainly
we learn what sort of thing it is to which the word ‘pig’ can and can’t be
properly applied, we don’t go through any kind of intermediate stage of
relating the word ‘pig’ to a lot of statements about the way things look, or
sound, or smell” (SS, 121). We are not given necessary and sufficient condi-
tions for what it is to be a pig.
Moreover, the empirical meaning of “pig” is not fixed. Austin allows
there may be other ways of assessing claims about the presence of pigs.
“Well, perhaps there is somewhere, recorded by some zoological authority,
a statement of the necessary and sufficient conditions for belonging to the
species pig. And so perhaps, if we use the word ‘pig’ strictly in that sense, to
say of an animal that it’s a pig will entail that it satisfies those conditions.
But clearly it isn’t this sort of entailment . . . [that] is particularly relevant
to the use that non-experts make of the word ‘pig’ ” (SS, 120). In addition,
there is no good reason to assume, in general, that the non-experts’ usage
is incorrect, that only the experts’ categorization tells us what a “real” pig
is. Both classifications may be useful and in context correct. Furthermore,
situations may arise that call for additional ways to divide up the animal
kingdom.
A typical response to Austin is that although he might be correct about
ordinary language, philosophy is no more constrained by our ordinary
ways of talking about pigs and seeing pigs, than physics is constrained in
its theories by the masses’ use of the term “mass.” Austin, of course, does
not maintain that inquiry must be bound by the use of words enshrined in
the Oxford English Dictionary. There is no denying that physics has devel-
oped useful ways to talk about mass that outrun anything to be garnered
from ordinary language. Problems arise, however, when it is thought that
physics or zoology is privileged, and gets to tell us what a “real” pig is. As in
so many other areas of philosophy, problems in perception arise when the
153

Berkeley and Austin on the Argument from Illusion 153

terms “real” and “really” are tacked on to relatively unproblematic ques-
tions about what is known or perceived. History shows that doing so only
results in self-inflicted epistemic and metaphysical wounds.11
Berkeley’s analysis of the situation is similar. In Principles of Human
Knowledge, he says, “Nothing seems of more importance, towards erecting
a firm system of sound and real knowledge, which may be proof against
the assaults of scepticism, than to lay the beginning in a distinct explication
of what is meant by thing, reality, existence: for in vain shall we dispute con-
cerning the real existence of things, or pretend to any knowledge thereof,
so long as we have not fixed the meaning of those words” (PHK, 89). There
is nothing wrong with everyday talk about what is or is not real. “In the
sense here given of reality, it is evident that every vegetable, star, mineral
and in general each part of the mundane system, is as much a real being
by our principle as by any other. Whether others mean anything different
by the term reality than I do, I entreat them to look to their own thoughts
and see” (PHK, 36).
In Alciphron, Berkeley’s opponent asks, “Do we see anything at all, or is
it altogether fancy and illusion?” . . . “Do we not, strictly speaking perceive
by sight such things as trees, houses, men, rivers, and the like?” Euphranor,
Berkeley’s defender, answers: “We do, indeed, perceive or apprehend those
things by the faculty of sight” (Alc, III, 10). Empirical inquiry justifies
claims that we see vegetables, stars, minerals, etc. Berkeley’s point is that it
does not warrant positing material substance, eternal absolutes, and other
things that lie beyond the realities revealed in experience.12
For Berkeley, we are usually justified in believing that what we see is a
cherry. We know how cherries look, smell, and feel, as well as how they
respond if acted upon. If something seems not to accord with these ideas
and expectations, we would not say it is a cherry. Perceptual veridicality lies
in establishing links within and between what we perceive by vision, touch,
and the other modalities. Knowledge is a matter of finding in Nature cog-
nitively and behaviorally relevant empirical correlations. It lies in detecting
useful regularities among ideas, not in discovering correspondences to the
Real. The “aim is only to know what ideas are connected together; and the
more a man knows of the connexions of ideas the more he is said to know
the nature of things” (TD, III, 245).

11
The relationship of this analysis to more recent claims that water is “really” H2O should be obvious.
12
It should be noted that Berkeley countenanced positing things too small to see, such as atoms,
although he rejected certain theoretical posits, for instance, absolute space and some conceptions of
the ether. See Schwartz (to appear).
4

154 Robert Schwartz

“There is a rerum natura and the distinction between realities and chi-
meras retains its full force” (PHK, 34). The skeptical belief that we cannot
distinguish realities from chimeras results from identifying the real with
a metaphysical conception of matter. Thus, Berkeley warns, “Matter, or
material substance, are terms introduced by philosophers; and as used by
them . . . are never used by common people. [S]o long as the names of
particular things [e.g. Austin’s “pigs” “chairs,” “tables,” “pictures,” “books,”
“flowers,” “pens,” “cigarettes”] are retained, the word matter should never
be missed in common talk. And in philosophical discourse it seems the
best way to leave it quite out” (TD, III, 261). Matter, material substance,
and similar ideas are inventions of philosophers and have no real place in
explaining the nature of perception. Philosophical worries over skepticism
vanish if “we do not amuse ourselves with the terms absolute, external, exist
and such like, signifying we know not what” (PHK, 88).

4 Veridical Perception
It is a truism that true or veridical statements, beliefs, and perceptions
are those that agree with reality. Austin has no trouble understanding
versions of the claim, as it occurs in ordinary discourse. Difficulties arise
not in speaking with common folk, but in speaking with philosophers.
Philosophers believe they have a clear picture of the correspondence rela-
tion that supposedly holds between words and the world. They run into
trouble, however, when they go beyond everyday talk and assume it is
possible to spell out in a non-question-begging way what is meant by “cor-
respondence” and “reality.” Neither the idea of word/world agreement, nor
the idea of Reality can withstand scrutiny. Austin warns, “It is essential
to realize that ‘true’ and ‘false’ . . . do not stand for anything simple at all;
but only for a general dimension of being a right and proper thing to say
as opposed to a wrong thing, in these circumstances, to this audience, for
these purposes and with these intentions” (1975: 145).
For Austin, truth cannot lie in a correspondence relation of language
to Reality, if Reality is understood as some delineable domain of “real”
things. As rehearsed earlier, “real” is a contrast word, and what is real in
one context is not the real thing in another. It makes no sense to think
that there is a set of “real” things that populate or constitute Reality. The
world has no ontology; it simply is, and there is no unique way to describe
what there is. Different category schemes and characterizations of experi-
ence are legitimate. Once this is understood, the epistemic and metaphysi-
cal importance philosophers accord truth is out of proportion to its use,
155

Berkeley and Austin on the Argument from Illusion 155

especially in the context of perceptual experience. Austin indicates in sev-
eral places that truth is a philosopher’s fetish. Although in certain formal
contexts, some notion of correspondence truth may have a role to play,
in everyday use and in the practice of science, appeals to this austere view
of truth can go only so far. There are many ways to get things right, to fit
the facts, to accord with the realities we encounter, and they cannot all be
usefully treated as a matter of correspondence. “There is no neat dichot-
omy between things going right and things going wrong; things may go
wrong . . . in lots of different ways” (SS, 13).
In arguing against foundationalism, we saw that Austin claims sensa are
dumb; they are not self-identifying. They have no assertive content until
they are classified. “Uncritical use of the direct object after know seems to
be one thing that leads to the view that (or to talking as though) sensa, that
is things, colours, noises, and the rest, speak or are labelled by nature, so
that I can literally say what (that which) I  see: it pipes up, or I read it off.
It is as if sensa were literally to ‘announce themselves’ or to ‘identify them-
selves’, in the way we indicate when we say ‘It presently identified itself
as a particularly fine white rhinoceros’. But surely this is only a manner
of speaking, a reflexive idiom in which the French, for example, indulge
more freely than the English. . . . If we choose to say that [sensa] ‘identify
themselves’ . . ., they share the birthright of all speakers, that of speak-
ing unclearly and untruly” (1946: 169; 1979: 97; second emphasis added).
In fact, for Austin, nothing is self-identifying. Identity, or the individua-
tion of objects, depends on classification. “That things are similar, or even
‘exactly’ similar I  may literally see, but that they are the same I  cannot
literally see  – in calling them the same colour a convention is involved
additional to the conventional choice of the name to be given to the colour
which they are said to be” (1950: 116, n. 9; 1979: 122, n. 2). Ontological
questions make no sense independent of the schemes of classification we
adopt for the purposes at hand.
Austin dismisses the claim that we should take seriously the idea that we
may be deceived by an evil demon or that we are always in a dream state.
For him, the issue is not whether there is something beyond our skin and
mind, but determining what (i.e., that which) there is. As W. V. Quine
was wont to say, there is a ready answer to the question “What is there?”,
namely, “Everything.” But this answer does not tell us what things or sorts
of things there are. To answer “the real” or “Reality” is no more helpful. To
make a substantive ontic claim, we need to employ classificatory schemes
to individuate that which these things are. Classificatory schemes, though,
are made, not found in the world. We construct them to meet our evolving
6

156 Robert Schwartz

needs, intellectual and practical, and language is the primary, but not the
only, tool we use to do the needed cutting, pasting, and grouping.
According to Austin, we employ different classificatory schemes for dif-
ferent purposes and in different contexts. “There is no one thing that we
‘perceive’ but many different kinds, the number being reducible if at all by
scientific investigation and not by philosophy” (SS, 4). Nor is there any
point assuming the concepts we employ aim to reflect “natural kinds” or
divide up the world at its inherent or essentialist joints. Objects grouped
together as the same on one account are not thought to be similar on
others. We saw that, as the term “pig” is used in everyday contexts, the
sentence “I see a pig” may, according to Austin, be true. Used according to
scientific zoological criteria, it may be false.
In How to Do Things with Words, Austin (1975: 143–144) offers a his-
torical example of the same phenomena. After presenting a brief account
of a well-known military event, he asks about the truth or falsity of the
sentence, “Lord Raglan won the battle of Alma.” Austin then goes on
to explain that it “may be judged true in [the context] of a school book,
[but] may not be so judged in a work of historical research.” Of course,
in both the pig example and that of Lord Raglan, it is necessary to keep
the conflicting claims apart. The assumed background schemes of clas-
sification that determine our assessment of their truth-values cannot be
put in play at the same time. What’s more, it is a mistake to insist that
only one member of the conflicting pair can be true, that only one can
reflect the way the world is. True sentences may be said to correspond to
the facts, as long as we recognize that what the facts are depends on the
classificatory schemes we adopt and the context in which the sentences
are uttered.
Berkeley, too, thought that ontological questions did not make sense
independent of the schemes of classification brought to bear. Berkeley
summarizes his position thus: “Knowledge, or certainty, or perception of
agreement of Ideas – as to Identity & diversity & real existence, Vanisheth;
of relation, becometh meerly Nominal; of Coexistence, remaineth”
(Philosophical Commentaries, 739). In Three Dialogues, he says to his critic,
“you superadded to your idea of the house the simple abstracted idea of
identity, whereas I did not; I would tell you I know not what you mean
by that abstracted idea of identity . . . Are you not yet satisfied, men may
dispute about identity and diversity, without any real difference in their
thoughts and opinions, abstracted from names?” (TD, III, 248). The iden-
tity of things is not found in the world independent of the names we
invent to group and individuate them.
157

Berkeley and Austin on the Argument from Illusion 157

Berkeley’s discussion of number is of a piece. Counting requires a speci-
fication of units or principles of individuation. “That number is entirely
the creature of the mind . . . will be evident to whoever considers that, the
same thing bears a different denomination of number, as the mind views
it with different respects. . . . Number is visibly relevant, and dependent on
men’s understanding” (PHK, 12 and NTV, 109). To ask how many things
or objects there are in reality simplicter has no coherent answer. Identity
conditions or criteria of individuation are not found in the world, but are
imposed on experience by the use of ideas or words. “Words are of arbi-
trary imposition and . . . men are used to apply the word ‘same’ where no
distinction or variety is perceived” (TD, III, 247).
For Berkeley, “An idea can be like nothing but an idea” (PHK, 8 and 90).
It makes no sense to compare an idea with the inarticulate world, as it is,
or as it is given, independent of our understanding. We test our ideas by
how they fit in with other ideas, not by how well they mirror the world.
And, like Austin, Berkeley allows that seemingly incompatible viewpoints
can both be true. In describing the difference between ordinary seeing
and looking through a microscope, Berkeley says, “A microscope brings
us, as it were into a new world: It presents us with a new scene of visible
objects, quite different from what we behold with the naked eye” (NTV,
85, emphasis added).
The immediate objects of experience in both cases are light and color.
They are the objects of experience that distinguish sight from our other
senses. “[I]n the case of microscopical eyes I  see only this difference, to
wit, that upon ceasing of a certain observable connexions betwixt the div-
ers perceptions of sight and touch which before enabled us to regulate
our actions by the eye, it would now be rendered utterly unserviceable
for that purpose” (NTV, 86). Thus “what seems smooth and round to the
touch may to sight, if viewed through a microscope, seem quite otherwise”
(NTV, 105). Both descriptions are correct. There is no unique way to per-
ceive or have beliefs about the surface that captures how it “really” is.
To appreciate Berkeley and Austin’s pluralist account of ontology, it
may be helpful to look at a problem raised by the physicist Eddington
(1928: ix–x). He pointed out that a table is solid at the everyday macro level
of description, although it is composed of more empty space than material
stuff at an atomic level of description. The question, which he and many
philosophers and scientists pondered at the time and still do, was whether
the ontologically real table is the one characterized as solid or the one
characterized as non-solid. Although there was much disagreement about
which it should be, it was widely thought it must be one or the other.
8

158 Robert Schwartz

By contrast, ontological pluralists challenge the presumption that there
can be only one option. They hold there are at least two useful ways of talk-
ing, each occurring within different organizational or categorical schemes,
one every day, and one scientific. Both table descriptions are correct, and
the facts they state may be correlated. Nonetheless, we cannot in the same
breath assert that the table is solid and not solid without falling into con-
tradiction. It makes sense, is true, in one context (or universe of discourse)
to say the table is smooth, in another that it is not.
In his otherwise quite positive review of The Concept of Mind, Austin
chides Ryle for what he calls Ryle’s doctrine of one world, “the belief that
he has to show that only one world exists” (1970:  49). More generally
Austin asks, “Yet what has ever been gained by this favourite philosophi-
cal pastime of counting worlds? And why does the answer always turn
out to be one or two, or some similar small, well-rounded, philosophi-
cally accepted number? Why, if there are nineteen of any thing, it is not
philosophy?” (1970:  48). Austin embraces pluralism. For him, there are
as many realities or worlds as there are useful, insightful ways to organize
experience. There is no need to show that they all can be reduced to physics
or some other physical base. It is also most unlikely it can be done. Failure
to appreciate this fact is responsible for a good deal of the problems with
philosophical accounts of truth, correspondence, and reality.
Questions concerning what the table “really” is, or what it is to per-
ceive the “real” table, cannot be answered, because they cannot be usefully
asked. Through inquiry, we may discover correlations that link the two
descriptions, but there is no epistemic or metaphysical reason to privilege
one description over the other. To call either the macro or the micro table
the real table is not to use the term “real” to make a substantive con-
trast. The distinction is neither to be found in every-day nor in scientific
discourse.
Austin’s account of truth is complex, nuanced, and controversial. It is
not possible to go into the details here. What I think is clear is that his anal-
ysis of truth and the problems explicating the “correspondence to Reality”
idea put stress on specifying the satisfaction or truth conditions that serve
to distinguish veridical from non-veridical perceptions. If no appearance is
inherently correct or incorrect, if perceptual truth is not a matter of resem-
blance or “mirroring,” and if there are alternative, but conflicting, correct
characterizations of that which is seen, what is the nature of the perception/
world agreement relation that veridicality claims presuppose? And what
does it add to the story to be told that veridical perceptions correspond to
reality, when the term “Reality” does not refer to anything specific?
159

Berkeley and Austin on the Argument from Illusion 159

Without plausible, non-question-begging answers to these questions, it
is hard to make sense of the claim that we are directly or indirectly in con-
tact with Reality, and it is hard to spell out what constitutes a Realist versus
an anti-Realist response. Austin, as Berkeley, argues that when we become
clear about thing, reality, and existence, the Argument from Illusion “really”
is unsound and poses no genuine skeptical problem that needs an answer.13

5 Austin and the New Problem

As mentioned at the start, while many hold that Austin’s analysis showed
convincingly that sense data could not be the objects of perception, they
continue to think it important to find out what is. And here Austin’s crit-
ics find him of no help.14 As Martin (2007: 27) says, to address these cur-
rent concerns “would seem to take us far from Austin’s official stance . . .
[it] requires us to take far more seriously the argument from illusion than
Austin does.” A  brief summary of the new problematic goes something
like this.
Perceiving is a relational activity holding between perceivers and what is
seen, i.e., the “object” of perception. Most people engaged in these debates
also admit that, phenomenally speaking, it is impossible for a person to tell
on the basis of introspection whether an experience is veridical, illusory,
or the product of a hallucination. But this seems to imply that what is
perceived directly, the object of perception, must be the same in all cases,
“mere” appearances, not the external world. On the other hand, percep-
tions seem to be transparent. It seems natural and unproblematic to claim
we perceive objects like chairs, cherries, and pigs directly or immediately.
So, at least in the case of veridical perception, its object must be something
material, not a mental intermediary.
Illusions, though, are a tricky case since by definition they are non-
veridical, and thus do not correspond to Reality. Yet, as in veridical percep-
tion, the object of perception of an illusion is typically something material.
We are just mistaken about the properties we attribute to it. Perhaps, the
correct account of matters, then, is to hold that the objects of perception
of veridical and illusory experiences are the same, the real thing, while

13
In Schwartz (2016), I examine related problems in the theory of vision that plague attempts to
specify the criteria for distinguishing veridical from non-veridical perceptions.
14
For claims that Austin’s work does not deal adequately with current concerns see, for example, Thau
(2004) and Martin (2007). Although not always concordant with the analysis I give here and in
Schwartz (2004a), for discussions of the continuing relevance of Austin’s ideas to the philosophy of
perception, see, for example, Putnam (1994) and Travis (2004).
0

160 Robert Schwartz

hallucinations are the outlier. Either solution, however, is problematic.
For, given that the appearances of veridical, illusory, and hallucinatory per-
ceptions are indistinguishable introspectively, it seems obvious that what is
seen immediately must be the same in all three cases. If, however, they are
indistinguishable, the immediate objects of perception will be intermedi-
aries that block our having direct perception of the world.
One response to this challenge is to claim that qualitatively identical
appearances need not be of the same type. Appearances are not only quali-
tative states, they have content. If appearances that look identical can have
different content, they need not be considered the same mental state. The
objects of perception in veridical, illusory, and hallucinatory experiences
can be distinguished. Of course, allowing that appearances have content
does not explain how it is possible for phenomenally identical appearances
to differ in their content.
A prominent solution to this problem is similar to one some philoso-
phers adopt in epistemology to distinguish mental states that are “mere”
beliefs from those that constitute knowledge. Phenomenally speaking the
beliefs are indistinguishable, but according to this “hybrid” model, the
factivity of true beliefs is constitutive of their nature. One is a mental state
of knowing; the other is “mere” belief. Likewise, the content of introspec-
tively indistinguishable appearances need not be the same, if they are typed
according to their causes or some other external factor.
One popular version of this approach is the so-called disjunctivist solu-
tion. Its name reveals its approach. The content of appearances is disjunc-
tive. The objects of veridical and illusory experiences are the same, the
physical objects that trigger them. In the case of hallucinations, the situ-
ation is different. The object of hallucinatory experiences is not due to
environmental stimuli. Rather it is solely a figment of the imagination.
Typically we take perception at face value, and the content of appearances
puts us in direct touch with the outside world. We do on occasion make
mistakes as in the case of illusions. Or ever so rarely we go wrong because
what we are experiencing is a hallucination. The possibility of illusions and
hallucinations poses no real epistemic problem since there are recognized
steps we can take to correct ourselves, should we have reason to doubt our
perceptions are veridical.
Critics reply that such disjunctive theories cannot solve the epistemic
problems. Their reasons for rejecting these solutions vary. A good many
argue that they simply cannot make sense of the claim that qualitatively
identical appearances should be treated differently. It is an a priori truth
that they must be typed the same. A number of others allow that while
161

Berkeley and Austin on the Argument from Illusion 161

indistinguishable beliefs may be typed differently, appearances are not the
same sort of mental state as beliefs. Appearances do have representational
content, but their content is non-conceptual. They do not represent the
same sort of worldly, mind-independent facts beliefs are supposed to report.
Finally, there are those critics who insist that appearances are purely quali-
tative states. They simply have no representational content at all. Hence,
disjunctivist and similar solutions that attribute content to appearances are
nonstarters. Champions of this view are among those willing to allow that
appearances are intermediaries and we do not have direct access to reality.
The epistemic gap between appearances and veridical perceptions is not
breached. Another route is needed to block skepticism.
Obviously, it is impossible here to examine the pros and cons of these
alternative positions. The issues are too complicated and convoluted to
explore in detail. What’s more, the options are not as simple as I  have
sketched them. There are multiple versions of all of these strategies.
And there are as many conflicts within each camp as there are between
them. There is even much disagreement among parties to the dispute as to
what the problem actually is. The literature on the topic is vast, and grow-
ing by the day.15
Fortunately, I do not think it necessary to enter into the fray in order to
give what I believe would be Austin’s response to current controversies over
the objects of perception. I think Austin would find such talk a bundle
of confusions, resting on misguided distinctions that cannot be substan-
tively cashed out. And, I  believe, he would argue his case much along
the lines explored in his challenging the claims associated with the older
discussions of the Argument from Illusion. I agree, though, with Martin’s
(2007:  29)  claim that “there is no simple matching between Austin’s
account and any of these contemporary views. For to flesh [them] out . . .
one has to go beyond the concerns that Austin himself thought appropri-
ate, and also find deeper concern within the tradition of debate about per-
ception than Austin was prepared to do.” Austin’s analysis of where things
stand would, nonetheless, be different from Martin’s. Austin would see no
reason to go beyond his expressed concerns, because he believes there are
no deeper issues worth exploring.
The term “object of perception” is not problematic when used in ordi-
nary contexts. If asked about what we see, answering “a pig,” “a statue of

15
More details on the nature of the problem and alternative solutions can be found in Martin’s paper,
as well as in Byrne and Logue (2009), Haddock and Macpherson (2011), Macpherson and Platchias
(2013).
2

162 Robert Schwartz

a pig,” “an animal of another species,” or “I do not know” are perfectly
good responses. At the same time, “object of perception” is not a technical
term in the scientific study of perception. Nor is there an obvious need to
develop one for such purposes. “Object of perception” is a philosophical
term, based on a slogan, “to see is to see something” – a slogan lacking a
home in common sense or vision theory.
Moreover, the tripartite division of visual experiences into the categories
“veridical,” “illusory,” and “hallucinatory” is itself not a well-defined clas-
sification scheme.
There are, for example, a large number of misperceptions that are not
treated as illusions by anyone. If judgments of size, distance, shape, and
color are evaluated with respect to how they correspond to scientific physi-
cal measurements, they are not very accurate. Misperceptions abound;
appearances mislead. Yet these mistakes typically are not labeled “illusions.”
If physical measurement, however, is not the standard by which veridical-
ity is assessed, what is? As Austin (1946:165; 1979: 94) points out, we can
almost always turn inaccurate judgments into veridical ones “if we take
refuge in a sufficiently rough description” of what counts to get it right.
Perceiving the yardstick to be four feet long is a mistake. Alternatively, if
four feet plus or minus one foot is taken to be accurate enough, the assess-
ment changes.
Defining hallucinations as cases of perception not having a material
thing as its object is also problematic. Whether due to mental disor-
der or drugs, seeing the trees in the backyard as animate and ready to
attack is likely to be called a “hallucination,” not an illusion. Yet in such
cases we are seeing physical objects. Austin has been criticized as being
inconsistent in his characterization of the objects of hallucinations.16 For
instance, sometimes he seems to claim that “seeing” a nonexistent oasis
in the desert is triggered by environmental factors. Sometimes he seems
to say that such experiences are entirely of our own making. I prefer to
think Austin’s response to such criticism would be that either the matter
is to be settled by science, or nothing significant hangs on the decision.
The fact is, while dividing experiences into veridical, illusory, and hallu-
cinatory may serve a purpose in everyday contexts, most philosophical
classifications of appearances do not divide them up into explanatory
scientific “natural kinds.”
More generally, the affective meaning of perceptual experience is to a
large extent constituted by the expectations and behavior triggered, not

16
See Thau (2004).
163

Berkeley and Austin on the Argument from Illusion 163

by categorical, determinate judgments of that which it is. No harm is
done allowing that perception also triggers belief. And no harm is done
accepting the claim that perception “P” is true or veridical, if and only
if P.  Nevertheless, this does not explain in an insightful way what it
means for veridical perceptions to correspond to Reality or even to have
propositional-like satisfaction or truth conditions.17
In the Stanford Encyclopedia of Philosophy entry on disjunctivism about
perception, the author makes clear that it is not easy to differentiate and
evaluate theorists’ numerous varying views on the topic, because their
accounts presuppose conflicting analyses of knowledge, representation,
mental states, reality, veridicality, inference, consciousness, content, mind-
independence, and much more (Soteriou 2014). As canvassed earlier in this
chapter, Austin has much to say about these and related ideas. In Sense and
Sensibilia and elsewhere, he argues that most philosophical treatments of
them rest on dualist assumptions he finds wanting. Faulty understanding
of the nature of “knowledge, the real, and reality” is responsible for under-
writing skeptical worries. Bogus distinctions between direct and indirect
perceptions lead to fruitless debates over perceptual inference and what
stands in the way of what. Discussions of representation rest on incorrect
views of the nature of concepts, semantic relations, and the functions of
language. And there is no settled epistemic or metaphysical account of the
mind-dependent/mind-independent distinction that is both plausible and
has useful work to do. Over and over again, Austin warns that the firm
intuitions and a priori commitments that underlie these distinctions are
not to be trusted.

6 Conclusion
According to Austin, the myriad assumptions relied on to get Argument
from Illusion controversies off the ground have no foundation in scientific
research. Nor do they conform to useful common-sense talk. Austin ques-
tions whether speculation on the issues serves a real purpose. The “deep”
problems, old and new, the Argument from Illusion supposedly poses for
philosophers are largely of their own making, and his primary aim is to
dissolve them. In doing so, Austin does not wish to come down on the side
of either Realism or anti-Realism. It is not that Austin is unable to decide
where the truth lies, rather he thinks the very distinction is specious.
Attempting to locate his position on the standard philosophical grid of

17
For support of these claims, see Schwartz (2016).
4

164 Robert Schwartz

options is to miss the point. In the final paragraph of Sense and Sensibilia,
Austin says the main lesson to be drawn is that “The right policy is not . . .
[that of ] trying to patch it up and make it work properly; that just can’t be
done. The right policy . . . is to dismantle the whole doctrine before it gets
off the ground” (SS, 142). I think Berkeley would agree.18

18
I wish to thank Margaret Atherton, David Rosenthal, and Savas Tsohatzidis for helpful comments
and criticism.
165

Ch apter  8

Austin on Perception, Knowledge, and Meaning

Krista Lawlor

1 Interpreting Austin on Perception

Austin’s Sense and Sensibilia (1962) generates wildly different reactions
among philosophers. On the one hand, some allow that the text offers
acute criticisms of the argument from illusion for sense data, but see little
further value in the work.1 Some dispute that the lectures achieve even this
much, and claim that Austin and sense data theorists simply talk past each
other.2 On the other hand, some have decidedly positive reactions but dif-
fer over the text’s main purpose: some see far-reaching ramifications for the
philosophy of perception;3 others see the work as a prime instance of an
ordinary language philosopher offering us therapy;4 while still others find
a substantive anti-skeptical agenda supported by complex argumentation.5
Philosophers will disagree of course, but the extent of disagreement about
Austin’s contribution is remarkable, with the main arguments, methodol-
ogy, and the whole point of the lectures under dispute.
Interpreting Austin on perception starts with a reading of his Sense and
Sensibilia, and interpreting Sense and Sensibilia, I believe, requires reading
into the lectures key ideas from Austin’s work on natural language and the
theory of knowledge. The lectures paint a methodological agenda, and
a sketch of some first-order philosophy, done the way Austin thinks it
should be done. Crucially, Austin calls for philosophers to bring a deeper
understanding of natural language meaning to bear as they do their tasks.
In consequence Austin’s lectures provide a fascinating start – but only a
start – on a number of key questions in the philosophy of perception. It is
easy to read either too little or too much into them; finding the right bal-
ance reveals an important view of perceptual knowledge.
1
Martin 2007; Snowdon 2014.
2
Robinson 1994; Thau 2004.
3
Putnam 1994; Travis 2008.
4
Locatelli 2014.
5
Leite 2011.

165
6

166 Krista Lawlor
My plan for this chapter is this: in the next section, I provide a brief
synopsis of Sense and Sensibilia.6 In the section following that, I discuss
the picture Austin sketches of the role of philosophers in rationally recon-
structing commonsense epistemology, and how this picture makes urgent
an account of the situation-dependent meaning of our utterances. I then
turn to Austin’s distinctive contribution to questions about the metaphys-
ics and epistemology of perception. I will argue (i) that he provides a frame-
work in which to address some of the central questions of contemporary
philosophy of perception, and (ii) this framework requires appreciation of
the situation-dependence of utterance meaning. In the conclusion, I speak
to the wider import of Austin’s lectures.

2 A Synopsis of Sense and Sensibilia

I think it useful to give a brief synopsis of the lectures. (Note that chapter
headings are not Austin’s but are to serve as convenient tags.)

2.1 Chapter I. Introduction

Austin announces he aims to examine the sense data theorist’s reasons for
holding the doctrine that “direct” perceptual awareness is always of sense
data and never of “material objects.” His concern is that the facts of percep-
tion are more diverse than the sense datum theorist allows. Austin does not
aim to defend an answer to the problem of perception7 – he won’t defend
direct realism against the sense data theory, for instance. Austin adds that
a “blinkering philosophical English” in use by sense data theorists – chiefly
A. J. Ayer and H. H. Price – is a distinct philosophical mistake; it feeds our
ability to ignore or distort the facts of perception.8

2.2 Chapter II. Deception, Directness, and Commonsense

Epistemology
Austin aims to show that Ayer’s attempted recapitulation of ordinary or
commonsense epistemology is inaccurate, and he suggests that the mis-
characterization is driven by an epistemological agenda. (We will hear
more about the agenda in chapter IX.) Ayer represents ordinary thinking

6
Unless otherwise indicated, parenthetical page references in the text are to that work.
7
Crane and French 2015.
8
Price 1932; Ayer 1940.
167

Austin on Perception, Knowledge, and Meaning 167

about perception as claiming, first, that we are deceived if and only if we
don’t perceive “material objects,” and, second, that we are merely “satis-
fied” with the extent to which our perceptual beliefs are justified. Austin
takes issue with both claims: First, common sense shows itself keenly inter-
ested in when perception is direct or indirect – on a natural construal of
those terms – and has it that not all cases of being deceived by the senses
are of the same kind. Second, commonsense epistemology is more than
merely “satisfied” with perceptual justification – in fact, the idea that one
might question ordinary perceptual claims made under paradigmatically
good circumstances is nonsense. (Austin will return to perceptual incor-
rigibility in chapters IX–X.)

2.3 Chapter III. Opening Charge Against the Argument from Illusion
The argument from illusion proceeds in two stages: first, that sometimes
what we are directly aware of is sense data, and second, that we are always
directly aware only of sense data. Ayer’s argument at the first stage is by
cases. Austin examines what Ayer says about each of his cases, and cannot
find reason to abandon ordinary ways of thinking about the perceptual
facts. A stick in water looks bent (we might say that for want of a better
description), though it doesn’t look exactly like a bent stick out of water.
Austin asks, must something straight look straight at all times and on all
occasions? He notes we readily acknowledge that the same thing looks
different to us, depending on a wide range of circumstances. Any premise
that it must be a different thing we see on every occasion of looking differ-
ent flies in the face of common sense.

2.4 Chapter IV. Looks, Appears, Seems; Meaning in Context

The chapter is an interlude from the argument from illusion in which
Austin details some important aspects of commonsense epistemology.
Austin details the way common sense finds it important to distinguish
the way a thing looks (or the look of a thing) from the way it appears or
seems. For instance, we might comment on someone’s looks, “he looks
guilty,” but reserve the use of “he appears guilty” for special occasions,
where we want to withhold judgment. We might use “he seems guilty”
when evidence of guilt is mounting, whether or not he has the look of a
guilty person. To talk about how things seem is already to express a judg-
ment (43). Austin cautions: “There is, of course, no general answer at all to
the question how ‘looks’ or ‘looks like’ is related to ‘is’; it depends on the
8

168 Krista Lawlor
full circumstances of particular cases” (39). So too, the circumstances of an
utterance help determine its meaning (41).

2.5 Chapter V. The Second Step of the Argument from

Illusion: Qualitative Difference
Austin briefly reconstructs the argument for the second step. Ayer and Price
both claim that (i) “veridical” and “delusive” perceptions are qualitatively the
same.9 Next (ii) recall that according to step 1 of the argument, in the “delu-
sive” case what we are aware of is a sense datum. But (iii) if what we are aware
of in the “veridical” case is very different in its nature – a material object and
not a sense datum – then one would expect experience to register a differ-
ence, and distinguish the two perceptions. But by (i) we do not distinguish
these perceptions, so by (iii) what we are aware of in both cases is the same –
a sense datum. Austin’s central point against the argument targets premise
(iii): it doesn’t follow from the fact that X is different from Y in nature that X
looks different than Y does. “If I am told that a lemon is generically different
from a piece of soap, do I ‘expect’ that no piece of soap could look just like a
lemon?” (50).

2.6 Chapter VI. Ayer’s Evaluation of the Argument from Illusion

Austin recaps Ayer’s position. Ayer himself is not convinced by the argu-
ment from illusion, at least if it aims to demonstrate that we are always
aware of sense data. In Foundations of Empirical Knowledge, he is moved
by the objection Austin notes above: why shouldn’t things very different in
nature look very much alike? According to Ayer, the argument from illu-
sion is not best understood as answering a question of fact – namely, what
in fact is it that we are directly aware of in perception? Rather, the conclu-
sion of the argument answers a question about how we might talk – what
is the best language for describing perceptual experience? Austin rejects
Ayer’s idea that philosophers have the linguistic freedom presupposed
by Ayer’s question10 (59). For instance, he suggests that philosophers are

9
Austin notes that Price’s use of this premise is different than Ayer’s, as Price takes himself to already
have established the doctrine of sense data and is concerned at this point with the question of
whether sense data are parts of the surfaces of objects.
10
Austin notes that Ayer’s officially tolerant stance about choices of linguistic framework is belied by
his actual view of what is in fact fundamental. Although Ayer represents his own defense of sense
data theory as not a theory about the nature of perception, but a choice of linguistic framework,
in fact, he insists that there are hard facts about sense data – they are what really exist, and though
169

Austin on Perception, Knowledge, and Meaning 169

constrained in what they mean by “real shape” by what is ordinarily meant
when people talk about the real shape of something. The point is pursued
in the next two chapters.

2.7 Chapter VII. “Real”

Prompted by “frequent and unexamined occurrences of ‘real’, ‘really’, ‘real
shape’ &c.” in the arguments, Austin gives an excursus on uses of the word
“real.” The word “real” is “highly exceptional . . . in . . . that, unlike ‘yellow’
or ‘horse’ or ‘walk’, it does not have one single specifiable, always-the-same
meaning” (64). It is “substantive hungry”  – we must be able to specify
some respect or other in which a thing is real, and have an answer to
the question “A real what?” (68). Further, negative uses are basic: knowing
what it is to be “a real duck” requires knowing specific ways of being not a
real duck (70). A thing that is not a real duck is not thereby immaterial or
nonexistent (68). And what is the shape of a real duck, anyway? The ques-
tion has no answer (67). Austin’s aim is to make us wary of any argument
that our perceptions must not be of a thing because they “fail to reveal its
real shape.”

2.8 Chapter VIII. Ayer on What is “Real”

Austin considers Ayer’s last chapter of Foundations in which he attempts
to “furnish an explanation of the use of the word ‘real’ as it is applied to
the characteristics of material things” (80, quoting Ayer). Ayer’s basic idea
is that those sense data that are “privileged” present the “real qualities” of
things, where what makes a sense datum privileged is its predictive value –
sensing a datum from middle distance gives one a better chance of predict-
ing the data one will sense from other distances. Austin complains that the
resulting judgments about “real” qualities don’t square with commonsense
judgments. (“That’s not the real color of his hair  – he dyes it” is not a
statement about the lack of predictive value of the relevant experience.) In
summary, Austin diagnoses Ayer’s failure as owing to a more fundamental
failure to attend to issues about natural language: it is fatal “to embark on
explaining the use of a word without seriously considering more than a
tiny fraction of the contexts in which it is actually used” (83).

we may choose to speak as if there were material things, this is a choice made for convenience (60).
Austin returns again to Ayer’s views about material things and their relation to sense data in chapter
IX, especially 106ff.
0

170 Krista Lawlor

2.9 Chapter IX. Ayer’s Argument for Sense Data

Ayer suggests that we may choose the sense datum language, which involves
a special philosophical sense of “perceive” on which “what is seen must really
exist and must really have the properties it appears to have” (102). Austin
protests: What would motivate choosing such a language for describing
the empirical world? We couldn’t use it to talk about “material things” as
they don’t always have the properties they appear to have. Ayer’s official
stance is that the new language has the advantage of avoiding ambiguities
found in natural language. But Austin reminds us Ayer has been unable to
document any such ambiguities. Austin concludes that Ayer’s (and Price’s)
real motive is to produce a species of statement that will be incorrigible
(103). Here we find out what Austin was alluding to in suggesting that an
unstated “agenda” lay behind the arguments of the sense datum theorists.

2.10 Chapter X. Sense Data Theory and the Epistemological Agenda

In the longest chapter of the book, Austin delves into the epistemologi-
cal agenda driving Ayer’s and Price’s sense datum theory, and into details
about commonsense epistemology.
Ayer (and to an extent Price) is attracted to two key epistemic claims: first,
that statements about sense data form a special class of incorrigible “obser-
vation sentences,” and second, that such sentences can serve as conclusive
evidence for other statements, and so serve as an epistemic foundation.
Austin argues that no such privileged class of sentences will be found, and
he notes that Carnap agrees, although for the wrong reasons, supposing
that the point holds because what counts as an “observation sentence” is
conventional; Austin argues that the point holds because sentences are the
wrong type of thing to privilege in the first place (111). No kind of sentence,
once uttered, is incapable of being amended or retracted (112). However,
Austin notes, a particular utterance made in particular circumstances may be
in fact incorrigible (114). (To have a name for it, I will hereafter call this de
facto incorrigibility.) The possibility of de facto incorrigibility is a tenet of
our commonsense epistemology, according to Austin.
What about the second claim, that some sentences serve as conclusive
evidence for others? In addition to reiterating his point about sentences
being the wrong type of thing to privilege, Austin identifies several key
commonsense epistemic principles about evidence. First, not all state-
ments require evidence for their support. Sometimes plain sight reveals the
presence of something – you see a pig in a pen – and “there is no longer
171

Austin on Perception, Knowledge, and Meaning 171

a question of collecting evidence . . . I  can now just see that it is” (115).
Second, statements about “material objects” may sometimes be conclusively
verified; it is not the case that conclusive verification is only secured for a
more privileged class of statement (117): suppose you see a pig, and that
the situation is one where further verification is called for. In that event,
you can prod the pig, or do some further biological tests. At a certain point
in your investigations, you have done quite enough (118) to conclusively
verify that it is a pig. You don’t have to rule out every statement whose fal-
sity is “entailed” by the claim that it is a pig (123). Austin goes on to claim
that this marks a distinctive problem for foundationalism. He has in mind
only foundationalisms built on distinctive sentence types (122–123).
We will return to de facto incorrigibility and conclusive verification in
what follows.

2.11 Chapter XI.
This chapter is devoted to critical discussion of places in Warnock’s book,
Berkeley,11 where Warnock accepts the “two-languages doctrine,” with
an “evidence-language” and a “material-object language” (142) along
Ayer’s lines.

3 A Picture of Philosophical Practice

Austin gave his lectures – later published as Sense and Sensibilia – in 1958
at the University of California, Berkeley. According to Searle, he was
delighted with his time in America and was eager to spread the word about
his vision of how to do philosophy.12 It is natural to imagine that he saw an
opportunity to demonstrate the virtues of his approach to philosophical
practice.
This approach is portrayed very clearly in the lectures. He simultaneously
exhibits and talks about his preferred philosophical methods throughout.
He gives an extended statement about best practices for philosophers in
chapter VII, after discussion of Ayer’s suggestion that we consider the
adoption of a sense data language:
most words are in fact used in a particular way already, and this fact can’t
be just disregarded. . . . it is advisable always to bear in mind (a)  that the
distinctions embodied in our vast and, for the most part, relatively ancient

11
Warnock 1953.
12
Searle 2014.
2

172 Krista Lawlor
stock of ordinary words are neither few nor always very obvious and almost
never just arbitrary; (b) that in any case, before indulging in any tampering
on our own account we need to find out what it is that we have to deal with;
and (c) that tampering with words in what we take to be one little corner
of the field is always liable to have unforeseen repercussions in the adjoin-
ing territory. And we must always be particularly wary of the philosophical
habit of dismissing some (if not all) the ordinary uses of a word as ‘unim-
portant’, a habit which makes distortion practically unavoidable. (62–63)
The picture Austin paints in Sense and Sensibilia is of ordinary people
with a natural set of concerns and questions about perception – when it
works, how it justifies, and so on. Moreover, ordinary people have answers
to many such questions, which form a commonsense theory that makes
use of important distinctions, and includes significant commitments and
explanations. Philosophers have a role in making our ordinary theoreti-
cal commitments explicit, and to find out about those commitments they
must attend to the ways ordinary people talk. Philosophers also have a
role in rationally reconstructing those commitments (that is why ordinary
language is not the last word):
Certainly, when we have discovered how any word is in fact used that may
not be the end of the matter; there is certainly no reason why, in general,
things should be left exactly as we find them; we may wish to tidy the
situation up a bit, revise the map here and there, draw the boundaries and
distinctions rather differently. (63)
If philosophers are to help sort out ordinary commitments, they must be
very careful of how those commitment are revealed by the language ordi-
nary people speak. That is to say, philosophers need a theory of natural
language meaning.
Although Austin does not provide this theory, he thinks a central feature
of it will be the situation-dependence of utterance meaning. Austin stresses
this point over and again. What we mean is always determined in the par-
ticular circumstance of utterance (16, 41, 111). At the end of chapter IV, he
pleads with philosophers to pay attention to the situation of the utterance:
it is not enough simply to examine the words themselves; just what is meant
and what can be inferred (if anything) can be decided only by examining
the full circumstances in which the words are used. (41)
Uttering a sentence on a given occasion involves a form of words that in
another circumstance would have a very different meaning or significance:
Consider, ‘That cloud is like a horse’ and ‘That animal is like a horse’. In
the case of the cloud, even if we had said it was exactly like a horse, we
173

Austin on Perception, Knowledge, and Meaning 173

should not have meant that one might easily mistake it for a horse, suc-
cumb to the temptation to try to ride it, &c. But if an animal is said to be
like a horse, then probably it might in some circumstances be mistaken for
a horse, someone might think of trying to ride it, &c. (41)
And most pointedly:
the question of truth and falsehood does not turn only on what a sentence
is, nor yet on what it means, but on, speaking very broadly, the circum-
stances in which it is uttered. (111)
That the truth or falsity of what we say does not depend on sentence mean-
ing alone, but is determined by sentence meaning in circumstances, is a
key insight of Austin’s. We will return to this insight and see how it informs
Austin’s positive contribution to a philosophical account of perception.

4 Austin’s Substantive Results: The Metaphysics

of Perception
Austin does not develop situation semantics in the lectures. For that we
must look elsewhere in his work.13 Consequently, his substantive contribu-
tions to the theory of perception can only be partial, by his own lights.
Nonetheless, Austin seeks to provide some compelling evidence that his
approach will be fruitful, by sketching some substantive claims. In this
section and the next, I discuss the substantive claims Austin makes about
the metaphysics and epistemology of perception.
The argument from illusion generates what we today call “the problem
of perception.” This metaphysical problem starts with a question: “What is
the direct object of perception?” Austin portrays common sense as reject-
ing this question. We directly see many different things, and they are not
of one kind, “material” as opposed to “immaterial,” say – or if these objects
are of one kind, it’s not the philosopher’s business to a priori decide what
that kind is: “There is no one kind of thing that we ‘perceive’ but many dif-
ferent kinds, the number being reducible if at all by scientific investigation
and not by philosophy” (4). Is commonsense theory of perception a sort of
naïve or direct realism, then? Austin adamantly is not arguing in favor of
realism either. His point is that “What is the direct object of perception?”
is a bad question: “So we are not to look for an answer to the question,
what kind of thing we perceive” (4).

13
A central resource is his paper “Truth” (Austin 1950; 1979: 117–133).
4

174 Krista Lawlor
This might lead us to suppose that Austin wants to turn away from
metaphysical problems of perception altogether. This impression is forti-
fied by Austin’s brief remarks at the end of chapter III. There he acknowl-
edges that Ayer’s case of mirages is a better case for the sense datum theorist
to use, but then proceeds in rapid succession to suggest that mirages are
none too like the “normal” case, and not very frequently encountered, and
anyway there will likely always be qualitative differences between experi-
encing a mirage and seeing an oasis. To make the matter of interpretation
even thornier, this last suggestion is one he goes on to repudiate explicitly
(52). All this suggests a dismissive and haphazard approach to the meta-
physics of perception.14 But it also suggests a dismissive approach that is
without a solid foundation: Austin can seem insufficiently sensitive to the
genuine worries behind the argument from illusion. For even if we do
reject “What is the direct object of perception?” as somehow a bad ques-
tion, certainly we can frame a reasonable question about the metaphysics
of perception this way: given that on some occasions one can seem to be
in perceptual contact with an oasis, say, when there’s no oasis before one,
what constitutes one’s perceptual state in the case when the oasis is there as
the seeing of an oasis?
My reading of Austin is that he does not at all ignore this last metaphysi-
cal question about perception, however badly he advertises his position. To
understand Austin’s position, it is vital that we register that he has a pre-
ferred way of addressing questions about the nature of a target phenom-
enon. His method is to answer such questions by first considering relevant
contrasts. So for instance, when we wonder about the nature of intentional
action, Austin has us consider all manner of kinds of attempted or almost-
action.15 When we wonder about the nature of seeing, similarly, Austin has
us consider various kinds of attempted or almost-seeing. In chapter IV,
Austin details the way we distinguish varieties of “almost seeing,” noting
that common sense finds it important to distinguish the way a thing looks
from the way it appears or from the way it seems (39–43). Through such
observations, we get an initial sense of what makes seeing different than

14
Austin emphasizes his negative program: “What we have above all to do is, negatively, to rid our-
selves of such illusions as ‘the argument from illusion’ ” (4). It is no wonder we find readers thinking
the lectures hold little more than negative critique. A related complaint is that the arguments for
sense data theory that Austin criticizes are just bad – bad in simple ways that require no special
methods or meta-philosophical approach to rebut them (Martin 2007). This complaint misses
the fact that Austin’s positive contribution in critiquing Ayer is different in kind – Austin is not
merely rebutting Ayer’s argument, but also demonstrating the ways that we can fail to ask the right
questions.
15
See the papers “A Plea for Excuses,” “Ifs and Cans,” and “Pretending” in Austin 1979.
175

Austin on Perception, Knowledge, and Meaning 175

various cases of “almost seeing.” What we find is that the relevant contrasts
are differentiated by a variety of factors, such as our readiness to make a judg-
ment, or the role of evidence in our judgment. For instance, seeing that F
is distinct from being in a state where it appears that F, in that we are ready
to judge that F in the former but not the latter case.
Let’s call this the “method of relevant contrasts.” The idea is that we
ought not try to answer the question “what constitutes one’s perceptual
state in paradigmatically normal cases as a seeing?” bare as it were. Rather,
we look at a range of nearby perceptual states, and try to determine what
it is that differentiates them from each other. We ask, “What constitutes
one’s perceptual state in paradigmatically normal cases as a seeing F, as
opposed to its only appearing to one as F?”
Note here that, in keeping with Austin’s philosophical methodology,
philosophers must look to ordinary ways of talking about the differen-
tiation of contrasting cases. That in turn means that philosophers must
consider a wide variety of concrete instances where we talk about the kinds
of almost-seeings. Don’t ask about what differentiates seeing F from its
appearing to one that F, but ask, of particular circumstances, what dif-
ferentiates seeing the pig there from its appearing that there’s a pig there?
Concrete instances are important, as Austin’s nascent theory of meaning
dictates that it is only with such circumstances in view that philosophers
can discern what we mean and what we are committed to by the utterances
that express our judgments. The method of relevant contrasts requires
attention to the situation in which our statements are made, if it is to
deliver reliable results.

5 Austin’s Substantive Results: The Epistemology

of Perception
Austin makes fascinating remarks in Sense and Sensibilia about the epis-
temology of perception, and to understand his contributions, we must
integrate them with his remarks about knowledge and knowledge claims.
The result is a position that speaks to some central preoccupations of con-
temporary epistemology of perception.
In what follows, I will confine myself to three tasks: first, I’ll sketch the
way I reconstruct Austin’s position on knowledge and knowledge claims
and show how this position makes sense of Austin’s position in Sense and
Sensibilia concerning de facto incorrigibility.16 Then, I’ll discuss his position

16
Lawlor 2013, 2015.
6

176 Krista Lawlor
concerning conclusive verification. Finally, I’ll briefly describe the upshot: a
distinctive position regarding the kind of reasons one has in virtue of see-
ing something.

5.1 Knowledge and Knowledge Claims

Most philosophers are familiar with the idea that Austin lays the ground
for a relevant alternatives account of knowledge on which knowledge is
true belief backed by evidence or reasons sufficient to rule out all the rel-
evant alternatives.17 An alternative is, roughly, anything the truth of which
would defeat one’s knowledge claim. What makes an alternative relevant? I
think it best to think of Austin as suggesting a reasonable person standard
applies: an alternative is relevant if a reasonable person worries about it
enough to become unwilling to give an assurance about the truth of the
target proposition. This standard of relevance best rationalizes our practice
of giving and evaluating assurances. Visiting the farm with a friend who
asks, “Are you sure that’s a pig?” you assure her, “Oh yeah, I know it’s a pig.
I worked for years on a pig farm.” An assurance is the speech act of vouch-
ing for the truth of a claim, wherein one takes responsibility for having
reasons to believe the claim, which reasons also tell against all reasonable
alternatives. What one will actually have vouched for with one’s assurance
depends in part on the meaning of the word “knows,” and in part on the
circumstances in which assurances are made. Recall here Austin’s insight:
the question of truth and falsehood does not turn only on what a sentence
is, nor yet on what it means, but on, speaking very broadly, the circum-
stances in which it is uttered. (111)
The meaning of “knows” might be simple and circumstance invariant,
roughly, has conclusive reason to think the target belief is true. But what one
means in the broader sense  – the truth conditions of one’s utterance of
“I know” given in an assurance – depends on the circumstances. What it
takes to have conclusive reasons depends on what the alternatives are, and
this is fixed by the circumstances. The alternatives that must be eliminated
for one’s knowledge claim to be true are all and only the reasonable alterna-
tives, and what is reasonable to worry about depends on the circumstances.
If one is on a small family farm in North America, it is not reasonable to
worry about whether it might be a peccary. The reasonable alternatives are

17
See his paper “Other Minds” (Austin 1946; 1979:  76–116). See also Sense and Sensibilia (1962:
118, 123).
177

Austin on Perception, Knowledge, and Meaning 177

that the creature is a cow or a goat, say. Those are all the alternatives that
need ruling out, and one’s visual experience of a pig is enough to rule them
out. If in the circumstances it is a pig, then one’s knowledge claim is true.
The resulting account of knowledge is fallibilist, where that means one’s
reasons or evidence can be sufficient for knowing but logically consistent
with the falsity of the known proposition.18 This is a feature of relevant
alternatives accounts of knowledge. Adding to this account an Austinian
situation semantics for knowledge claims, we can resolve some key dif-
ficulties that fallibilist theories confront. I  haven’t space to demonstrate
the point fully here, though I  will sketch how situation semantics aids
epistemology.
I want next to briefly develop how this account of knowledge and
knowledge claims integrates with observations that Austin isolates for dis-
cussion in his chapter X, namely, the issues of de facto incorrigibility and
conclusive verification. His discussion of these topics casts light on some
preoccupations of recent epistemology of perception. Recent epistemology
of perception is concerned with (i) whether seeing provides a special sort
of reason or justification – a “factive reason” such that one could not have
that reason without the attendant perceptual belief being true; and (ii)
how it might be that, if seeing does not provide such factive reasons, see-
ing provides reasons or justification sufficient for knowledge. What Austin
says about de facto incorrigibility and conclusive verification bears on both
these questions.

5.2 De Facto Incorrigibility

After claiming that the actual agenda of sense data theorists is to find a
privileged class of incorrigible sentences (chapter IX), Austin begins his
examination of incorrigibility by saying, “The pursuit of the incorrigible is
one of the most venerable bugbears in the history of philosophy” (104). He
notes that it is a mistake to search for incorrigibility in statements about
the way things look:
[D]escriptions of looks are neither ‘incorrigible’ nor ‘subjective’. . . . [C]ertainly
someone might say, ‘It looks heliotrope’, and then have doubts either as to
whether ‘heliotrope’ is right for the colour this thing looks, or (taking another
look) as to whether this thing really looks heliotrope. There is certainly nothing
in principle final, conclusive, irrefutable about anyone’s statement that so-and-
so looks such-and-such. (42)

18
Cohen 1988.
8

178 Krista Lawlor
Interestingly, Austin has a diagnosis of why this is so, and it has to do with
what we mean with our “looks” talk:
the way things look is, in general, just as much a fact about the world, just
as open to public confirmation or challenge, as the way things are. I am not
disclosing a fact about myself, but about petrol, when I say that petrol looks
like water. (43)
One’s commitment in saying “so and so looks such and such” is precisely
that others will find the look of the thing a certain way, and in light of this,
we should expect descriptions of looks to be corrigible.
In spite of these remarks, it is a mistake to suppose that Austin is dis-
missive about incorrigibility. Austin says no kind of sentence, once uttered,
is incapable of being amended or retracted. (112) However, Austin notes,
a particular utterance made in particular circumstances may be in fact
incorrigible:
If I carefully scrutinize some patch of colour in my visual field, take careful
note of it, I know English well, and pay scrupulous attention to just what
I’m saying, I may say, ‘It seems to me now as if I were seeing something
pink’; and nothing whatever could be produced as showing that I had made
a mistake. But equally, if I watch for some time an animal a few feet in front
of me, in a good light, if I prod it perhaps, sniff, and take note of the noises
it makes, I may say, ‘That’s a pig’; and this too will be ‘incorrigible’, nothing
could be produced that would show that I had made a mistake. (114)
Utterances have de facto incorrigibility when “nothing could be produced
that would show I had made a mistake.” Austin stresses that this sort of
incorrigibility is recognized by our commonsense epistemology.
The problem for the philosopher is to make sense of the common-
sense claim that Austin voices, that sometimes “nothing whatever could
be produced that would show I had made a mistake.” This is something
we might say on occasion, but how on earth might it be true? The task
for philosophers is to give a story about our commitments in making
such claims – and in turn show that the idea of de facto incorrigibility is
intelligible.
What are the truth conditions for an utterance of “nothing could be
produced that would show I had made a mistake”? Considering sentence
meaning alone in isolation from circumstances might suggest to us the
sentence when uttered is true when there is no possible world in which
the creature before one is not a pig. But such a world is after all possible,
and so if the truth conditions of our utterance are simply fixed by sentence
meaning, then what we say is false.
179

Austin on Perception, Knowledge, and Meaning 179

Here is where we can bring Austin’s insights to bear. What the modal
claim “nothing could be produced that would show I had made a mistake”
commits us to depends on the circumstances. The truth or falsehood of
such an utterance is a function of sentence meaning in the circumstances
of utterance. Austin notes that utterances are only incorrigible with respect
to the circumstances in which they are made:
many kinds of sentences may be uttered in making statements which are
in fact incorrigible – in the sense that, when they are made, the circum-
stances are such that they are quite certainly, definitely, and un-retractably
true. (115)
How do circumstances help determine the truth or falsity of the modal
claim? Here is one way to make this idea more precise. Suppose the case is
one where I see a pig and the reasonable alternatives are that the animal is a
large goat or a small cow, and my visual experience rules out these alterna-
tives. “I know it’s a pig” is true. Suppose R is an alternative to the proposi-
tion, P, that the creature is a pig. The truth conditions for an utterance of
“R could be produced as a cogent ground for retracting the utterance that
P” are that R is in the circumstances an uneliminated reasonable alter-
native. Consequently, the truth conditions for an utterance of “R could
be a cogent ground for retracting the utterance that P” are fixed by the
same reasonable alternatives that fix the truth conditions of an assurance
“I know P” in the same circumstances.19 This way of making the idea pre-
cise seems to be on Austin’s mind when he says
surely there will be plenty of cases in which what we say by [the] utterance
will in fact be incorrigible – cases in which, that is to say, nothing whatever
could actually be produced as a cogent ground for retracting [it]. (114)
Utterances are de facto incorrigible in particular circumstances when there
are no cogent grounds in those circumstances for retraction. We’ll say in such
circumstances “nothing whatever could be produced that would show
I had made a mistake,” and that will be true; the circumstances are also
such that “I know it” is also true.
I want to note that my interpretation on this point may align with
some things that Charles Travis says about how to interpret Austin.20 Travis
frames the problem of perception this way: let a “ringer” be a condition
where, if S were in it, the condition would be indistinguishable to her from
19
Note that this reading makes sense of Austin’s claims in “Other Minds” (1946: 159ff; 1979: 88ff)
about “reasonable precautions” being sufficient to show it “can’t be” anything else.
20
Travis 2008. Travis, it should be noted, has been calling for philosophers to appreciate the impor-
tance of Austinian semantics for a long while.
0

180 Krista Lawlor
a condition in which she seems to see that p, yet in which S does not know
p. Now we want to accept this claim:
Unmistakability:  if S knows p, then it is for S unmistakable that
p. Specifically:
(i) It is not the case that for all S can see, a ringer holds instead of p
(ii) S sees that this is so (“That what he sees does not admit of ringers is
part of what he sees” in knowing p.21)
But, Travis continues, we want to reject this claim:
Distinguishability:  If S knows p, then S can distinguish his state from a
ringer case.
The question then is, how one can hope to “grasp oneself as seeing
what excludes a ringer” since Distinguishability is rejected? If we reject
Distinguishability, then it seems to follow that for all one can see, one
might be in a ringer condition. So it seems that it cannot be true that
one knows unmistakably that p on the basis of seeing p. As I understand
Travis, he suggests that the way forward is to realize that “might” is an
“occasion-sensitive” term. It is possible that Sid doesn’t face a pig, but an
animatronic robot; however, on this occasion, it is not true that it might
be an animatronic robot. If I’ve got Travis right on this point, he should
find congenial my account of the truth conditions for “nothing could be
produced to show I had made a mistake,” and he should treat “might”
similarly.22
Philosophers have the task of making cogent our commonsense episte-
mic commitments. For philosophers to take up their task, it is necessary
for them to keep in view the situation dependence of the meaning of our
utterances. As we have seen, we can thereby make sense of commonsense
commitments about the incorrigibility of our perceptual claims.

5.3 Conclusive Verification

Austin’s main concern in passages where he discusses seeing and evidence
(chapter X) is to counter Ayer’s contention that we cannot be incorrigi-
ble about, or conclusively verify statements about, ordinary middle-sized

21
Travis 2008: 292.
22
It’s not clear that Travis would want to adopt all that I have formulated above. For instance, it
is not entirely clear to me what Unmistakability comes to, whether it expresses a commonsense
claim about seeing, or whether Austin would advocate it. Also, Travis inclines to seeing Austin as a
disjunctivist (Travis and Kalderon 2013).
181

Austin on Perception, Knowledge, and Meaning 181

dry goods (“material objects”). His main complaint, as we have seen,
is that Ayer imagines a type of sentence that can have a special episte-
mic role. But sentences, as distinct from utterances made in particu-
lar circumstances, “cannot be divided up at all” (123) into those which
are incorrigible, those which provide evidence for other sentences, or
those which can (or cannot) be conclusively verified; it is not sentences,
but utterances made in particular circumstances that are true or false.
Moreover, even if we correct for Ayer’s mistake, and consider utterances
in circumstances, it is just not the case that utterances about “material
objects” will have lesser epistemic standing than utterances about experi-
ences or sense data.
He then illustrates this complaint with various examples. Some show
that utterances in circumstances about ordinary “material objects” can be
conclusively verified. Other examples show that sometimes such utter-
ances may not stand in need of “verification.” (Is that your house? Have
you verified it is?) And others suggest that not all statements about material
objects need evidence: at the barnyard, one gathers evidence that there are
pigs about, but when one sees the pig, one does not merely have evidence.
One just sees that it is a pig (115).
Does Austin then hold that in every case “seeing is knowing”? Certainly
that would be an overly strong claim, and as it happens not Austin’s view.
He is explicit that it is possible that one might see something and yet need
to do more in order to have “conclusive verification” (118–119). Seeing the
pig might not be enough – if for instance one has reason to think it’s a
peccary in captivity, then one might need to prod the pig or do a further
test. In such a case, seeing the animal is not enough to have conclusive ver-
ification or knowledge. A relevant alternatives account of knowledge lets
us say why this is so – which alternatives are relevant depends on the cir-
cumstances, which include the target of one’s knowledge claim and what it
is reasonable to worry about.
Return now to the questions that occupy recent philosophy of
perception: (i) Does seeing provide a special sort of reason or justification –
a “factive reason” such that one could not have that reason without the
attendant perceptual belief being true? Epistemological disjunctivists
claim that seeing gives one a special epistemic standing that does not share
a common epistemic factor with merely seeming to see; seeing provides
factive reasons, while merely seeming to see does not.23

23
For instance in McDowell 1994b; Pritchard 2012.
2

182 Krista Lawlor
Given what I’ve attributed to Austin so far, I think he is not best under-
stood as an epistemic disjunctivist. First and most generally, there is good
reason to think Austin is a fallibilist, and epistemic disjunctivism is strongly
motivated by impatience with fallibilism. Second, as we have seen, Austin
countenances the possibility that one might need to do more than see the
pig to have “conclusive verification” or knowledge  – one might need to
prod the pig, if that is what is needed to eliminate reasonable alternatives.
This suggests at least that he would resist the idea that seeing always pro-
vides factive reasons.
Finally, and most importantly: as we have noted, there is reason to sup-
pose that Austin thinks an utterance’s de facto incorrigibility rests on what
we might call de facto conclusive grounds, and there is a difference between
de facto conclusive grounds and factive reasons.
Let me explain. First, about the epistemic grounds of incorrigible utter-
ances:  Austin suggests that statements about ordinary “material objects”
may be “conclusively verified.” One way to explicate this idea is in terms
of de facto incorrigibility. As we’ve seen, an utterance is de facto incor-
rigible when nothing could be produced in the circumstances as a cogent
grounds for retracting it. One’s epistemic grounds in making the utterance
eliminate what are in the circumstances all the reasonable alternatives to
its truth. In such a case, the statement is conclusively verified. The epis-
temic grounds one has in making a conclusive verification are grounds
one has when one has done all one needs to eliminate what are in the cir-
cumstances all the reasonable alternatives. Conclusive verification involves
having de facto conclusive grounds. But now, we can contrast de facto con-
clusive grounds with factive reasons. Factive reasons are reasons you would
not have unless the target proposition is true. De facto conclusive grounds
are grounds that eliminate what are in the circumstances all the reasonable
alternatives to the target proposition. It might be that one has de facto
conclusive grounds for what is a false proposition. You see what looks for
all the world like a pig in the barnyard, but it is a peccary shipped in from
South America on a trial basis. Your evidence gives you de facto conclusive
grounds for thinking it is a pig, but it isn’t.
For these reasons, I  suggest Austin would resist the claim that seeing
provides a factive reason.
This makes our second question more pressing: (ii) If seeing provides
less than factive reasons, how can it suffice for knowing?
This question can be developed in different ways. One way has us ask-
ing, how can the experience of seeming to see provide justification sufficient
for knowing, given that this experience is consistent with the falsity of
183

Austin on Perception, Knowledge, and Meaning 183

one’s attendant perceptual belief? Here the key is just to make sense of
fallibilism about perceptual knowledge. I  argue at length elsewhere that
Austin’s reasonable alternatives theory of knowledge, combined with his
situation semantics, provides a coherent fallibilism.24 Seeing can provide
reasons sufficient for knowing. One can know it’s a pig by looking, because
one can gain de facto conclusive grounds just by looking. If in the circum-
stances, the only reasonable alternatives are eliminated by the look of the
thing, then one has sufficient reasons for knowledge by seeing the thing.
And if in fact it is a pig, then seeing can also give one knowledge.
Another way to push the issues raised by (ii) is more overtly skeptical:
your seeming to see there’s a pig is consistent with this all being a dream,
or your being a brain-in-a-vat merely fed experiences by an evil scientist.
How then can your visual experience be the basis of knowledge if you do
not know that you’re not just dreaming it all up? A version of this skeptical
argument is given sharper form by Barry Stroud.25 Stroud argues that our
concept of knowledge has as a necessary condition that if one is to know
anything about the world, one must know that it’s not the case that all
of one’s present experiences are merely dream experiences; further Stroud
claims that the condition cannot be filled.
Recently, Adam Leite finds in Sense and Sensibilia a substantive rejoinder
to the Cartesian external world skepticism articulated by Stroud.26 Much
of Leite’s article suggests that Austin makes a straightforward response to
the skeptic, accepting the “Cartesian condition,” and claiming that the
condition is filled: I do know that I’m not merely dreaming all this; I know
my experiences are not disconnected from the world around me.
I think an alternative approach to skepticism is more readily attributed
to Austin. The basic idea is to reject the Cartesian condition. You’re at the
farm, staring at the pig, and perhaps you’ve even prodded it a bit. You don’t
need to know that you’re not dreaming this all up in order to know it’s a
pig. If we pay attention to our actual practices of assurance-giving, the
challenges and defenses we’re ready to make, we’ll see this is so.27
What more needs be done to make good on this response to Stroud’s
Cartesian skepticism? Two large tasks now face us. One is to respond to

24
Lawlor 2013.
25
Stroud 1984.
26
Leite 2011. I am not certain how to reconcile Leite’s claim that circumstances determine whether
an alternative is defeated with his circumstance-independent claim that we never have reason to
consider “merely metaphysical” possibilities when we worry about what we know.
27
Mark Kaplan makes a strong case that Descartes’ condition is not part of our concept of knowl-
edge. Kaplan 2000.
4

184 Krista Lawlor
Stroud’s rejoinder that ordinary linguistic practice or judgments do not
reveal the meaning of our words, the content of our concepts, or the
truth conditions of our utterances; an alternative skeptic-friendly seman-
tic account is coherent, Stroud insists. A  second large task is to address
apparent inconsistencies in ordinary knowledge ascriptions. For instance,
it seems a bit of common sense that we know what follows from what we
know. But “closure” principles raise difficulties. You say you know it’s a pig.
So do you thereby know it’s not a peccary that happens to look just like a
pig? You hesitate. So what is it? If the answer is yes, then we’d like a story
about how you could know this on the basis of a visual experience that is
exactly the experience you would have were it a pig-like peccary. And if the
answer is no, then how can you claim to know it’s a pig in the first place?
Handling these questions is a large task, and the discussion is longer
than space permits, so I will just say this much. First, Stroud’s suggestion
that a skeptical semantics is a viable alternative semantics for our language
echoes Ayer’s suggestion that his sense datum language is a viable alterna-
tive language for describing the empirical world. If we want to turn back
Stroud’s skeptic on this front, we do well to return to Austin’s criticisms
of Ayer on this point. Second, on the matter of closure generated para-
doxes: these paradoxes can be successfully handled if we develop Austin’s
insights about the circumstance dependence of meaning. The paradoxes
arise because we are not attentive to the way the truth-values of knowledge
claims depend on the situation talked about.28

6 Conclusion
It is easy to read both too much and too little into Sense and Sensibilia.
Austin does not in the lectures offer a systematic theory of perception or
perceptual knowledge, and we read too much in if we suppose that such a
theory lies entirely within its pages. Austin aims for his lectures to illustrate
his ideas about how philosophy should be pursued, using the theory of
perception as an example. In order to demonstrate the fruitfulness of his
approach, he delivers substantive results about perception. His contribu-
tions are partial, and necessarily so, given his own picture of what the work
requires.
Austin’s picture is that common sense encodes important commit-
ments about perception, and the role of philosophy in clarifying and ratio-
nally reconstructing these commitments therefore demands attention to

28
For a compact statement of the situation semantic response, see Lawlor 2015.
185

Austin on Perception, Knowledge, and Meaning 185

ordinary language. Ordinary talk reveals commonsense theory, but only if
we understand what utterances really commit speakers to. And to under-
stand that, an account of natural language meaning is essential equipment.
But we are cautioned repeatedly: the right theory of natural language mean-
ing will note that utterance meaning depends on the situation in which a
thing is said. Austin’s picture, which ultimately calls for the development
of “situation semantics,” has great import for philosophical practice. The
theory of perception is where Austin chooses to advertise his picture. But
the moral embedded in it applies quite broadly. His picture of how to do
philosophy requires philosophers to attend to the situation dependence of
the language in which we express our commitments. We read too little into
the lectures if we fail to see this fact.
While Austin’s target in the lectures is perception, and the tangles phi-
losophers can get into if they aren’t sufficiently attentive, his moral is meant
for all philosophers as they tackle a wide range of issues, not just those of
perception. Austin’s work is a provocation to philosophy as practiced, and
an invitation to do things differently. The provocation is still apt, and the
invitation still largely not accepted. Most urgently, Austin demonstrates
the need for a theory of natural language meaning sufficient for our pur-
poses as philosophers. Here more than on any other point, the moral of the
lectures is of great importance, and not yet widely appreciated.
6

Ch apter  9

Enough is Enough
Austin on Knowing
Guy Longworth

1 Introduction
J. L. Austin’s main discussion of knowledge is in ‘Other Minds’ (Austin
1946; 1979: 76–116). The essay gives rise to numerous questions, both local
and global. Ostensibly, the topic of the essay is knowledge of other minds.
However, explicit discussion of that topic is postponed until the essay’s
twenty-seventh page. Apparently by way of preamble, the main body of
the essay comprises an analysis of aspects of our ordinary treatment of
expressions of knowledge, stippled with tantalizing pronouncements about
knowing in general. One aim of my discussion is to address a global ques-
tion about the function of Austin’s more general discussion of knowledge.
How, if at all, does it further pursuit of our knowledge of other minds?
More local questions arise about Austin’s pronouncements.
I’ll also address a more local question. Austin invites us to consider a
natural way of treating a claim to the effect that a goldfinch is present:
If you have asked ‘How do you know it’s a goldfinch?’ then I may reply
‘From its behaviour’, ‘By its markings’, or, in more detail, ‘By its red head’,
‘From its eating thistles’. . . . You may object: . . . But that’s not enough:
plenty of other birds have red heads. (Austin 1946: 154–155; 1979: 83)

Reflecting on the extent to which one who knows is required to be in a

position to address such challenges, he writes the following:
Enough is enough: it doesn’t mean everything. Enough means enough to
show that (within reason, and for present intents and purposes) it ‘can’t’ be
anything else, there is no room for an alternative, competing, description of
it. It does not mean, e.g., enough to show it isn’t a stuffed goldfinch. (Austin
1946: 156; 1979: 84)

Austin’s pronouncement is both enticing and elusive: enticing, in that it

seems to expose a significant limit on our obligations as knowers; elusive,
186
187

Enough is Enough: Austin on Knowing 187

in that Austin fails to resolve or vindicate the alleged limit. What does he
mean to claim about knowing? Should we believe him?
Philosophical positions rarely form in a vacuum, and so one way to
enhance one’s understanding of a position is by attending to its roots. On
the topic of knowledge, Austin’s avant-couriers were his teacher, H. A.
Prichard, and Prichard’s teacher, John Cook Wilson, and so it is on their
work that I shall focus. (Important influences who won’t be considered
here include G. E. Moore, especially his 1905–1906.) I’ll begin by sketching
the shared core of Cook Wilson’s and Prichard’s views about knowledge –
the epistemological component of Oxford Realism – before pointing to
ways in which Austin’s position emerges naturally from theirs. I’ll suggest
that Austin would have viewed his discussion of knowledge, not only as a
preamble to a treatment of knowledge of other minds, but as a case study.
And I’ll suggest that Austin’s pronouncement is best understood as a par-
tial characterization of a necessary, but insufficient condition on knowing.
We’ll see that when Austin’s pronouncement is so understood, there is no
reason to take him to be proposing that someone might know that a gold-
finch is present without being in a position to know that it isn’t stuffed.
Section 2 comprises a sketch of six core commitments of Oxford
Realism. Section 3 explains how Austin’s general discussion of knowledge
constitutes a case study of our knowledge of other minds. Sections 4 and 5
develop an interpretation of Austin’s project on which his pronounce-
ment leaves intact that being in a position to know is closed under known
entailment.

2 Oxford Realism

2.1 Knowledge as Primitive

The first and most basic commitment of Oxford Realism is that knowledge
is primitive. Thus, Cook Wilson writes:
Perhaps most fallacies in the theory of knowledge are reduced to the pri-
mary one of trying to explain the nature of knowing or apprehending. We
cannot construct knowing—the act of apprehending—out of any elements.
I  remember quite early in my philosophic reflection having an instinc-
tive aversion to the very expression ‘theory of knowledge’. (Cook Wilson
1926: 803)
Prichard echoes:
Knowledge is sui generis and therefore a ‘theory’ of it is impossible.
Knowledge is simply knowledge, and any attempt to state it in terms of
8

188 Guy Longworth

something else must end in describing something which is not knowledge.
(Prichard 1919: 245)
According to Cook Wilson and Prichard, knowledge is a distinct kind, and
cannot be constructed out of elements distinct from knowledge. If the goal
of a theory of knowledge is to say how knowledge is constructed, then the
goal is unachievable; there can be no such theory.

2.2 Knowledge as Akin to Proof

The first commitment of Oxford Realism is bound up with a second:
knowing amounts to, or is equivalent to, possession of proof:
In knowing, we can have nothing to do with the so-called ‘greater strength’
of the evidence on which the opinion is grounded; simply because we know
that this ‘greater strength’ of evidence of A’s being B is compatible with A’s
not being B at all. (Cook Wilson 1926: 100)
The view is not that one who knows must possess a cogent derivation of
what they know from premises that are distinct – and, perhaps, known.
That is, the view is not that knowing is equivalent to possession of a proof.
Rather, the view is that knowing is itself proof of what is known. Where
one knows, one has a conclusive guarantee of that which one knows. One’s
standing with respect to that which one knows is incompatible with falsity.
Thus, meeting a threshold condition on strength of evidence could not
suffice for knowing if meeting that condition were consistent with falsity.
As Prichard notes, Cook Wilson appears to have derived his model of
knowing from reflection on mathematics:
The point of departure of Cook Wilson’s views lay in his unwavering con-
viction of the truth of mathematics. In mathematics we have, without real
possibility of question, an instance of knowledge; we are certain, we know.
(1919: 302)
The model of mathematics is useful in presenting a case in which it is
plausible that we can possess absolute guarantees of truth. Furthermore,
it presents a case in which we can possess such guarantees both with respect
to basic and derived truths, and so a case in which the distinction between
possession of proof and possession of a proof is operative. However, the
model also points to a delicate issue about the notion of incompatibility
that figures in Cook Wilson’s exposition. For suppose that incompatibil-
ity of a standing with falsity amounted to the impossibility of having that
standing with respect to a proposition whilst the proposition is false. On
the natural assumption that true mathematical propositions are necessarily
189

Enough is Enough: Austin on Knowing 189

true, it would be impossible to have any standing with respect to such
propositions whilst those propositions were false. And in that case, incom-
patibility with falsity would fail to characterize a distinctive property of the
standing of one who knows. Thus, the operative notion of incompatibility
must be more demanding than the simple modal notion.

2.3 Knowledge as a State of Mind

The third commitment of Oxford Realism is a partial corollary of the first
two: knowing is a state (or frame, or condition) of mind. We can recon-
struct a path to the commitment as follows. If one’s standing when one
knows is to furnish one with a guarantee against falsity, then it must make
a difference to one’s subjectivity. Thus, one’s standing must partly comprise
a mental state. Suppose, then, that that mental state did not suffice for
knowing. In that case, knowing would comprise that mental state together
with whatever extra-mental elements provide a guarantee against falsity.
But in that case, knowing would be constructible out of elements. Hence,
since knowing is at least partly mental, and is not constructible out of
elements, it must be wholly mental. That is, knowing is a mental state,
occupancy of which state is incompatible with falsity. (See also McDowell
1982; Williamson 2000.)

2.4 Knowing as Distinct from Believing

Connected with the third commitment is a fourth: knowing is not a form
of believing. In particular, knowing is not believing whilst meeting further
conditions. In fact, Cook Wilson makes the stronger claim that knowing
excludes believing:
Belief is not knowledge and the man who knows does not believe at all what
he knows; he knows it. (Cook Wilson 1926: 100)
Prichard brings together the previous four commitments in a way which
again echoes Cook Wilson:
Knowing is not something which differs from being convinced by a differ-
ence of degree of something such as a feeling of confidence, as being more
convinced differs from being less convinced, or as a fast movement differs
from a slow movement. Knowing and believing differ in kind as do desir-
ing and feeling, or as do a red colour and a blue colour. (Prichard 1950: 87)
Both Cook Wilson and Prichard are drawn to the stronger claim that
knowing excludes believing due to their positive views about the distinctive
0

190 Guy Longworth

nature of believing  – in particular, that believing is a matter of holding
something true on broadly evidential grounds whilst recognizing that one’s
grounds fail to decide the issue. By contrast, some contemporary thinkers
will be willing to adopt a more minimal conception of believing, or to
accept the existence of a more general kind of state of mind that encom-
passes believing. For example, they will be willing to allow a conception on
which believing, or some more general sort of state of mind, is a matter of
holding something true in a way that is potentially responsive to evidence.
However, the crucial claim here is that knowing isn’t itself a form of believ-
ing. And one could consistently endorse that claim whilst allowing that
knowing doesn’t exclude believing, or even that knowing entails believing.
I can see no grounds for thinking that Cook Wilson or Prichard would
have denied that knowing was, or at least entailed occupancy of, a frame of
mind of the more general sort. (See Williamson 2000: 41–8.)

2.5 The Accretion

The fifth commitment of Oxford Realism concerns our capacities to know
which frames of mind we occupy. The seemingly implausible strength of
this commitment, together with its seeming independence from other
commitments, has led Charles Travis to label it the Accretion (Travis 2005).
Cook Wilson presents the commitment in the following passage:
[knowledge cannot be one of ] two states of mind . . . the correct and the
erroneous one . . . quite indistinguishable to the man himself. [For] as the
man does not know in the erroneous state of mind, neither can he know in
the other state. (Cook Wilson 1926: 107)
The first thought contained here is that subjects must be in a position to
distinguish any state of knowledge from other ‘erroneous’ states of mind,
at least in principle. The second is that if a state were not in the required
sense distinguishable by its subject from ‘erroneous’ states, then  – since
those other states are, by assumption, not states of knowledge – that state
could not be a state of knowledge.
Cook Wilson’s view may have been sponsored by an argument like the
following. In order for a state to be a case of knowledge, it must be dif-
ferent in kind from any ‘erroneous’ state. Furthermore  – and, perhaps,
because the kinds in question are mental kinds – the required difference
must have a subjective reflection: it must make a difference to how things
are, or seem, from the subject’s perspective. If the difference between the
target state and its ‘erroneous’ ringers were blankly external to the way
191

Enough is Enough: Austin on Knowing 191

things are for the subject, then how things were subjectively for the subject
of either kind of state would be compatible with their not knowing. And
in that case, even if they occupied the target state, they wouldn’t know.
A final step in the argument is required in order to connect the require-
ment that the difference between knowing and its ‘erroneous’ ringers be
reflected subjectively with the further requirement that the subject be in a
position to distinguish the two states – that is, to tell the two states apart.
What, more precisely, does Cook Wilson mean by claiming that sub-
jects must be in a position to distinguish states of knowing from ringers?
Prichard offers the following elaboration:
We must recognize that when we know something we either do, or by
reflecting can, know that our condition is one of knowing that thing, while
when we believe something, we either do or can know that our condition is
one of believing and not of knowing: so that we cannot mistake belief for
knowledge or vice versa. (Prichard 1950: 88)

Prichard’s elaboration of the Accretion invokes two conditions on knowing:

(i) If one knows p, then one is in a position, at least in principle, to know
by reflection that one knows p.
(ii) If one believes p without knowing p, then one is in a position, at
least in principle, to know by reflection that one believes p without
knowing p.
Condition (i) does not obviously entail condition (ii). It is consistent to
hold that one might fail to know p whilst being unable to know by reflec-
tion that one failed to know p even if one also held that if one knew p,
then one would be in a position to know that one knew p. To take one sort
of example, one might reasonably hold that a severely drunk person can
fail to know that they are drunk (and can even believe that they are sober)
whilst at the same time being precluded by their drunkenness from know-
ing that they don’t know that they are drunk. And holding that would
seem perfectly consistent with also holding that someone who is sober, and
who knows that they are, would be in a position to know that they know
that. (See Williams 1978: 309–313; Soteriou 2016: 117–156.)
Furthermore, when conditions (i) and (ii) are distinguished, it becomes
apparent that meeting the former condition would be enough to render
states of knowing distinguishable from their ringers. On at least one rea-
sonable understanding, one can distinguish Fs from Gs just in case one can
activate knowledge that presented Fs are not Gs. Similarly, one can distin-
guish Gs from Fs just in case one can activate knowledge that presented
2

192 Guy Longworth

Gs are not Fs. Thus, distinguishability is asymmetrical. Very often the
required capacities run in step:  one can activate knowledge that a pre-
sented sheep isn’t a wolf when, and only when, one can activate knowledge
that a presented wolf isn’t a sheep. In those cases, the asymmetry doesn’t
matter. But in the sorts of cases we’ve just considered, the capacities can
come apart: one can activate knowledge that a case of one’s sobriety isn’t a
case of one’s severe drunkenness, even though one cannot activate know-
ledge that a case of one’s severe drunkenness isn’t a case of one’s sobriety.
One can distinguish one’s knowing from one’s occupying ringer states if
one can activate knowledge that one’s state is not a ringer state. And that
can be so even if one cannot distinguish one’s occupying a ringer state from
one’s knowing. Cook Wilson’s requirement that knowing be distinguish-
able by its subjects from ignorance can be implemented by condition (i);
condition (ii) is needless.
Even if we treat the Accretion as incorporating only condition (i), it
seems implausibly demanding. For according to condition (i), that one
knows is, in Timothy Williamson’s sense, a luminous condition: for every
case α, if in α one knows, then in α one is in a position to know that one
knows. And Williamson has offered powerful arguments that no condition
which obtains only sometimes – and in particular no such condition that
one knows – is luminous (Williamson 2000: 93–123).

2.6 Being under the Impression

The Accretion presents Oxford Realism with a difficulty. Given the lumi-
nosity of one’s epistemic position, how is it possible for one to make
mistakes? More carefully, how is it possible for one to make mistakes
that one cannot, by reflection, correct? Cook Wilson offers the following
example:
we see at a little distance a person whom ‘we mistake for an acquaintance’
and without hesitation perform some act which it would be a liberty to take
with anyone but an acquaintance, do something in fact which we rightly say
we should not have done if we had ever suspected he was not an acquaint-
ance. We did not act on an opinion that it was our friend; for, in forming an
opinion, we are aware that the evidence is insufficient and, if we had thought
that, we should never have done the act. It seems more like belief; but if we
had consciously made it a matter of belief, we should have distinguished it
from knowledge, and then again, ex hypothesi, we should not have done the
act. Probably one answer offered would be that, though we didn’t know, we
thought we knew. But this will not suffice. Apart from the criticism we have
193

Enough is Enough: Austin on Knowing 193

already passed on this phrase itself, if we really thought we knew, we must
have reflected and must have thought the evidence conclusive, whereas,
ex hypothesi, any reflection shows it could not be conclusive. (Cook Wilson
1926: 109–110)
Cook Wilson fails to detail what no Edwardian would have risked from a
position of ignorance. His response to the challenge of explaining its per-
formance is the invocation of a further species of attitude: being under the
impression. This is the sixth commitment of Oxford Realism. Being under
the impression is – like opinion or belief – a mode of holding something
to be true. However, unlike belief or opinion, being under the impression
need not be installed or sustained by reflection. Indeed, being under the
impression is incompatible with, and so apt to be destroyed by, reflection
(Cook Wilson 1926: 108–113).
Being under the impression falls outside the scope of the Accretion.
One cannot know by reflection that one is under an impression, since
reflection would release one from its hold. Thus, being under the impres-
sion can help to explain how subjects can hold things to be true in a way
precluded by the activation of knowledge. It can explain how unreflective
mistakes are possible. However, it’s natural to think that some mistakes
can withstand reflection. If one were ensconced in a standard sceptical
scenario, then one might be subject to mistakes both about how things
were – e.g. that one had hands – and about one’s attitudes to how things
were – e.g. that one knew that one had hands. And it’s natural to think that
no amount of reflection would remedy one’s situation.
There are two broad routes via which an attempt might be made
to exploit the state of being under the impression in order to explain
apparently reflection-free error. The first would be to deny that genu-
inely reflective errors are possible. Genuine reflection would reveal that
one doesn’t know that one has hands and, so, that one isn’t entitled to
hold true that one has hands. The claim would be that we are commonly
less reflective than we take ourselves to be – that is, than we are under
the impression of being. The second route would be to extend the first
by seeking to explain barriers to reflection. The claim would be that in
certain circumstances genuine reflection cannot be undertaken. Severe
drunkenness might block reflection. More delicately, being the subject
of a sceptical scenario might prevent the activation of knowledge that
would otherwise enable reflection.
That completes my sketch of the six core commitments of Oxford
Realism:  (1)  knowledge is primitive; (2)  knowledge is akin to proof;
4

194 Guy Longworth

(3)  knowledge is a state of mind; (4)  knowledge is distinct from belief;
(5) the Accretion; (6) mistakes depend on the unreflective state of being
under the impression. (See also Marion 2000a, 2000b; Travis 2005; Travis
and Kalderon 2013.)

3 Other Minds
There are numerous echoes of Oxford Realism in Austin’s work. Cook
Wilson and Prichard both present commitment (1), knowledge as primi-
tive, as precluding one form of theory of knowledge. Austin agrees:
there could be no general answer to the questions what is evidence for what,
what is certain, what is doubtful, what needs or does not need evidence, can
or can’t be verified. If the Theory of Knowledge consists in finding grounds
for such an answer, there is no such thing. (Austin 1962: 124)
Commitment (2), knowledge as akin to proof, reverberates more widely.
Thus, for example, Austin contrasts (inconclusive) evidence with what set-
tles a question:
The situation in which I would properly be said to have evidence for the
statement that some animal is a pig is that, for example, in which the
beast itself is not actually on view, but I can see plenty of pig-like marks
on the ground outside its retreat. If I find a few buckets of pig-food, that’s
a bit more evidence, and the noises and the smell may provide better evi-
dence still. But if the animal then emerges and stands there plainly in view,
there is no longer any question of collecting evidence; its coming into view
doesn’t provide me with more evidence that it’s a pig, I can now just see
that it is, the question is settled. (Austin 1962: 115; cp. 1946: 176–182; 1979:
105–111)
Austin makes a closely related point here:
saying ‘I know’ is taking a new plunge. But it is not saying ‘I have performed
a specially striking feat of cognition, superior, in the same scale as believing
and being sure, even to being merely quite sure’: for there is nothing in that
scale superior to being quite sure. (Austin 1946: 171; 1979: 99)
It is natural to read Austin as appealing here to a distinction between the
accumulation of grounds for sureness of belief and the achievement of a
position that differs not merely in degree, but in kind.
The distinction between evidence and proof feeds into the idea of
knowledge as a state of mind distinct from belief (commitments (3) and
(4)). At first blush, Austin might be read as rejecting the first idea in the
following passage:
195

Enough is Enough: Austin on Knowing 195

If we like to say that ‘I believe’, and likewise ‘I am sure’ and ‘I am certain’,
are descriptions of subjective mental or cognitive states or attitudes, or what
not, then ‘I know’ is not that, or at least not merely that: it functions differ-
ently in talking. (Austin 1946: 150; 1979: 78–79)
However, the key phrase is ‘or at least not merely that’, and the main point
is to distinguish knowing from believing. The claim is that ‘I know’ is not
merely a description of a subjective mental state. That is either because
knowing differs from believing in not being merely a subjective mental
state, or because ‘I know’ has functions over and above describing the sub-
jective mental state of knowing. (The latter idea figures in Austin’s infa-
mous comparison of ‘I know’ with ‘I promise’. See Austin 1946: 169–175;
1979: 97–103.)
There is evidence, then, if not proof, that Austin accepted at least the
first four commitments of Oxford Realism. That supports a straightfor-
ward answer to our first question. A  general discussion of knowledge
might be useful preparation for a discussion of knowledge of other minds.
However, insofar as Austin views knowledge itself as a state of mind, such
a discussion assumes a more central role. For the question how we can
know about what other people know is viewed as a special case of the more
general question of how we can know about other’s minds.
Furthermore, there are forces internal to Oxford Realism that make it
pressing to address questions about our knowledge of what other people
know. One source of pressure is the idea that knowledge is primitive. It is
apt to seem to follow from that idea that there are no independently speci-
fiable criteria by which to discern whether or not someone knows. And in
that case, we can’t account for our knowledge of whether or not someone
knows by appeal to our knowledge of whether or not they satisfy such cri-
teria. A second source of pressure arises from the Oxford Realist treatment
of the idea that knowing is a state of mind and, in particular, the Accretion.
For by making the reflective subject decisively authoritative about whether
or not they know, the Accretion problematizes the idea that other people
might have access to ways of determining whether or not the subject knows
other than via recourse to the subject’s avowals of their own reflective view
about whether or not they know. The first two sources of pressure sponsor
a third: the threat of dogmatism. Suppose, first, that reflective subjects can
be decisively authoritative about whether or not they know and, second,
that there are no independently discernible conditions to which appeal can
be made in attempting to show that subjects are wrong to take themselves
to know. In that case, it would be hard to see how a challenge could be
mounted to someone’s sincere and reflective claim to know. The situation
6

196 Guy Longworth

would be akin to one in which a subject reflectively avows that they believe
something. Typically, the most that one would achieve by challenging such
an avowal would be to encourage the subject to reflect again. So questions
about how we can know whether other people know arise naturally from
engagement with Oxford Realism. (R. G. Collingwood’s central complaint
about Oxford Realism was that it enabled dogmatism (Collingwood 1939).
The threat was live: Cook Wilson thought he knew that non-Euclidean
geometry is impossible. Cook Wilson 1926:  456, 561. See also Prichard
1950: 99–100.)
As Austin in effect observes, the Oxford Realists’ denial that knowing can
be reconstructed out of other materials – that is, their claim that there are
no independently specifiable sufficient conditions for knowing – leaves open
that there are independently specifiable necessary conditions on knowing.
One such condition to which the Oxford Realists make explicit appeal is
that if one knows, then one can’t be wrong. (See e.g. Cook Wilson 1926: 69;
Prichard 1950: 88.) Opining falsely is incompatible with knowing. It is to
that condition that Austin points in his comment to the effect that talk
about knowing functions differently from talk about believing. Whilst fal-
sity is no immediate threat to the claim that someone believes, it excludes
the claim that they know. So, if it is belief ’s compatibility with falsity which
is supposed to ground the idea that belief is merely subjective, and which in
turn grounds the idea that thinkers are decisively authoritative about what
they believe, then that ground is unavailable with respect to knowing.
Austin’s central project involves seeking to discern further necessary
conditions on knowing, as revealed in our ordinary handling of challenges
to a thinker’s claims to know. A  typical challenge might begin with the
question how the thinker knows, raised with the aim of having the thinker
reveal sources of their standing, with those revelations potentially giving
rise to further challenges. Thus, for example,
If you have asked ‘How do you know it’s a goldfinch?’ then I  may reply
‘From its behaviour’, ‘By its markings’, or, in more detail, ‘By its red head’,
‘From its eating thistles’. (Austin 1946: 154; 1979: 83)
Austin focuses on challenges to the presumption that the thinker possesses
a standing equivalent to proof:
You may object: . . . But that’s not enough: plenty of other birds have red
heads. (Austin 1946: 155; 1979: 83)
If the thinker’s standing were exhausted by their awareness of the bird’s red
head, then – according to the challenger – their standing would be com-
patible with falsehood, and so could not amount to proof.
197

Enough is Enough: Austin on Knowing 197

Superficial appearances notwithstanding, we can see Austin as pursuing
questions about our knowledge of other minds throughout ‘Other Minds’.
In doing so, he seeks to discern necessary conditions on knowing. In the
following section, we’ll consider some of the conditions that he discerns.

4 Reasons for Doubt
The question at issue is whether some particular thinker knows some par-
ticular fact. The answer is pursued by a challenger who seeks to determine
whether the claimant meets necessary conditions on possessing proof:
It is in the case of [this] objection that you would be more inclined to say
right out ‘Then you don’t know’. Because it doesn’t prove it, it’s not enough
to prove it. (Austin 1946: 155; 1979: 84)
One natural reading is the following. It is a necessary condition on know-
ing that a goldfinch is present that one’s standing is equivalent to proof. It
is a natural consequence of that condition that one who knew would be
in a position to articulate that standing, so as to show that they had the
required standing. And showing that one had the required standing would
amount to providing proof. If the best that a thinker could do by way
of articulating their standing didn’t amount to proof, then their incapac-
ity would provide reason to think that they lacked proof and, so, failed
to know.
On the natural reading, it would be reasonable to expect that if a
candidate knower were able successfully to navigate all appropriate chal-
lenges to their claim to know, then their doing so would constitute a
conclusive defence of their claim. That is a perspective from which it is
natural to read Austin’s further comments on the obligations attending
the candidate:
(a) If you say ‘That’s not enough’, then you must have in mind some more
or less definite lack. ‘To be a goldfinch, besides having a red head it
must also have the characteristic eye-markings’: or ‘How do you know
it isn’t a woodpecker? Woodpeckers have red heads too’. If there is
no definite lack, which you are at least prepared to specify on being
pressed, then it’s silly (outrageous) just to go on saying ‘That’s not
enough’.
(b) Enough is enough: it doesn’t mean everything. Enough means enough
to show that (within reason, and for presents intents and purposes) it
‘can’t’ be anything else, there is no room for an alternative, competing,
description of it. It does not mean, e.g., enough to show it isn’t a stuffed
goldfinch. (Austin 1946: 156; 1979: 84)
8

198 Guy Longworth

My focus in this section will be (a). (We’ll return to (b) in the next section.)
The central thought in (a) seems to be that there can be reason to doubt
that a candidate knows something only if there is a reason for doubting
that the candidate’s standing amounts to proof that is available to the
challenger – that is, a reason that is specifiable, and so known. On the face
of it, this represents a significant restriction on the candidate’s obligations.
They are required to be in a position to respond to a challenge only if that
challenge isn’t silly (outrageous); and a challenge meets that condition only
if it is backed by reasons that are specific and available. Austin expands on
the apparent restriction:
The doubt or question ‘But is it a real one?’ has always (must have) a special
basis, there must be some ‘reason for suggesting’ that it isn’t real, in the sense
of some specific way, or limited number of ways, in which it is suggested
that this experience or item may be phoney. (Austin 1946: 159; 1979: 87)
According to the natural reading, where no challenge that is appropriately
grounded in available reasons is forthcoming, the candidate knows.
By way of comparison, consider a similar idea expressed by J. M. Hinton:
the apparent rigorism which runs ‘I need more than there is here, before
I stop saying that an item is not known’ is at the same time the apparent
laxism, ‘I need no more than there is here, to make me admit an item
“not p” as epistemically possible—not known not to be the case’. (Hinton
1989: 232)
Believing that something is known and believing that something is not
known should be treated symmetrically. Insofar as one should believe
something only if one has reasons for believing it, one should believe that
something is known only if one has reasons for believing that it is known.
But for the same reason, one should believe that something is not known
only if one has reasons for believing that it is not known. Thus, one should
believe that a candidate does not know that a goldfinch is present – that
from their perspective it is epistemically possible that it’s not the case that
a goldfinch is present – only if one has reasons for believing it. Doubt and
belief are equally reason hungry.
Hinton’s proposal that belief and doubt be treated symmetrically doesn’t
impose any substantive restriction on the obligations attending candidate
knowers. In particular, it doesn’t insulate their claims to know from defeat
by countermanding reasons on grounds of the unavailability of those rea-
sons. To what extent does Austin’s proposal impose such a restriction? We
can pursue that question by considering a development of Austin’s pro-
posal due to Adam Leite:
199

Enough is Enough: Austin on Knowing 199

If one recognizes that there is no reason in favour of some possibility that
would undermine one’s authority, competence, or reliability regarding a
certain domain, then (other things being equal) one may reasonably believe
things in that domain even if one lacks adequate independent grounds for
believing that the possibility does not obtain, and one may reasonably dis-
miss as groundless the suggestion that it does obtain. (Leite 2011: 94)

If one recognizes (and so knows) that there are no reasons for believing
that not-p is epistemically possible, then there are no reasons for believing
that not-p is epistemically possible; and in that case, one may reasonably
believe p. Although Leite doesn’t present his proposal as covering the case
of knowledge, a natural extension would be the following. If one recog-
nizes that there is no reason in favour of any possibility the obtaining of
which would undermine a claim to know, then one is in a position to
dismiss as groundless any objections to that claim, and so to endorse the
claim. Indeed, that would be a way of possessing proof.
Leite’s proposal seems close to Hinton’s plea for symmetry and, to that
extent, not to impose substantive restrictions on candidate knowers’ obli-
gations. However, Leite’s applications of the proposal reveal that, as he
understands it, it differs from Hinton’s.
Leite seeks to apply his proposal in order to undercut forms of scepti-
cism that are based on an alleged inability to know that one isn’t dream-
ing. Leite considers two types of case that might be thought to sponsor
a threat to knowledge based on one’s current experience. The first type
of case is that of a dream that is phenomenologically indistinguishable
from wakeful experience. Here, the threat to knowledge is supposed to
arise because one lacks positive reasons for believing that one isn’t cur-
rently suffering such a dream. If it were an epistemic possibility that one
is currently suffering such a dream, then it is plausible that its being so
would preclude one from exploiting one’s current experience in order to
know things about one’s environment. Hence, the fact that one has – and
as a matter of principle can have – no positive reasons for excluding that
epistemic possibility seems to undercut one’s claim to know. However,
the nature of such a dream not only rules out one’s acquiring positive
reasons for thinking that one is not suffering one. It also rules out one’s
acquiring reasons for thinking that one is suffering one. Thus, on reflec-
tion, one can recognize that one can never be apprised of reasons for
believing that one is now suffering such a dream. And now, according
to Leite (2011), his proposal operates to deliver the result that the puta-
tive epistemic possibility can reasonably be dismissed as groundless. Thus,
in serving to exclude what might otherwise have seemed a threatening
0

200 Guy Longworth

possibility, Leite’s proposal is shown to impose a substantive restriction
on the obligations of knowers.
Given the substantive restriction on the obligations of knowers, admis-
sible challenges to claims to know must be grounded on facts or reasons
that are themselves available, and so knowable. Thus, the only type of
dream the possibility of which could pose a threat to one’s claim to know
would be a dream which was phenomenologically distinguishable from
one’s actual experience, and that is the second sort of case that Leite con-
siders. In this case, he allows that we have some reasons for believing that
we might now be suffering such a dream since we have, after all, suffered
such dreams in the past. However, we can in principle establish that we are
not suffering such a dream, by exploiting their phenomenological distin-
guishability from wakeful experience. We therefore have the resources to
exclude the possibility.
Unlike Leite’s initial proposal, his applications rely on restricting admis-
sible challenges to those grounded on available reasons. That suggests that
the transition from proposal to applications involved slippage. On closer
inspection, that is what we find.
This can be seen by comparing the following principles:
1. If one recognizes that there are no reasons that support a possibility,
then the possibility may be dismissed.
2. If one doesn’t recognize that there are reasons that support a possibility,
then the possibility may be dismissed.
3. If one recognizes that one can’t recognize that there are reasons that
support a possibility, then the possibility may be dismissed.
Leite’s initial proposal is a version of principle 1.  However, his applica-
tions rely not on one’s recognizing that there are no reasons that support a
possibility, but rather on one’s recognizing that one cannot recognize that
there are such reasons. Thus, Leite’s applications rely on a version of prin-
ciple 3, rather than 1. We can most clearly see the difference between those
principles by reflecting on the difference between principles 1 and 2. 1, as
we observed, relies on one’s recognizing, and so knowing, that there are no
reasons that support a possibility. Since knowing is factive, that principle
delivers that there are no such reasons. By contrast, principle 2 relies only
on one’s failure to recognize any reasons that support a possibility. That
would entail that there are no such reasons only on the assumption that
we are incorrigible with respect to the geography of reasons. Having noted
the weakness of principle 2, by contrast with 1, it becomes obvious that
principle 3 represents no significant advance. For again, one’s recognizing
201

Enough is Enough: Austin on Knowing 201

that one is debarred from recognizing reasons in support of a possibility
leaves open that that is due to limitations on our ability to track whatever
reasons there are, rather than to the absence of such reasons. Thus, prin-
ciple 1 is quite plausible, but is of no assistance in dealing with, for exam-
ple, phenomenologically indistinguishable dreams. In order to exploit that
principle, one would have to recognize, and so know, that there are no
reasons for believing that one is suffering such a dream. By contrast, prin-
ciple 3 might sustain the dismissal of the possibility that one is suffering
such a dream, but only at a cost to plausibility. For meeting its condition
is consistent with getting things wrong, for reasons of which one is una-
ware even in principle. (That is the central message of Barry Stroud’s use
of Thompson Clarke’s example of the plane spotters. Stroud 1984: 67–74;
Clarke 1972.)
The upshot is that principles 2 and 3 impose no plausible restriction
on the obligations of knowers, since one can adhere to the restriction and
yet fail to know. And principle 1, whilst plausible, imposes no substantive
restriction. With that information in hand, let’s return to consideration
of Austin’s proposal. Is it a version of the insubstantial principle 1, or is it
rather a version of the substantial but implausible principle 3?
It is neither. It can’t be the second because the gap between a thinker’s
being unaware, even in principle, of reasons that would undermine their
claim to know and, on the other hand, there being no such reasons, is too
large and too obvious to have escaped Austin’s attention. And it can’t be the
first because Austin builds into his proposal that operative reasons must
be available. But if Austin’s proposal isn’t to be understood in either of
those ways, how is it to be understood?
At the outset, I suggested that Austin’s proposal seems to impose a sub-
stantive restriction on the range of challenges that a knower should be in a
position to address. And I suggested that his proposal seems to allow that
someone able to answer all such challenges would thereby count as know-
ing. If we adhere to both of those suggestions, we are forced to read Austin
as proposing a restriction on what knowers should be in a position to do
that is either insubstantial or implausible. Since Austin explicitly endorses
the first suggestion, we should reject the second: it is no part of Austin’s
proposal that being able to deal with all available reasons for doubt is a
sufficient condition on knowing.
John McDowell provides a useful model for the proposed reading
of Austin. McDowell presents a view of the standing, and obligations,
of knowers comprising three main claims. First, knowing p is a matter of
possessing reasons that suffice for truth. (Compare Oxford Realism (2) and
2

202 Guy Longworth

(3).) However, second, one can meet that condition without knowing spe-
cifiable, non-question-begging reasons that suffice for that truth. (Compare
Oxford Realism (1).) For example, one’s possession of reasons for holding p
might be constituted by one’s seeing p or one’s remembering p, where those
standings could not be specifiable independently of appeal to one’s know-
ing p. Appeal to those standings would therefore beg the question against a
challenge to one’s claim to know. One might possess reasons that don’t beg
the question in that way – for example, facts about the way things look to
one. But there is no general expectation that it will be possible to recon-
struct sufficient conditions for truth on the basis of those reasons. What is
the role of those reasons, if not to constitute one’s standing as a knower?
McDowell’s third claim addresses that question. Knowing requires that
one is doxastically responsible. That is, it requires that one is appropriately
sensitive to non-question-begging reasons for or against the truth of what
one knows. (McDowell 1994a; 1998: 414–443.) As I’ll explain, that neces-
sary condition on knowing is central to the proposed reading of Austin.
McDowell presents his model in the following passages:
There is a completely cogent argument from the fact that someone, say,
sees that things are thus and so to the conclusion that things are thus and
so. But that argumentative transition cannot serve to explain how it is that
the person’s standing with respect to the fact that things are thus and so is
epistemically satisfactory. . . . Genuinely mediated epistemic standings, on
the conception I have in mind [ – that is, the governing conception that
McDowell rejects – ], would have to consist in the cogency of an argument
whose premises do not beg the relevant question of epistemic standing.
(McDowell 1994a: 196–197; 1998: 416)
If one’s takings of things to be thus and so are to be cases of knowledge, they
must be sensitive to the requirements of doxastic responsibility. We could
not conceive remembering that things are thus and so, say, as a standing in
the space of reasons if a subject could count as being in that position even
if he were not responsive to the rational force of independently available
considerations – the material to which the governing conception appeals.
But we can separate that point from the idea that one can reconstruct the
epistemic satisfactoriness of the standing in terms of the rational force of
those considerations. (McDowell 1994a: 206–207; 1998: 429–430)
Austin endorses the proposed distribution of epistemological labour. In
accord with Oxford Realism, he views knowing as possession of conclusive
reasons. However, in order to have access to such reasons, one must be
appropriately sensitive to independently specifiable, non-question-begging
reasons. Crucially, that obligation is limited. Although one must be able
to address appropriate challenges to one’s claim to know, such challenges
203

Enough is Enough: Austin on Knowing 203

must be based on reasons that are both specific and available. As we saw,
the limitations that Austin proposes are implausible if taken to exhaust the
obligations on knowers. However, when construed as limitations on the
demands of doxastic responsibility, where the latter is a necessary but not
sufficient condition on knowing, they appear defensible.
In the next section, the proposed reading of Austin is extended in order
to encompass part (b) of the target passage. As we’ll see, this allows for a
reading of that passage which is less dramatic, but more plausible, than an
alternative reading due to Mark Kaplan (2011).

5 Enough is Enough
In the previous section, I suggested that we read Austin as aiming to char-
acterize only a necessary condition on knowing, according to which know-
ing requires appropriate sensitivity to non-question-begging reasons for or
against what one claims to know. The conditions on doxastic responsibility
are weaker than the conditions on knowing, since knowing requires pos-
session of conclusive reasons and, so, reasons that, in effect, beg the ques-
tion against challenges. In this section, the proposed reading is applied to
Austin’s claim, in (b), that enough is enough: that meeting the obligations
that attend knowing that there is a goldfinch requires being in a position
to do enough to show that ‘there is no room for an alternative, competing,
description of it. It does not mean, e.g., enough to show it isn’t a stuffed
goldfinch’ (Austin 1946: 156; 1979: 84).
One might read the passage as suggesting that the description of some-
thing as stuffed needn’t compete with its description as a goldfinch, because
it’s possible for something to be both. (Austin was sceptical about the idea
that claims about ordinary things carry a determinate range of entail-
ments: 1946: 159–161; 1979: 88–89; 1962: 118–124.) Another way of reading
(b) would be as allowing that one might be in a position to show that
something is a goldfinch without (yet) being in a position to show that it
isn’t stuffed because one hadn’t realized that being a goldfinch entails not
being stuffed. Kaplan proposes a more tantalizing reading of the passage.
On Kaplan’s reading, Austin’s proposal is that one might be in a position to
meet sufficient conditions on knowing that something is a goldfinch, and
also that its being a goldfinch entails that it isn’t stuffed, and yet not be in
a position to know that it isn’t stuffed. (Kaplan 2011)
Kaplan’s reading can be developed via the sixth component of Oxford
Realism, the appeal to a state of being under the impression. On this read-
ing, knowing that there is a goldfinch, and that its being a goldfinch entails
4

204 Guy Longworth

that it isn’t stuffed, would impose an obligation to hold true that it isn’t
stuffed. However, one can meet that condition without knowing that it
isn’t stuffed. For one might hold it true unreflectively in being under the
impression that it isn’t stuffed (Kaplan 2011: 60–72).
On the assumption that Austin follows the Oxford Realists in identi-
fying knowledge with possession of proof, Kaplan’s reading would have
Austin allowing that one can have proof that there’s a goldfinch, and that
that entails that it isn’t stuffed, without being able to convert those proofs
into proof that it isn’t stuffed. But if one’s purported proofs leave it open
whether it’s stuffed, then one might reasonably doubt that what one has
are proofs. It’s imaginable that one might have done enough to show that
something is a goldfinch, and that its being so entails that it isn’t stuffed,
without having done anything further to show that it isn’t stuffed. But one
might wonder why Austin would have accepted the demand that showing
that something isn’t a stuffed goldfinch must proceed independently of
showing that it is a goldfinch.
Furthermore, and as noted by Leite, Kaplan’s reading is in tension with
Austin’s explicit animadversions about assumptions:
To say . . . that we are making assumptions and taking things for granted
whenever we make an ordinary assertion, is of course to make ordinary
assertions look somehow chancy. (Austin 1962: 138; Leite 2011: 91, fn. 24)

Austin’s general target is the idea that one isn’t always in a position to
know all the things that are entailed by things that one takes oneself to
know. In that case, Austin suggests, ordinary claims to know would look
somehow chancy, since those claims would be dependent on things about
which one was strictly ignorant. One’s standing with respect to the things
that one claims to know would appear no better than one’s standing with
respect to the mere assumptions on which those claims to know depend.
Austin’s response casts doubt on the conjunctive claim (i) that there are
things entailed by things that one takes oneself to know and (ii) that one is
confined to assuming those things. Since Kaplan’s reading relies on Austin’s
endorsing that conjunction, the reading is undermined.
We’ve seen, in the previous section, that an alternative interpretation of
Austin’s project is available. According to the alternative, Austin’s discus-
sion of what a knower must be in a position to show is concerned not with
the totality of a knower’s reasons, but only with those that sponsor their
doxastic responsibility. Knowing requires being appropriately sensitive to
non-question-begging reasons for or against what one claims to know. In
this case, it requires being appropriately sensitive to non-question-begging
205

Enough is Enough: Austin on Knowing 205

reasons for or against the claim that there is a goldfinch. However, one’s
non-question-begging reasons needn’t furnish proof that there’s a gold-
finch. So, although a proof that there is a goldfinch would be convertible,
inter alia, into a proof that it isn’t stuffed, the reasons that figure in that
proof are liable to beg the questions whether it’s a goldfinch and, thus,
whether it’s stuffed. Alternatively, one’s non-question-begging reasons for
holding that there is a goldfinch can leave open whether it’s a goldfinch –
for example, by leaving open whether it’s stuffed. Thus, we can find pas-
sage (b) intelligible without reading Austin as allowing that someone could
know that there is a goldfinch, and that that entails that it isn’t stuffed,
whilst being incapable of knowing that it isn’t stuffed.

6 Conclusion
I’ve suggested that we can better understand some otherwise puzzling
aspects of ‘Other Minds’ if we read that work against the background of
Oxford Realism. First, we can discern evidence that Austin agreed with the
Oxford Realists in viewing knowing as a mental state. In that way, we can
see the whole of Austin’s essay as addressing its titular topic. Second, we
can discern evidence that Austin agreed with the Oxford Realists in view-
ing knowing as primitive. In that way, we can see him as seeking to uncover
necessary conditions on knowing – including, especially, the requirement
of doxastic responsibility – without treating those conditions as elements
in a conjunctive reconstruction of knowing. And that, in turn, makes space
for a plausible reading of some of Austin’s tantalizing pronouncements.1

1
I’m grateful for discussion and comments to audiences at Oxford, Porto, and Warwick, and to Bill
Child, Thomas Crowther, Naomi Eilan, Elizabeth Fricker, Anil Gomes, Mark Kalderon, Hemdat
Lerman, Ian Phillips, Johannes Roessler, Matthew Soteriou, and Charles Travis.
6

Ch apter  10

Knowledge and Knowledge-Claims

Austin and Beyond
Stephen Hetherington

1 Introduction
J. L. Austin’s contributions to epistemology, although narrowly focused,
have been widely influential. The narrowness reflects the fact that
talk  – more generally, the use of language  – was Austin’s overt episte-
mological topic. Still, it was knowledge-talk. Knowledge-attributions
and knowledge-denials  – more so than knowledge-possessions and
knowledge-lacks – were apparently his direct concern. Nevertheless, he
essayed some ideas whose influence is still taking shape, it seems, within
contemporary epistemology. There are many current philosophers whose
overt aim is first and foremost to understand knowledge-attributions
or knowledge-denials  – and the possession or the lack of knowledge
itself only secondarily and consequently, if at all. These philosophers are
said to be epistemologists, on the ground that they discuss utterances
about knowledge, say. In this broad sense, theirs is an Austinian meth-
odology (as section 5 will explain more fully), even if it is not always
employed in the service of Austinian theses.1 This chapter will discuss
those key Austinian theses about knowledge-talk. But we will also meet
a way of using such theses as part of a picture of what knowledge itself
might be. In that respect, we will see how to move, in an Austinian way,
beyond Austin’s overt focus. Graham (1977: 138) describes that focus in
this way:  “Austin does not give us a theory about what it is to know
something. He gives us a theory about what it is to say that you know
something.” Still, we will find, it is possible to turn that into a theory of
knowledge itself.2

1
See Dickerson (2006), however, for an explicitly Austinian contextualism.
2
Lawlor (2013: 10) also sees this. Her approach’s details are quite different to this chapter’s, however,
seemingly reflecting our respective epistemological preferences.

206
207

Knowledge and Knowledge-Claims: Austin and Beyond 207

2 Knowledge-Claims as Performatives?
We should begin with what is probably Austin’s most distinctive proposal
as to how to categorise a knowledge-claim. For this, we turn to his paper
“Other Minds” (Austin 1946; 1979: 76–116), where we encounter his con-
ception of knowledge-claims as what he called performatives.
What did Austin have in mind with the idea of a performative? Section 3
will explain how, in his later work, performatives were conceived of as just
one species within a relevant genus – how Austin aimed to uncover what
underlies or generates the idea of a performative. Right now, though, we
may consider the idea as it arose, initially and more narrowly, in “Other
Minds,” where it was being applied by Austin specifically to a restricted
kind of knowledge-talk. His proposal is one of the twentieth century’s
more widely quoted epistemological ideas. Here are its key elements:
When I say ‘I know [that S is P]’, I  give others my word: I  give others my
authority for saying that ‘S is P’.
To suppose that ‘I know’ is a descriptive phrase, is only one example of the
descriptive fallacy, so common in philosophy. . . . Utterance of obvious ritual
phrases, in the appropriate circumstances, is not describing the action we are
doing, but doing it (‘I do’). (1946: 171, 174; 1979: 99, 103)
That is how Austin regards “first person singular, present indicative tense”
(1946: 170; 1979: 98) uses of the word “know.” (This is why I described his
focus as being on a restricted form of knowledge-talk.) His favoured anal-
ogy is between “I know” and “I promise.” Just as one’s saying “I promise
to dance at your wedding” is one’s promising (other things being equal),
one’s saying “I know that you are soon to be married” is – analogously –
one’s . . . what? Well, the analogy is not perfect and immediate, because
(Austin would allow) one’s saying “I know that you are soon to be married”
is not one’s knowing. Hence, whereas the fact of promising can be con-
stituted by the appropriate use of “I promise,” the fact of knowing is not
constituted by any appropriate use of “I know.” In short, to say “I promise/
know that p” is not equally well one’s promising/knowing that p; it is the
former – promising – but not the latter – knowing. No matter: Austin’s
surrounding discussion makes clear that “I know that p” is a shorthand. It is
a shorthand for something like “I give you my word that p,” or like “I give
you (and relevant others) my authority for saying that p.” And, of course,
to give one’s word or authority is at least analogous to promising.
Thus we are led to Austin’s performative conception of such cases of
knowledge-talk. “I promise to act so that p” (or, equally, “I promise to
8

208 Stephen Hetherington

do A”) is a use of language that – by being used (and only by being used) –
achieves non-descriptively more than what “She promises to act so that
p” (or “She promises to do A”) achieves descriptively. The same is true,
mutatis mutandis, of “I know that p” and “She knows that p.” The latter is
not a performative, but the former is. “She knows” is a performance by the
speaker, but it is not using a word (“knows”) that it is also applying to its
performance, thereby rendering that use of the word a performative; but
all of that is achieved by “I know.”
And what follows from that distinction? We might be concerned, as
was Roderick Chisholm (1966: 15–18; 1969: 110–112), that the observation
in question falls short of establishing Austin’s conclusion, especially about
performatives as a distinct and explanatory class. The reason for the con-
cern is as follows:
Even if (as Austin claims) “I know that p” is being used in a performative
way, the performance succeeds only because there are conditions that entitle
the speaker to perform in that way with those words. And these conditions
may be described as perhaps what it is for one to have the knowledge that
p. In other words, even if Austin is right about the performative nature of
“I know,” this does not entail that “I know” is not also describing – where
successful – the person’s knowing.

Austin might reply that, on the contrary, even the overtly descriptive
“She knows that p” need only be interpreted as describing an actual
or possible performative, such as “She is entitled to say, ‘I know that
p.’ ” But Chisholm would reply, in turn, that this is simply describing
something further about the person that is her having the knowledge –
this state of affairs being what makes her entitled to say “I know that
p.” Suppose that Austin was to say that there is something conceptually
primary about “I know that p” as against “He knows that p” – so that “I
know that p” does not need to be adverting to an independently consti-
tuted state of knowing, with “He knows that p” being understood only
as saying that if he was to say “I know that p” he would be doing what-
ever one does say via saying “I know that p.” But then Chisholm would
insist on the following (1966: 17): “To suppose that the performance of
the nondescriptive function is inconsistent with a simultaneous per-
formance of the descriptive function might be called . . . an example of
the performative fallacy.” The point is that performatives work because of
circumstances beyond themselves – and these circumstances (Chisholm
would say) are either implicitly or explicitly being described by success-
ful uses of “I know.”
209

Knowledge and Knowledge-Claims: Austin and Beyond 209

So, there seems to be a stand-off. Did Chisholm reveal Austin to have
raised merely an interesting, but not deeply explanatory and not rationally
inescapable, possibility only about one restricted kind of knowledge-talk?

3 Generalising Austin’s Proposal

3.1 Doing More Things with More Words

Austin’s 1946 “Other Minds” was followed by his 1955 William James
Lectures at Harvard, published as How To Do Things with Words (Austin
1975 – for short: HDW). As we saw a moment ago, “Other Minds” intro-
duced us to the idea of “I know that p” being used as a performative.
But HDW generalised the idea of some utterances being performatives. It
enlarged upon both the range of utterances being discussed and the range
of categorisations being offered of them. For instance, the notion of a per-
formative utterance became absorbed within the more general category
of illocutionary forces that an utterance can possess. Previously, Austin
acknowledges, he had offered us simply “a dichotomy of performatives
and constatives [roughly: purely descriptive utterances], which we see [he
now realises] had to be abandoned in favour of more general families of
related and overlapping speech acts” (1975: 150). Austin then concludes
HDW with the details of his newly described “attempt to classify” those
families.
In doing this, he maintains his earlier focus upon “the first person sin-
gular present indicative active form” (1975:  150). With some hesitation
(1975: 151, 152), he proceeds to offer us “five very general classes” (1975: 151).
The present chapter, however, is not an account of Austin’s philosophy of
language in general; my emphasis is upon his philosophy of knowledge-
language in particular. Accordingly, I will focus upon just two of those five
classes of Austin’s. These two are the ones most clearly applicable to under-
standing knowledge-talk. They are Category 1 (Verdictives) and Category
5 (Expositives). And here is Austin’s “rough idea of each”:
The first, verdictives, are typified by the giving of a verdict, . . . by a jury,
arbitrator, or umpire. But they need not be final; they may be, for example,
an estimate, reckoning, or appraisal. It is essentially giving a finding as to
something – fact, or value – which is for different reasons hard to be certain
about. (1975: 151)
The fifth, expositives, are difficult to define. They make plain how our utter-
ances fit into the course of an argument or conversation, how we are using
words, or, in general, are expository. (1975: 152)
0

210 Stephen Hetherington

At least when used properly, verdictives are responses to evidence. We
could call them epistemic responses, as against simply responses as such.
They bear correlatively and consequently upon the verdict’s truth or fal-
sity (1975: 153). Instances of expositives could also sometimes be regarded
as verdictives (1975: 161); indeed, some could fit into the other categor-
ies, too. So, the category of expositives in particular is a loosely grouped
one. Nevertheless, regardless of where they are to be located within
Austin’s taxonomy, expositives are at least part of our striving to use and
to develop evidence – by considering it and/or marshalling it and/or etc.
Austin describes expositives as being “the clarifying of reasons, arguments,
and communications” (1975: 163). It is unsurprising, then, that the closest
he approaches to telling us directly about uses of “know” is his including
it – albeit preceded by a “?”– in his “list of expositives” (1975: 162). But
it is surprising, given the centrality of “I know” to his initial account in
“Other Minds,” that so little is said by Austin – and, even then, it is only a
tentative and passing suggestion on his part – about how “knows” should
be understood within HDW’s expanded account. Still, let us see what
epistemological progress can be made by expanding upon this suggestion
of Austin’s.

3.2 Knowledge-Practicalism
I believe that some such progress is possible; for Austin has said enough, we
will find, to allow us to enrich reciprocally both his account and a recent
epistemological proposal as to the nature of knowledge. The proposal in
question is mine (Hetherington 2011a, 2011b, 2013b). I call it practicalism,3
and section 3.3 will indicate how we might blend it with Austin’s own
account of knowledge-talk (from section 3.1). First, though, I will provide a
brief overview of practicalism on its usual non-Austinian terms. This over-
view will highlight the category of what I call a knowing action.
Austin was wary of reifying knowledge as an inner state, or indeed
a special state at all, of a person; and I  share that caution. Accordingly,
knowledge-practicalism is a theory that conceives of knowledge’s nature
as something somehow diffused and decentralised, even for a particular
person’s knowledge of a particular fact. An example will illustrate the gen-
eral approach being envisaged. Suppose that you know that you are seeing

3
It is not the ideal name. But I use it because (as I will explain) the core of the proposal is about
knowledge-how – which has often been called practical knowledge, to distinguish it from theoretical
or contemplative knowledge, say.
211

Knowledge and Knowledge-Claims: Austin and Beyond 211

a kookaburra. Practicalism allows that this knowledge need not be a belief,
although you could in fact have an appropriate belief on a given occasion
of having the knowledge. Nor (allows practicalism) need the knowledge be
based on evidence, although you could in fact be using such evidence on a
given occasion of having the knowledge. What, then, is knowledge – if it is
not obliged to be a belief and based on evidence? How is it that knowledge
could be something that need not be a belief and need not be based on
evidence? The key to knowledge-practicalism is its view of knowledge as
always being, most fundamentally, an ability or power (most likely, a com-
plex one in practice). And I believe that the most helpful way to conceive
of knowledge, towards that interpretive end, is as always being some sort
of knowledge-how. More fully, the idea is that any instance of knowledge-
that is some knowledge-how to act in some or all of the ways that are typ-
ically indicative of the presence of the pertinent knowledge-that.
Thus (to continue the example at hand), practicalism tells us that in
knowing that you are seeing a kookaburra (and not merely as a causal
consequence of your having this knowledge), most likely you have much
knowledge-how:  you know how to reason and/or to respond and/or to
question and/or to answer and/or to move and/or etc. More specifically,
you know how to perform some or all of these actions, in ways that are
directed relevantly at the specific truth that you are seeing a kookaburra.
Note, too, that we can even include within these options your knowing
how to do some or all of whatever actions are typically indicative of the
presence of the belief that you are seeing a kookaburra. In other words,
knowledge-practicalism does not deny that knowing can include believ-
ing:  no Platonic incompatibility between knowing that p and believing
that p is being thrust upon us by a practicalist conception of the former.
Asked whether you are seeing a kookaburra, “yes” is probably your reply.
Asked by someone else what you are seeing, your answer is almost cer-
tainly “A kookaburra”. Maybe you know how to describe a kookaburra, for
instance, so as to distinguish it at least somewhat from other Australian
birds. Similarly, you might – although, again, you need not – have in mind
a more or less detailed visual representation of the bird. And so on, for one
relevant-but-optional ability or power after another.
What then links together all of those pieces of knowledge-how (these
abilities or powers, as I am assuming knowledge-how in general to be),
so that they can coalesce into a given instance of knowledge-that?4 What

4
See Snowdon (2003) for doubts about knowledge-how’s being an ability, at any rate. See
Hetherington (2011a: 46–47) for a reply to such doubts.
2

212 Stephen Hetherington

makes them “relevant” (even if optional) for the example in question? The
answer is that all of them are directed, in whatever way is apt for the spe-
cific abilities, at a shared p – in this case, the proposition that you are see-
ing a kookaburra. And then we may begin to conceive of your knowledge
that you are seeing a kookaburra as being a form of knowledge-how on
your part. That is, practicalism implies that in knowing even a particular
truth p, one has knowledge-how  – potentially quite a lot of it, gener-
ating a complex combination in the particular case of these potentially
distinguishable cases of knowledge-how, all of them somehow bearing
upon that particular truth that p. Many epistemologists will concede that,
when one knows that p, one can, as a result, have some or all of those
abilities:  “Knowing that p allows you to describe, to question, to rep-
resent, etc.” But knowledge-practicalism’s claim is not merely causal in
that way. Instead, it is constitutive; and in that sense, it is metaphysically
stronger. It says that the knowledge that p is the knowledge-how, as this
has been characterised. And it says that, in describing such a concatena-
tion of abilities, this is all that we need to say if we are to capture what it
is to know that p.
On that practicalist approach, therefore, any instance of knowledge that p
is in effect nothing beyond a potential – a principled potential – for various
p-related actions. These actions can vary from person to person, and from
moment to moment, even for a single p: generally, not everyone will know
that p in the same ways; and often a given person can know that p in dif-
ferent ways at different times. That sort of variability, across persons and/or
times, is standardly registered by epistemologists’ allowing there to be evi-
dential disparities across those persons and/or times. Practicalism, too, allows
for those evidential disparities. But practicalism also allows for many more
epistemically relevant disparities to be constitutive of the presence of different
instances, across persons and/or times, of knowledge that p (for a given p).
What practicalism thereby directs us to are actual or possible actions that do
or would manifest or express the associated knowledge-how – and, ipso facto,
the knowledge-that itself. Two people could have the same evidence for p,
say, even as one of them is better at answering questions arising from that evi-
dence: she might, in that way, know p more deeply than the other person does.
And (to return to the important example of knowledge and its relationship to
belief) when you believe that p, this could itself be an expression on your part
of the knowledge that p – as are (I hope we will agree) your alerting others to
its being the case that p, your gathering evidence for p, and so on.
I should expand on the fact that this knowledge-practicalist
suggestion – which, I realise, is only being sketched, not established, by
213

Knowledge and Knowledge-Claims: Austin and Beyond 213

my present remarks5 – would reverse a piece of epistemological orthodoxy.
Epistemology has long assumed that a belief that p would be a constitu-
ent of any instance of knowledge that p: the belief that p is, in effect, the
“thing” that is – once it is suitably embellished with various epistemic
properties – the knowledge that p. Recent years, though, have seen a less
widespread epistemological commitment to at least part of that picture,
largely due to Timothy Williamson’s (2000: ch. 1) rejection of the project
of analysing the concept of knowledge in order to list knowledge’s key con-
stituents, such as belief. My suggestion here does not depend on or borrow
from Williamson’s thinking. Still, we should be correlatively emboldened
to consider alternative relationships between knowledge and belief; and
knowledge-practicalism offers us one such approach. On this approach,
knowledge is not a belief, the possession of which enables one – once the
belief has those desired further epistemic properties – to perform vari-
ous actions (cognitive or otherwise); rather, knowledge is whatever complex
ability enables one to perform such actions – including the belief-enabled
ones but even, I suggest, including the believing itself.
Those ways of acting are thus means of expressing or manifesting one’s
knowledge that p, once this has been conceived of as knowledge-how (the
complex sort of ability gestured at just now). And many of those ways of
acting then fall into the category of what I have elsewhere called know-
ing actions (2013b). The distinctive feature of such actions is not (as their
name might convey) that they are automatically and infallibly indicators
of one’s actually having knowledge. Rather, their distinctive feature is that
they typically tend  – or are even typically intended  – to convey knowl-
edge. For instance, riding a bicycle manifests knowledge-how; not typi-
cally, however, so as to convey one’s having the knowledge-how. In contrast
to that example, answering “yes” to “Are you looking at a kookaburra?” is
typically intended to convey one’s knowledge that one is seeing a kooka-
burra. Hence, when one does in fact have that knowledge, such an action –
providing that answer  – will in fact typically succeed in conveying this
knowledge. So, riding a bicycle is not a knowing action, even though it is a
skilled action, one that manifests or expresses the knowledge-how to ride a
bicycle. In contrast, answering “yes” to “Are you looking at a kookaburra?”
is a knowing action; for it is a skilled action, one that manifests or expresses
some knowledge-how so as to convey – skilfully – the possession of this
knowledge-how.

5
Again, though, a fuller argument on behalf of a practicalist conception of knowledge-possession may
be found elsewhere (Hetherington 2011a, 2011b, 2013b).
4

214 Stephen Hetherington

3.3 An Austinian Enrichment of Knowledge-Practicalism

Section 3.2’s overview of knowledge-practicalism is significant in this discus-
sion of Austin because many – maybe most – of those kinds of action that
would be regarded by knowledge-practicalism as manifesting or express-
ing one’s knowledge that p will, it seems, be linguistic in some way. I have
mentioned a few of them – one’s answering a question, one’s describing a
state of affairs, etc. Some of these will be overt utterances to an audience.
Yet even when such an utterance does not actually occur, knowledge-
practicalism countenances possible such utterances. And, of course, there
is soliloquising, again both actual and possible. More generally, then, there
will be actual or possible spoken or thought utterances to oneself or to
others. Moreover, many – maybe most – of these actual or possible uses
of language will be knowing actions (the category of action introduced a
moment ago, at the close of section 3.2). So, this is where Austin’s account
(in section 3.1) is especially helpful to knowledge-practicalism, because
he is describing what we are free to interpret as some of the knowledge-
practicalism’s microstructure – details within the programmatic structure
I have sketched. By the same token, this is also, I propose, where Austin’s
account can strengthen its own credentials insofar as we wish it to be an
account that contributes to our understanding of knowledge as such, rather
than merely our understanding of knowledge-talk.
Recall Austin’s discussion of verdictives and expositives. A knowledge-
practicalist can say that what Austin’s discussion of these gives us is an
insightful and fecund taxonomy of at least some of the possible ways for
one’s knowledge that p to be expressed or manifested. Obviously, these
ways are actions. Indeed (to repeat my point of a moment ago), they are
knowing actions. What I did not also mention a moment ago is that know-
ing actions are collectively a species within the genus of what Austin’s
Oxford colleague Gilbert Ryle (1945, 1949) called intelligent actions. For
Ryle, an action is intelligent (in his intended sense) insofar as it is express-
ing or manifesting knowledge-how. What practicalism then tells us is that
knowledge-how is what knowledge-that is; and so, at least sometimes,
our actions – some of our intelligent ones – are manifesting or express-
ing knowledge-that in manifesting or expressing knowledge-how. Which
intelligent actions do this? The knowing (intelligent) actions do so, I am
saying. In contrast, for example, riding a bicycle is an intelligent action,
but not a knowing one (as section 3.2 mentioned). Once more, though,
answering the question “What are you seeing?” with “A kookaburra” in the
situation described earlier is a knowing action (and thereby an intelligent
215

Knowledge and Knowledge-Claims: Austin and Beyond 215

one). And if knowledge-practicalism is right, then the latter action is an
expression or manifestation of some knowledge-how – and thereby of some
knowledge-that, given the action’s being a knowing one. This is a generic
picture, I acknowledge. As I will now explain, however, we can render it
less so, with Austin’s help.
Consider, first, his class of verdictives. Austin said that these can, but
need not, be “official” (1975: 153). What matters more right now, I add, is
what makes them also a species of knowing action – namely, their being
intended to convey knowledge. Sometimes, verdictives, such as the follow-
ing ones (1975: 153), also make true the known truth:
acquit, convict, find (as a matter of fact), hold (as a matter of law), grade,
rank, rate, assess, value
Sometimes, verdictives, such as the following ones (1975: 153), are attempts
more to decide on the evidential support that is present for the known truth:
calculate, reckon, estimate, describe, characterise, diagnose, analyse
In each of these instances, though, the same general point persists – which
is that it is natural to regard the relevant utterances of verdictives as actions
that at least can be manifesting or expressing one’s knowing that p.
The same general point is true of expositives. Austin’s grouping of his
examples from within this category is not accompanied by any explanation
of the grouping’s details. Here are just some of those putative expositives
(1975: 162–163):
affirm, deny, describe, analyse [these last two appeared also in Austin’s list
of verdictives], class, identify, testify, report, inform, apprise, tell, answer,
accept, concede, agree, correct, revise, postulate, deduce, interpret, distin-
guish, explain, understand
Section 3.1 noted that Austin also includes “know” here, while preceding
it with “?” (1975: 162). The same is true of “doubt” and “believe”: they
appear here, preceded by “?” (1975: 162). That is significant, since these two
terms are traditionally linked by epistemologists with the term “know.”
Including them here is appropriate (and so I am not sure why Austin
includes the “?”).
Let us take stock. Austin is highlighting many terms that are more or less
openly epistemic in their content. And, of course, it is various uses of these
terms with which he is concerned. Now, knowledge-practicalism regards
knowledge itself in terms of what one does or can do – actual or possible
actions, presumably some (even if not all) of them overtly cognitive. And
6

216 Stephen Hetherington

Austin is directing our attention to various relevant types of action: they
are ones with epistemic connotations or denotations, ones that are per-
formed with words. So, my correlatively constructive suggestion is this:
What Austin has thereby described is in fact much of what one can do in
having knowledge. That is, he has described actual or possible linguistic
actions that are typically performed not simply by having knowledge, or as
a result of having it, but instead in having it. And this tells us, in turn, that
the pertinent part of his discussion is revealingly and helpfully congruent
with knowledge-practicalism.
The epistemological moral is thus clear. We can derive an enriched – a more
detailed and less programmatic – knowledge-practicalism, by allowing it
to be a more Austinian knowledge-practicalism  – one that draws upon
some of the linguistic specificities described and taxonomised by Austin,
especially in HDW.6
As I observed earlier, this result is also methodologically significant for
our understanding, in turn, of the epistemological potential in Austin’s
thinking. It enables us to regard his comments on knowledge-talk as being –
immediately and transparently – a contribution to our understanding of
knowledge itself, insofar as (according to knowledge-practicalism) knowl-
edge is inherently a potential to perform various sorts of action, including
ones that are instances of knowledge-talking. Austin’s account (as I  am
interpreting it) of knowledge-talk is directly about a significant proportion
of those kinds of action. Hence, once we combine his account with my
knowledge-practicalism, we may regard his account as telling us directly
about what – it transpires – is a conceptually essential component of what
knowledge itself is. It would be easy for epistemologists to overlook this
potential understanding of Austin’s account – and, as a result, not to do
justice to the epistemological potential within his account. But knowledge-
practicalism bridges that gap. It allows us to rectify that potential lack of
epistemological understanding, by treating Austin-on-knowledge-talk as
Austin-on-part-of-what-it-typically-is-to-know.

4 Knowledge and Fallibility

As it happens, too, knowledge-practicalism lends itself easily to a
knowledge-fallibilism. Insofar as any instance of knowledge-that is an
6
In a similar vein, Wittgenstein’s discussion of rule-following may also be understood in knowledge-
practicalist terms (Hetherington 2011a: sec. 2.3). Ryle, Wittgenstein, and Austin: three philosophical
giants of that period – and each of them interpretable to some extent in terms of his reaction to
questions about knowledge-that and knowledge-how. (But I will not pursue that point further in
this chapter.)
217

Knowledge and Knowledge-Claims: Austin and Beyond 217

instance of knowledge-how – an ability or power, say – it also admits of
fallibility, if the abilities or powers themselves admit of it. And presum-
ably they do: in general, it seems, our abilities or powers are fallible. For
example, most people know how to ride a bicycle; surely only ever fallibly,
though, with imperfections and limitations in their cycling techniques and
skills. This is true of professional cyclists, too. We might not even know
what it would be to have a perfect cycling skill – in effect, an infallible
cycling skill.
Correlatively, if section 3.3 is correct to interpret Austin’s views on
knowledge-talk as blending naturally and advantageously with a practical-
ism about knowledge’s nature, we might expect to find Austin endorsing a
sort of knowledge-fallibilism. And so he claims to do. This emerges, as we
might also expect, in some of his further comments on knowledge-talk. In
particular, what we find is Austin commenting on the phenomenon of a
person’s claiming to know a truth even while – with the same breath – con-
ceding what amounts to a kind of fallibility in the knowing of that truth.
More specifically, it is the phenomenon of utterances with this form: “I
know that p although I might be mistaken as to p.”
That sort of combination is generally called a concessive knowledge-
attribution, and it has been much discussed by epistemologists in recent
years.7 Those discussions have been directed at the question of whether
such utterances are consistent. In that way, the issue has been treated
as a test case that needs to be confronted by any would-be knowledge-
fallibilist – a case that tests the consistency of knowledge-fallibilism itself.
Any utterance with the form “I know that p although I might be mistaken
as to p” could be offered in the spirit simply of acknowledging the fallibil-
ity that there is in knowing that p. Accordingly, the question of whether
knowledge-fallibilism is true8 could be thought to depend, for a start, on
whether there can be consistent utterances with that form.
As I said, this is an issue currently before the court of epistemological
opinion. It is also an issue on which – before it had been formalised into
the general epistemological issue that it now is – Austin had something
to say. Again, his key remarks are in “Other Minds” (1946; 1979: 76–116).
He is considering the question “How do you know?” as it is used in spe-
cific contexts, when being addressed to specific people. In practice, reasons

7
See, e.g., Stanley (2005), Dougherty and Rysiew (2009), Dodd (2010), and Hetherington (2013a).
8
We may regard knowledge-fallibilism for present purposes as the thesis that it is possible for at least
some instances of knowledge to be fallible – and that it is thereby possible for at least some instances
of knowledge to be such that there are relevant possible situations where the same evidence, used in
the same way, leads to a false belief. For more on how we should formulate knowledge-fallibilism,
see Hetherington (1999, 2001, 2002, 2016b, forthcoming).
8

218 Stephen Hetherington

for questioning, let alone for doubting, whether one’s interlocutor has
the knowledge she claims to have will generally be cited or expected; and
these reasons can take various forms in various settings (a fact upon which
Austin enlarges also in Sense and Sensibilia [1962, ch. X]). An acknowledge-
ment simply of one’s epistemic fallibility per se, of course, allows or even
encourages one to append “although I might be wrong [as to p]” to any
of one’s claims of the form “I know that p” – yet without thereby feeling
that one is undermining the aptness of the knowledge-claim. And this
combination is possible (Austin-the-avowed-fallibilist would say) because
the fallibility per se is not a barrier to knowing: since ordinary language
displays such a pattern, we should respect it, thereby regarding ourselves –
even when we claim knowledge that p – as simultaneously living with the
possibility that not-p.
But there is more to be said about this issue, particularly because we
do not standardly concede simply our generic fallibility as such on those
occasions. We often say why or how we are fallible in the specific setting.
Austin (1946: 169–173; 1979: 97–101) centres his discussion of the issue
upon a contrary form of claim, “If I know, I can’t be wrong”;9 and this
leads into his account (discussed in section 3.1) that “I know” amounts
to giving one’s guarantee – making a kind of promise. What we may now
add to the earlier exposition is that Austin is not saying that the promise in
question is that one will be able to transcend the fallibility as such that one
embodies – the fallibility that is inherent within any “human intellect and
senses” (1946: 169; 1979: 98). Rather, Austin is saying that the promise in
your saying “I know” is that you have no “concrete reason to suppose that
you may be mistaken in this case” (1946: 170; 1979: 98; my emphasis). That
is how we are to understand the following:
You are prohibited from saying ‘I know it is so, but I may be wrong,’ just
as you are prohibited from saying ‘I promise I will, but I may fail.’ If you
are aware you may be mistaken, you ought not to say you know, just as, if
you are aware you may break your word, you have no business to promise.
(1946: 170; 1979: 98)
Now, is this a fallibilist way of thinking about knowledge? Austin
clearly believes that it is (as does Krista Lawlor [2013: 17]), even though

9
His account in Sense and Sensibilia (1962)  – written particularly in response to some of Ayer’s
work (1940) – argued that there is no class of sentences that are so epistemically privileged as to
be incorrigible: no such class awaits us at the foundations (if there are such things or places) of our
empirical knowledge. These days, epistemologists speak more of fallibility-versus-infallibility than of
corrigibility-versus-incorrigibility. The epistemological spirit remains the same, though – a concern
with whether even ordinary knowledge must somehow rest upon a kind of extraordinary knowledge.
219

Knowledge and Knowledge-Claims: Austin and Beyond 219

he is endorsing the idea that knowing precludes the possibility of being
mistaken  – a form of words that connotes knowledge-infallibilism. The
underlying thinking (Austin would say) remains fallibilist because the
possibility in question is one that even a knowledge-fallibilist would,
in the circumstance, allow to be an appropriate reason for denying the
knowledge-claim. This is because the possibility in question would be a
“concrete reason” for that denial.10
Contemporary epistemologists will espy a familiar structure to that
analysis; and many will be comfortable with Austin’s apparent view of
himself as accommodating an interpretation of people, even when they
know a truth, as doing this fallibly. Think of those epistemologists, in
particular, who would say that knowing that p precludes there being any
relevant alternatives to p that are not eliminated by one’s evidence for p.11
Lawlor (2013: 45) says, correctly, that “Austin’s ‘Other Minds’ is an early
statement of relevant alternatives theories of knowledge.” The language
used in relevant alternatives accounts is new since Austin’s time, but their
accompanying idea remains Austinian in spirit. As Austin famously said
(1946: 156; 1979: 84),
Enough is enough: it doesn’t mean everything. Enough means enough to
show that (within reason, and for present intents and purposes) it ‘can’t’
be anything else, there is no room for an alternative, competing, descrip-
tion of it. It does not mean, for example, enough to show it isn’t a stuffed
goldfinch.
Fred Dretske (1970) advocated much the same moral with an example
about the possibility of disguised mules, when discussing what is needed
for knowing that one is looking at some zebras in a zoo. But, as I said, the
same general epistemological message persists (even if, as Lawlor explains,
Austin’s is stated in terms of reasonable alternatives rather than relevant
ones). And seemingly that message is intended to be fallibilist, even
though it tells us that knowledge requires that all alternatives or concrete
reasons for doubt be eliminated or removed (remember Austin’s “I can’t
be wrong”). This is nonetheless intended to be a fallibilist picture, because
those alternatives or reasons for doubt are not required to include all of
the possible merely-because-we-are-fallible alternatives or reasons for doubt
(remember Austin’s expectation that the reasons for doubt are generally to
be “concrete” ones).

10
For more on Austin on this, see Warnock (1989: 33–4) and Kaplan (2006).
11
See Dretske (1970) and Goldman (1976) for seminal motivations of the idea of a relevant alterna-
tive, and for accounts of that idea’s significance within epistemological analyses of knowing.
0

220 Stephen Hetherington

As far as it takes us, such an approach – by Austin and, later, by Dretske
and others12 – deserves epistemological respect. It was at least sufficient for
Austin’s immediate purpose – of establishing a defensive strategy, in terms
of knowledge-talk, against the argument from illusion, which was one of
his recurring philosophical foes (see, especially, Sense and Sensibilia [1962]).
But what epistemological price did Austin pay for framing his response in
that way? In particular, does he succeed in retaining a fallibilist conception
of knowing, insofar as this is to be revealed by reflection upon how we
claim or deny knowledge?
This is not the place to attempt to answer that question fully. But I will
offer what I hope is one helpful line of thought. Although Austin does talk
here of reasons, he does not develop a systematic account of what makes
for epistemic reasonableness – more particularly, of what constitutes suf-
ficient concreteness in a reason for doubt. All that he does seem to require
are some minimal conditions, of actuality and realistic specificity: judging
by his discussions, some particular reason will have been mentioned, and
it will have described a realistic way in which the evidence that has been
taken to support p is nonetheless compatible with not-p. The emphasis
by Austin would then be on the way in which that compatibility obtains,
not on the statement of compatibility as such; for he wants specificity –
further detail – in the doubt and thereby in that compatibility.
Thus, when discussing one’s asking “How do you know it’s a goldfinch?”
(1946: 154; 1979: 83), Austin considers a few possible objections to the asso-
ciated knowledge-claim, one of which is this:
(2) But that’s not enough:  plenty of other birds have red heads. What
you say doesn’t prove it. For all you know, it may be a woodpecker.
(1946: 155; 1979: 83)
Austin continues:
It is in the case of objection (2) that you would be more inclined to say right
out ‘Then you don’t know.’ Because it doesn’t prove it, it’s not enough to
prove it. (1946: 155; 1979: 84)
And
[i]f you say ‘That’s not enough,’ then you must have in mind some more
or less definite lack. . . . If there is no definite lack, which you are at least
prepared to specify as being present, then it’s silly (outrageous) just to go on
saying ‘That’s not enough.’ (1946: 155–156; 1979: 84)

12
For a recent endorsement and development of the idea as it is formulated in the contemporary
idiom of relevant alternatives, see Bradley (2014).
221

Knowledge and Knowledge-Claims: Austin and Beyond 221

Still, what makes this a fallibilist (rather than infallibilist) conception of
what is required if one is to claim to know that p? Here I hesitate. Again,
what seems to be regarded by Austin as sufficient for ‘keeping it fallibil-
ist’ (as we might now say) is the fact of a specific possible circumstance’s
actually being mentioned in what is recognisably a sensible conversation.13
Yet no knowledge-fallibilist, I suggest, should accept even that combina-
tion as a sufficient ground for denying a knowledge-claim. Far from it; we
might indeed say that it is precisely a mark of knowledge-fallibilism that it
allows people to have – and to claim – knowledge that p even when they
acknowledge explicitly, and in specific terms, that their evidence for p is at
least somewhat mixed (that is, when not all of it is in support of p, even if
most of it is). No knowledge-fallibilist should be committed to expecting
that the path of inquiry, say, to her claiming some particular knowledge
will need to have been free of all distracting or potentially misleading data.
Nor should she be committed to expecting that she will be able – once she
decides that it is true that p – to gaze back upon the path of inquiry along
which she has travelled, only to see it now swept clear of all such distract-
ing or potentially misleading data (those pieces of evidence that might
have – but have not – diverted her away from what she has now concluded
is the truth that p). Such is the cognitive life of an avowed fallibilist – a
form of life that can forever remain tolerant of explicitly acknowledged
alternative and conflicting evidence. Such is the cognitive life of an avowed
fallibilist – a form of life available also to those who are prepared to claim
knowledge on various occasions of admitting their fallibility.
Austin, however, does not do justice to that form of cognitive life. And
this suggests that, ultimately, any knowledge-fallibilist might do well to
regard Austin as just another unwitting knowledge-infallibilist wearing
some fallibilist garb.14 Seemingly, Austin did not think of himself in that
way; but this could have been a lack of some self-knowledge on his part.
It is a committed knowledge-fallibilist who will say, “I do know that p,

13
In this respect, the thinking here is clearly a precursor to at least some contemporary contextualist
attempts to treat the mere mentioning of a sceptical possibility, for example, as sufficing to make
that possibility relevant to whatever knowledge-claim is being considered. The idea is that even if a
person can truly be said, in a normal setting, to know that p, she might truly be denied knowledge
that p once the question explicitly arises in conversation of whether she knows that she is not a
brain in a vat, say, being unwittingly manipulated to believe that p. (For a clear instance of this
contextualist idea, see Lewis 1996.) See section 5 for a little more on contextualism and Austin.
(Only a little more, though; for extended discussion, see Lawlor [2013: ch. 3].)
14
I say “just another” because the phenomenon of unwittingly responding as an infallibilist, even
while regarding oneself as a fallibilist, about knowledge is more common among epistemologists, it
seems to me, than is widely realised. For a case study partly of this phenomenon, see Hetherington
(2016a).
2

222 Stephen Hetherington

although my evidence does include some data in favour of not-p. I might
be mistaken, then, in believing that p.  For now, though, I  maintain  –
fallibly, I accept – that I do know that p.” Austin’s approach is fundamen-
tally inhospitable to this conception of how a knowledge-fallibilist might
at times – and indeed should – wish to talk.

5 Conclusion
Austin’s influence on contemporary epistemology has been indirect but
significant. When current contextualists, for instance, sketch the history of
their general approach, many stop short of mentioning Wittgenstein and
Austin. Yet seemingly this is historical myopia on their part. Wittgenstein’s
(1969) version of contextualism generated a line of discussion focused
especially on questions about knowledge’s foundations and epistemic
regress: witness the investigations and applications of the Wittgensteinian
idea by David Annis (1978), Crispin Wright (1985), and Michael Williams
(2001). But that particular pattern of contextualist discussion is not the one
that has taken an increasingly determinate and discussed shape in recent
years at the hands of Stewart Cohen (e.g. 1986), David Lewis (1996), and
Keith DeRose (e.g. 2009) – a tradition that, whenever the question arises,
is usually said to stretch as far back as Dretske (1970). We saw, however (in
section 4), that the Dretskean approach is apparently indebted, in turn,
to what may now be recognised as having been Austin’s own form – even
if it was a proto-form – of contextualism about at least some knowledge-
attributions. As to whether this approach is one that epistemologists would
do well to encourage far into the future, that is a separate question, and not
one that I have examined in this chapter.
What I have offered is an account of how Austin’s influence on contem-
porary epistemology has the potential to be even more pronounced than
it has been until now. My account has been gestural and programmatic,
but optimistic, mainly with section 3’s explanation of what I  regard as
Austin’s actual and latent contribution to our epistemological understand-
ing of knowledge itself (rather than just of knowledge-talk), courtesy of
his approach’s congruence with knowledge-practicalism. Austin himself, of
course, did not describe that potential congruence. In this respect, section 3
adapts and extends what Austin did say. And we thereby gain access to a
line of thought that aims to strengthen the epistemologically distinctive
insight behind Austin’s account of utterances of “I know” as performatives.
My hypothesis, again, is that knowledge-practicalism enables Austin-on-
knowledge-talk to become Austin-on-knowledge-itself.
223

Bibliography

Alston, W. P. (2000). Illocutionary Acts and Sentence Meaning. Ithaca, NY: Cornell

University Press.
Annis, D. B. (1978). “A Contextualist Theory of Epistemic Justification.” American
Philosophical Quarterly 15: 213–219.
Anscombe, G. E. M. (1959). An Introduction to Wittgenstein’s Tractatus. London:
Hutchinson.
Atherton, M. (1990). Berkeley’s Revolution in Vision. Ithaca, NY:  Cornell
University Press.
Austin, J. L. (1946). “Other Minds.” Proceedings of the Aristotelian Society,
Supplementary Volume 20: 148–187.
(1950). “Truth.” Proceedings of the Aristotelian Society, Supplementary Volume
24: 111–128.
(1957). “A Plea for Excuses.” Proceedings of the Aristotelian Society 57: 1–30.
(1958). “Pretending.” Proceedings of the Aristotelian Society, Supplementary
Volume 32: 261–278.
(1962). Sense and Sensibilia. Reconstructed from the Manuscript Notes by G. J.
Warnock. Oxford: Oxford University Press.
(1963). “Performative-Constative,” tr. G.  J. Warnock. In C.  E. Caton (ed.),
Philosophy and Ordinary Language, pp. 22–33. Chicago:  University of
Illinois Press. (Originally published as “Performatif-Constatif ” in Cahiers
de Royaumont, Philosophie No. IV, La Philosophie Analytique, Paris: Minuit,
1962.)
(1966). “Three Ways of Spilling Ink.” The Philosophical Review 75: 427–440.
(1970). “Intelligent Behaviour:  A  Critical Review of The Concept of Mind.”
In O. P. Wood and G. Pitcher (eds.), Ryle:  A  Collection of Critical Essays,
pp. 45–51. New York: Doubleday. (Originally published in The Times Literary
Supplement, April 7, 1950.)
(1975). How to Do Things with Words, 2nd edn., ed. by J.  O. Urmson and
M. Sbisà. Oxford: Oxford University Press. (1st edn., 1962.)
(1979). Philosophical Papers, 3rd edn., ed. by J. O. Urmson and G. J. Warnock.
Oxford: Oxford University Press. (1st edn., 1961.)
Ayer, A. J. (1940). The Foundations of Empirical Knowledge. London: Macmillan.
(1967). “Has Austin Refuted the Sense-Datum Theory?” Synthese 17: 117–140.

223
4

224 Bibliography
Bach, K. (1975). “Performatives Are Statements Too.” Philosophical Studies 28:
229–236.
Bach, K. and Harnish, R. (1979). Linguistic Communication and Speech Acts.
Cambridge, MA: The MIT Press.
Barwise, J. and Etchemendy, J. (1987). The Liar: An Essay on Truth and Circularity.
Oxford: Oxford University Press.
Barwise, J. and Perry, J. (1983). Situations and Attitudes. Cambridge, MA:  The
MIT Press.
Bauer, N. (2015). How to Do Things with Pornography. Cambridge, MA: Harvard
University Press.
Baz, A. (2012). When Words Are Called For:  A  Defense of Ordinary Language
Philosophy. Cambridge, MA: Harvard University Press.
Beardsley, M. C. (1962). “The Metaphorical Twist.” Philosophy and Phenomeno-
logical Research 22: 293–307.
Berkeley, G. (1948–1957). The Works of George Berkeley, Bishop of Cloyne, ed. by
A. A. Luce and T. E. Jessop. Vols. 1–9. Edinburgh: Thomas Nelson.
Bird, A. (2002). “Illocutionary Silencing.” Pacific Philosophical Quarterly 83: 1–15.
Black, M. (1962). Models and Metaphors:  Studies in Language and Philosophy.
Ithaca, NY: Cornell University Press.
Bradley, D. (2014). “A Relevant Alternatives Solution to the Bootstrapping and
Self-Knowledge Problems.” The Journal of Philosophy 111: 379–393.
Brandom, R. (1994). Making It Explicit:  Reasoning, Representing, and Discursive
Commitment. Cambridge, MA: Harvard University Press.
Butler, J. (1997). Excitable Speech: A Politics of the Performative. London: Routledge.
Byrne, A. and Logue, H. (eds.) (2009). Disjunctivism:  Contemporary Readings.
Cambridge, MA: The MIT Press.
Cappelen, H. and Lepore, E. (2007). Language Turned on Itself: The Semantics and
Pragmatics of Metalinguistic Discourse. Oxford: Oxford University Press.
Cattell, R. (1978). “On the Source of Interrogative Adverbs.” Language 54:
61–77.
Cavell, S. (1969). Must We Mean What We Say? Cambridge: Cambridge University
Press.
(1979). The Claim of Reason:  Wittgenstein, Skepticism, Morality, and Tragedy.
Oxford: Oxford University Press.
(1984). Themes Out of School: Effects and Causes. Chicago: University of Chicago
Press.
(1994). A Pitch of Philosophy:  Autobiographical Exercises. Cambridge, MA:
Harvard University Press.
(1997). Contesting Tears:  The Hollywood Melodrama of the Unknown Woman.
Chicago: University of Chicago Press.
Chaucer, G. (2008). Troilus and Criseyde, ed. by L. D. Benson. Third Edition.
Oxford: Oxford University Press.
Chisholm, R. M. (1966). Theory of Knowledge. Englewood Cliffs, NJ: Prentice-Hall.
(1969). “Austin’s Philosophical Papers.” In K. T. Fann (ed.), Symposium on J. L.
Austin, pp. 101–106. London: Routledge.
225

Bibliography 225
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA:  The
MIT Press.
Clark, H. H. (1996). Using Language. Cambridge: Cambridge University Press.
(2004). “Pragmatics of Language Performance.” In L. R. Horn and G. Ward
(eds.), The Handbook of Pragmatics, pp. 365–382. Oxford: Blackwell.
Clark, H. H. and Carlson, T. B. (1982). “Speech Acts and Hearers’ Beliefs.” In
N. V. Smith (ed.), Mutual Knowledge, pp. 1–36. New York: Academic Press.
Clark, H. H. and Gerrig, R. (1990). “Quotations as Demonstrations.” Language
66: 764–805.
Clarke, T. (1972). “The Legacy of Skepticism.” The Journal of Philosophy 69: 754–769.
Cohen, S. (1986). “Knowledge and Context.” The Journal of Philosophy 83: 574–585.
(1988). “How to Be a Fallibilist.” Philosophical Perspectives 2: 581–605.
Collingwood, R. G. (1939). An Autobiography. Oxford: Clarendon Press.
Cook Wilson, J. (1926). Statement and Inference, ed. by A. S.  L. Farquharson.
Oxford: Clarendon Press.
Crane, T. and French, C. (2015). “The Problem of Perception.” In E. N. Zalta
(ed.) The Stanford Encyclopedia of Philosophy. Winter 2015 Edition. Stanford,
CA: Center for the Study of Language and Information.
Crary, A. (2007). Beyond Moral Judgment. Cambridge, MA:  Harvard University
Press.
Davidson, D. (1967). “The Logical Form of Action Sentences.” In N. Rescher
(ed.), The Logic of Decision and Action, pp. 81–95. Pittsburgh:  University of
Pittsburgh Press.
(1969). “On Saying That.” Synthese 19: 130–146.
(1979). “Quotation.” Theory and Decision 11: 27–40.
DeRose, K. (2009). The Case for Contextualism. Oxford: Clarendon Press.
Dickerson, A. B. (2006). “A Reasonable Contextualism (or, Austin Reprised).” In
S. Hetherington (ed.), Aspects of Knowing: Epistemological Essays, pp. 115–131.
Oxford: Elsevier.
Dodd, D. (2010). “Confusion about Concessive Knowledge Attributions.”
Synthese 172: 381–396.
Dougherty, T. and Rysiew, P. (2009). “Fallibilism, Epistemic Possibility, and
Concessive Knowledge Attributions.” Philosophy and Phenomenological
Research 78: 123–132.
Dretske, F. I. (1970). “Epistemic Operators.” The Journal of Philosophy 67:
1007–1023.
Eddington, A. (1928). The Nature of the Physical World. New York: MacMillan.
Fiengo, R. (2007). Asking Questions: Using Meaningful Structures to Imply Ignorance.
Oxford: Oxford University Press.
(to appear). As We Commonly Say.
Fine, K. (2017). “Truthmaker Semantics.” In B. Hale, C. Wright and A. Miller
(eds.), A Companion to the Philosophy of Language, 2nd edn, pp. 556–577.
Oxford: Wiley Blackwell.
Fischer, E. (2011). Philosophical Delusion and Its Therapy: Outline of a Philosophical
Revolution. London: Routledge.
6

226 Bibliography
Forguson, L. W. (1973). “Locutionary and Illocutionary Acts.” In I. Berlin et al.,
Essays on J. L. Austin, pp. 160–185. Oxford: Oxford University Press.
Frege, G. (1879). Begriffsschrifft (selections). In M. Beaney (ed.), The Frege Reader,
pp. 47–78. Oxford: Blackwell, 1997.
(1880/81). “Boole’s Logical Calculus and the Concept-Script.” In G. Frege,
Posthumous Writings, ed. by H. Hermes, F. Kambartel, and F. Kaulbach, pp.
9–46. Chicago: University of Chicago Press, 1979.
(1891). “Function and Concept.” In M. Beaney (ed.), The Frege Reader, pp. 130–148.
Oxford: Blackwell, 1997.
(1892). “On Sense and Reference.” In P. Geach and M. Black (eds.), Translations
from the Philosophical Writings of Gottlob Frege, pp. 56–78. Oxford:
Blackwell, 1960.
(1893). Grundgesetze der Arithmetik, I (selections). In M. Beaney (ed.), The Frege
Reader, pp. 194–223. Oxford: Blackwell, 1997.
(1918). “Thought.” In M. Beaney (ed.), The Frege Reader, pp. 325–345. Oxford:
Blackwell, 1997.
(1967). The Basic Laws of Arithmetic: Exposition of the System, ed. and tr. by M.
Furth. Berkeley and Los Angeles: University of California Press.
Fricker, M. (2007). Epistemic Injustice. Oxford: Oxford University Press.
Garvey, B. (ed.) (2014). J. L. Austin on Language. Basingstoke: Palgrave Macmillan.
Gaynesford, M. de (2009a). “Illocutionary Acts, Subordination and Silencing.”
Analysis 69: 488–490.
(2009b). “Incense and Insensibility: Austin on the ‘Non-Seriousness’ of Poetry.”
Ratio 22: 464–485.
(2011a). “Speech, Action and Uptake.” In M. de Gaynesford (ed.), Agents and
Their Actions, pp. 121–137. Oxford: Blackwell.
(2011b). “How Not to Do Things with Words.” The British Journal of Aesthetics
51: 31–49.
(2013). “Speech Acts, Responsibility and Commitment in Poetry.” In P.
Robinson (ed.), The Oxford Handbook of Contemporary British and Irish
Poetry, pp. 617–637. Oxford: Oxford University Press.
(2017). The Rift in the Lute:  Attuning Poetry and Philosophy. Oxford:  Oxford
University Press.
Geach, P. (1972). “Quotation and Quantification.” In P. Geach, Logic Matters,
pp. 205–209. Oxford: Blackwell.
Ginzburg, J. and Cooper, R. (2014). “Quotation via Dialogical Interaction.”
Journal of Logic, Language, and Information 23: 287–311.
Giorgi, A. (2016). “Integrated Parentheticals in Quotations and Free Indirect
Discourse.” In A. Capone, F. Kiefer and F. Lo Piparo (eds.), Indirect Reports
and Pragmatics: Interdisciplinary Studies, pp. 471–488. Berlin: Springer.
Goffman, E. (1952). “On Cooling the Mark Out: Some Aspects of Adaptation to
Failure.” Psychiatry 15: 451–463.
(1959). The Presentation of Self in Everyday Life. New York: Anchor Books.
(1967). Interaction Ritual:  Essays in Face-to-Face Behavior. New  York: Anchor
Books.
227

Bibliography 227
(1974). Frame Analysis: An Essay on the Organization of Experience. New York:
Harper & Row.
(1983). “Felicity’s Condition.” American Journal of Sociology 89: 1–53.
Goldman, A. I. (1970). A Theory of Human Action. Princeton, NJ:  Princeton
University Press.
(1976). “Discrimination and Perceptual Knowledge.” The Journal of Philosophy
73: 771–791.
Goodall, G. (1987). Parallel Structures in Syntax. Cambridge:  Cambridge
University Press.
Graham, K. (1977). J. L.  Austin:  A  Critique of Ordinary Language Philosophy.
Hassocks: The Harvester Press.
Green, M. (2015). “Speech Acts.” In E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy. Summer 2015 Edition. Stanford, CA: Center for the Study of
Language and Information.
Grice, P. (1989). Studies in the Way of Words. Cambridge, MA:  Harvard
University Press.
Grimshaw, J. (2015). “The Light Verbs Say and Say.” In I. Toivonen, P. Csúri and
E. van der Zee (eds.), Structures in the Mind: Essays on Language, Music,
and Cognition in Honor of Ray Jackendoff, pp. 79–100. Cambridge, MA: The
MIT Press.
(Ms). “Quotes, Clausal Complements, and Parentheticals.” Manuscript, Rutgers
University.
Gustafsson, M. (2011). “Introduction.” In M. Gustafsson and R. Sørli (eds.), The
Philosophy of J. L. Austin, pp. 1–31. Oxford: Oxford University Press.
Gustafsson, M. and Sørli, R. (eds.) (2011). The Philosophy of J.  L. Austin.
Oxford: Oxford University Press.
Habermas, J. (1984). The Theory of Communicative Action, tr. T.  McCarthy.
Boston: Beacon Press.
Haddock, A. and Macpherson, F. (eds.) (2011). Disjunctivism: Perception, Action,
Knowledge. Oxford: Oxford University Press.
Hanks, P. W. (2015). Propositional Content. Oxford: Oxford University Press.
Hart, H. L. A. (1949). “The Ascription of Responsibility and Rights.” Proceedings
of the Aristotelian Society 49: 171–194.
Hatfield, G. and Epstein, W. (1979). “The Sensory Core and Medieval Foundations
of Early Modern Perceptual Theory.” Isis 70: 363–384.
Helmholtz, H. (1950). Treatise on Physiological Optics, ed. by J. Southall. New York:
Dover.
Hetherington, S. (1999). “Knowing Failably.” The Journal of Philosophy 96:
565–587.
(2001). Good Knowledge, Bad Knowledge:  On Two Dogmas of Epistemology.
Oxford: Clarendon Press.
(2002). “Fallibilism and Knowing that One Is Not Dreaming.” Canadian
Journal of Philosophy 32: 83–102.
(2011a). How to Know:  A  Practicalist Conception of Knowledge. Oxford:
Blackwell.
8

228 Bibliography
(2011b). “Knowledge and Knowing: Ability and Manifestation.” In S. Tolksdorf
(ed.), Conceptions of Knowledge, pp. 73–100. Berlin: De Gruyter.
(2013a). “Concessive Knowledge-Attributions:  Fallibilism and Gradualism.”
Synthese 190: 2835–2851.
(2013b). “Skeptical Challenges and Knowing Actions.” Philosophical Issues
23: 18–39.
(2016a). Knowledge and the Gettier Problem. Cambridge: Cambridge University
Press.
(2016b). “Understanding Fallible Warrant and Fallible Knowledge:  Three
Proposals.” Pacific Philosophical Quarterly 97: 270–282.
(forthcoming). “Skepticism and Fallibilism.” In D. Machuca and B. Reed
(eds.), Skepticism: From Antiquity to the Present. London: Bloomsbury.
Hill, G. (1998). The Triumph of Love. New York: Houghton Mifflin Company.
Hinton, J. M. (1989). “Scepticism:  Philosophical and Everyday.” Philosophy
64: 219–243.
Hornsby, J. (1994). “Illocution and its Significance.” In S. L. Tsohatzidis (ed.),
Foundations of Speech Act Theory:  Philosophical and Linguistic Perspectives,
pp. 187–207. London: Routledge.
(2006). “Speech Acts and Performatives.” In E. Lepore and B. C. Smith (eds.),
The Oxford Handbook of Philosophy of Language, pp. 893–909. Oxford:
Oxford University Press.
Ingarden, R. (1931). Das Literarische Kunstwerk. Halle: Niemeyer.
Jacobson, D. (1995). “Freedom of Speech Acts? A Response to Langton.” Philo-
sophy and Public Affairs 24: 64–79.
Jubien, M. (2001). “Propositions and the Objects of Thought.” Philosophical
Studies 104: 47–62.
Kaplan, D. (1989a). “Demonstratives:  An Essay on the Semantics, Logic,
Metaphysics, and Epistemology of Demonstratives an Other Indexicals.” In
J. Almog, J. Perry, and H. Wettstein (eds.), Themes from Kaplan, pp. 481–563.
Oxford: Oxford University Press.
(1989b). “Afterthoughts.” In J. Almog, J. Perry, and H. Wettstein (eds.), Themes
from Kaplan, pp. 565–614. Oxford: Oxford University Press.
Kaplan, M. (2000). “To What Must an Epistemology Be True?” Philosophy and
Phenomenological Research 61: 279–304.
(2006). “If You Know, You Can’t Be Wrong.” In S. Hetherington (ed.),
Epistemology Futures, pp. 180–198. Oxford: Clarendon Press.
(2011). “Tales of the Unknown: Austin and the Argument from Ignorance.” In
M. Gustafsson and R. Sørli (eds.), The Philosophy of J. L. Austin, pp. 51–77.
Oxford: Oxford University Press.
Katz, J. J. and Postal, P. M. (1964). An Integrated Theory of Linguistic Descriptions.
Cambridge, MA: The MIT Press.
Kratzer, A. (2014). “Situations in Natural Language Semantics.” In E. N. Zalta
(ed.), The Stanford Encyclopedia of Philosophy. Spring 2014 Edition. Stanford,
CA: Center for the Study of Language and Information.
Kukla, R. and Lance, M. (2009). Yo! And Lo! The Pragmatic Topography of the Space
of Reasons. Cambridge, MA: Harvard University Press.
229

Bibliography 229
Langton, R. (2009). Sexual Solipsism:  Philosophical Essays on Pornography and
Objectification. Oxford: Oxford University Press.
Laugier, S. (2005). “Austin after Cavell.” In R. B. Goodman (ed.), Contending with
Stanley Cavell, pp. 82–99. Oxford: Oxford University Press.
(2009a). “Transcendentalism and the Ordinary.” European Journal of Pragma-
tism and American Philosophy 1: 1–17.
(2009b). Wittgenstein: Les sens de l’usage. Paris: Vrin.
(2013). Why We Need Ordinary Language Philosophy. Chicago:  University of
Chicago Press.
Laugier, S. and Al-Saleh, C. (eds.) (2011). J. L. Austin et la philosophie du langage
ordinaire. Hildesheim: Olms.
Lawlor, K. (2013). Assurance:  An Austinian View of Knowledge and Knowledge
Claims. Oxford: Oxford University Press.
(2015). “Précis of Assurance.” Philosophy and Phenomenological Research 90:
194–204.
Leite, A. (2011). “Austin, Dreams, and Scepticism.” In M. Gustafsson and R.
Sørli (eds.), The Philosophy of J.  L. Austin, pp. 78–113. Oxford:  Oxford
University Press.
Lemmon, E. J. (1962). “On Sentences Verifiable by their Use.” Analysis 22: 86–89.
Lewis, D. (1979). “Scorekeeping in a Language Game.” Journal of Philosophical
Logic 8: 339–359.
(1996). “Elusive Knowledge.” Australasian Journal of Philosophy 74: 549–567.
Locatelli, R. (2014). “Sense and Sensibilia and the Significance of Linguistic
Phenomenology.” In B. Garvey (ed.), J. L. Austin on Language, pp. 141–157.
Basingstoke: Palgrave Macmillan.
MacKinnon, C. (1987). Feminism Unmodified:  Discourses on Life and Law.
Cambridge, MA: Harvard University Press.
Macpherson, F. and Platchias, D. (eds.) (2013). Hallucination:  Philosophy and
Psychology. Cambridge, MA: The MIT Press.
Maddy, P. (2017). What Philosophers Do? Skepticism and the Practice of Philosophy.
Oxford: Oxford University Press.
Maier, E. (2014a). “Pure Quotation.” Philosophy Compass 9: 615–630.
(2014b). ‘Mixed Quotation: The Grammar of Apparently Transparent Opacity.”
Semantics & Pragmatics 7: 1–67.
Maitra, I. (2009). “Silencing Speech.” Canadian Journal of Philosophy 39:
309–338.
(2012). “Subordinating Speech.” In I. Maitra and M. K. McGowan (eds.),
Speech and Harm: Controversies over Free Speech, pp. 95–118. Oxford: Oxford
University Press.
Marion, M. (2000a). “Oxford Realism:  Knowledge and Perception I.” British
Journal for the History of Philosophy 8: 299–338.
(2000b). “Oxford Realism: Knowledge and Perception II.” British Journal for
the History of Philosophy 8: 485–519.
Martin, M. G. F. (2007). “Austin: Sense and Sensibilia Revisited.” London
Philosophy Papers, School of Advanced Study, University of London. http://
sas-space.sas.ac.uk/id/eprint/631.
0

230 Bibliography
Matushansky, O. (2008). “On the Linguistic Complexity of Proper Names.”
Linguistics and Philosophy 21: 573–627.
McDowell, J. (1982). “Criteria, Defeasibility, and Knowledge.” Proceedings of the
British Academy 68: 455–479.
(1994a). “Knowledge by Hearsay.” In B. K. Malital and A. Chakrabarti (eds.),
Knowing from Words, pp. 195–224. Dordrecht: Kluwer.
(1994b). “The Content of Perceptual Experience.” The Philosophical Quarterly
44: 190–205.
(1998). Meaning, Knowledge, and Reality. Cambridge, MA: Harvard University
Press.
Mikkola, M (2008). “Contexts and Pornography.” Analysis 68: 316–320.
Millikan, R. G. (1998). “Proper Function and Convention in Speech Acts.” In L. E.
Hahn (ed.), The Philosophy of P. F. Strawson, pp. 25–43. Chicago: Open Court.
Moltmann, F. (1992). Coordination and Comparatives. PhD Thesis, MIT,
Cambridge, MA.
(2003). “Nominalizing Quantifiers.” Journal of Philosophical Logic 35: 445–481.
(2013). Abstract Objects and the Semantics of Natural Language. Oxford: Oxford
University Press.
(2014). “Propositions, Attitudinal Objects, and the Distinction between Actions
and Products.” Canadian Journal of Philosophy 43: 679–701.
(2015a). “Truth Predicates in Natural Language.” In T. Achourioti, H. Galinon,
J. Martínez Fernández and K. Fujimoto (eds.) Unifying the Philosophy of
Truth, pp. 57–83. Berlin: Springer.
(2015b). “A Predicativist Semantics of Modals Based on Modal Objects.” In
T. Brochagen, F. Roloefson, and N. Theiler (eds.) Proceedings of the 20th
Amsterdam Colloquium, pp. 296–302. Institute for Logic, Language, and
Computation, University of Amsterdam.
(2017a). “Partial Content and Expressions of Part and Whole.” Philosophical
Studies 174: 797–808.
(2017b). “Cognitive Products and the Semantics of Attitude Verbs and Deontic
Modals.” In F. Moltmann and M. Textor (eds.), Act-Based Conceptions of
Propositional Content: Contemporary and Historical Perspectives, pp. 254–289.
Oxford: Oxford University Press.
Moore, G. E. (1905–1906). “The Nature and Reality of Objects of Perception.”
Proceedings of the Aristotelian Society 6: 68–127.
Munro, P. (1982). “On the Transitivity of the ‘Say’ Verbs.” Syntax and Semantics
15: 301–318.
Neale, S. (2001). Facing Facts. Oxford: Oxford University Press.
Ogden, C. K. and Richards, I. A. (1923). The Meaning of Meaning. London:
Routledge.
Owens, D. (2012). Shaping the Normative Landscape. Oxford:  Oxford University
Press.
Pagin, P. and Westerståhl, D. (2010). “Pure Quotation and General Compo-
sitionality.” Linguistics and Philosophy 33: 381–415.
Potts, C. (2007). “The Dimensions of Quotation.” In C. Barker and P. Jacobson
(eds.), Direct Compositionality, pp. 405–431. Oxford: Oxford University Press.
Price, H. H. (1932). Perception. London: Methuen and Co.
231

Bibliography 231
Prichard, H. A. (1919). “Professor John Cook Wilson.” Mind 28: 297–318.
(1950). Knowledge and Perception: Essays and Lectures. Oxford: Clarendon Press.
(1968). Moral Obligation and Duty and Interest:  Essays and Lectures. Oxford:
Oxford University Press.
(2002). Moral Writings, ed. by J. MacAdam. Oxford: Oxford University Press.
Pritchard, D. (2012). Epistemological Disjunctivism. Oxford:  Oxford University
Press.
Putnam, H. (1994). “Sense, Nonsense, and the Senses: An Inquiry into the Powers
of the Human Mind.” The Journal of Philosophy 91: 445–517.
(1999). The Threefold Cord:  Mind, Body, and World. New  York:  Columbia
University Press.
Recanati, F. (1978). La transparence et l’énonciation. Paris: Seuil.
(1980). Les énoncés performatifs. Paris: Minuit.
(1991). “Postface: Du positivisme logique à la philosophie du langage ordi-
naire.” In J. L. Austin, Quand dire, c’est faire, tr. G. Lane, pp. 185–203. Paris:
Seuil.
(2000). Oratio Obliqua, Oratio Recta:  An Essay on Metarepresentation.
Cambridge, MA: The MIT Press.
Reichenbach, H. (1947). Elements of Symbolic Logic. New York: Free Press.
Reimer, M. (2006). “The Metaphor of Correspondence.” ProtoSociology 23:
93–110.
Reimer, M. and Camp, E. (2006). “Metaphor.” In E. Lepore and B. C. Smith
(eds.), The Oxford Handbook of Philosophy of Language, pp. 845–863.
Oxford: Oxford University Press.
Reinach, A. (1913). “Die Apriorischen Grundlagen Des Bürgerlichen Rechtes.”
Jahrbuch für Philosophie und phänomenologische Forschung 1: 685–847.
(1983). “The A  Priori Foundations of Civil Law,” tr. J.  F.  Crosby, Aletheia 3:
1–142. Originally published as Reinach 1913.
Rickless, S. (2013). Berkeley’s Argument for Idealism. Oxford:  Oxford University
Press.
Rizzi, L. (1997). “The Fine Structure of the Left Periphery.” In L. Haegeman
(ed.), Elements of Grammar:  Handbook in Generative Syntax, pp. 281–337.
Dordrecht: Kluwer.
Robinson, H. (1994). Perception. London: Routledge.
Ross, J. (1970). “On Declarative Sentences.” In R. Jacobs and P. Rosenbaum (eds.),
Readings in English Transformational Grammar, pp. 222–272. Waltham, MA:
Ginn and Co.
Ryle, G. (1945). “Knowing How and Knowing That.” Proceedings of the Aristotelian
Society, 46, 1–16.
(1949). The Concept of Mind. London: Hutchinson.
Saka, P. (1998). “Quotation and the Use-Mention Distinction.” Mind 107: 113–135.
Saul, J. (2006). “Pornography, Speech Acts, and Context.” Proceedings of the
Aristotelian Society 106: 229–248.
Schwartz, R. (1994). Vision:  Variations on Some Berkeleian Themes. Oxford:
Blackwell.
(2004a). “To Austin or Not to Austin:  That’s the Disjunction.” Philosophical
Studies 120: 255–263.
2

232 Bibliography
(ed.) (2004b). Perception. Oxford: Blackwell.
(2006). Visual Versions. Cambridge, MA: The MIT Press.
(2012). “The Illusion of Visual Illusions.” In C. Calabi (ed.), Perceptual Illusions,
pp. 25–43. London: Palgrave Macmillan.
(2016). “Perceptual Veridicality.” Philosophical Topics 44: 381–403.
(to appear). “Berkeleian Instrumentalism: From Substance to Space.” In L.
Berchielli (ed.), Empiricist Theories of Space. Berlin: Springer.
Searle, J. R. (1968). “Austin on Locutionary and Illocutionary Acts.” The
Philosophical Review 77: 405–424.
(1969). Speech Acts:  An Essay in the Philosophy of Language. Cambridge:
Cambridge University Press.
(1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge
University Press.
(1989). “How Performatives Work.” Linguistics and Philosophy 12: 535–558.
(1998a). “ Truth: A Reconsideration of Strawson’s Views.” In L. E. Hahn
(ed.), The Philosophy of P. F. Strawson, pp. 385–401. Chicago: Open Court.
(1998b). Mind, Language and Society:  Philosophy in the Real World.
New York: Basic Books.
(2014). “Recollections of J.  L. Austin.” In B. Garvey (ed.), J. L.  Austin on
Language, pp. 1–10. Basingstoke: Palgrave Macmillan.
(2015). Seeing Things as They Are:  A  Theory of Perception. Oxford:  Oxford
University Press.
Searle, J. R. and Vanderveken, D. (1985). Foundations of Illocutionary Logic.
Cambridge: Cambridge University Press.
Shakespeare, W. (2002). The Oxford Shakespeare: Complete Sonnets and Poems, ed.
by C. Burrow. Oxford: Oxford University Press.
Snowdon, P. (2003). “Knowing How and Knowing That:  A  Distinction
Reconsidered.” Proceedings of the Aristotelian Society 104: 1–29.
(2014). “Austin on the Philosophy of Perception.” In B. Garvey (ed.), J. L. Austin
on Language, pp. 161–175. Basingstoke: Palgrave Macmillan.
Soames, S. (2003). Philosophical Analysis in the Twentieth Century, Volume 2: The
Age of Meaning. Princeton, NJ: Princeton University Press.
(2010). What Is Meaning? Princeton, NJ: Princeton University Press.
Soteriou, M. (2014). “The Disjunctive Theory of Perception.” In E. N. Zalta (ed.)
The Stanford Encyclopedia of Philosophy. Summer 2014 Edition. Stanford, CA:
Center for the Study of Language and Information.
(2016). Disjunctivism. London: Routledge.
Stanley, J. (2005). “Fallibilism and Concessive Knowledge Attributions.” Analysis
65: 126–131.
Strawson, P. F. (1950). “Truth.” Proceedings of the Aristotelian Society, Supplementary
Volume 24: 129–156.
(1964). “Intention and Convention in Speech Acts.” The Philosophical Review
73: 439–460.
(1998). “Reply to Ruth Garrett Millikan.” In L. E. Hahn (ed.), The Philosophy of
P. F. Strawson, pp. 44–48. Chicago: Open Court.
Stroud, B. (1984). The Significance of Philosophical Scepticism. Oxford: Clarendon
Press.
233

Bibliography 233
Thau, M. (2004). “What Is Disjunctivism?” Philosophical Studies 120: 193–253.
Travis, C. (1981). The True and the False: The Domain of the Pragmatic. Amsterdam:
Benjamins.
(2001). Unshadowed Thought:  Representation in Thought and Language.
Cambridge, MA: Harvard University Press.
(2004). “The Silence of the Senses.” Mind 113: 57–94.
(2005). “A Sense of Occasion.” The Philosophical Quarterly 55: 286–314.
(2008). Occasion-Sensitivity: Selected Essays. Oxford: Oxford University Press.
(2013). “As a Matter of Fact.” Proceedings of the Aristotelian Society, Virtual Issue
1: 42–51.
(2015). Review of Searle 2015. Notre Dame Philosophical Reviews, July 6, 2015.
Travis, C. and Kalderon, M. E. (2013). “Oxford Realism.” In M. Beaney (ed.),
The Oxford Handbook of the History of Analytic Philosophy, pp. 489–517.
Oxford: Oxford University Press.
Twardowski, K. (1912). “Actions and Products: Some Remarks on the Borderline of
Psychology, Grammar, and Logic.” In K. Twardowski, On Actions, Products,
and Other Topics in Philosophy, ed. by J. Brandl and J. Woleński, pp. 103–132.
Amsterdam: Rodopi, 1999.
Ulrich, W. (1976). “An Alleged Ambiguity in the Nominalizations of Illocutionary
Verbs.” Philosophica 18: 113–127.
Urmson, J. O., Quine, W. V. O., and Hampshire, S. (1969). “A Symposium on
Austin’s Method.” In K. T. Fann (ed.), Symposium on J. L. Austin, pp. 76–97.
London: Routledge.
Vries, M.  de (2008). “The Representation of Language within Language:  A
Syntactico-Pragmatic Typology of Direct Speech.” Studia Linguistica 62:
39–77.
Waldron, J. (2012). The Harm in Hate Speech. Cambridge, MA:  Harvard
University Press.
Warnock, G. J. (1953). Berkeley. Harmondsworth: Penguin Books.
(1989). J. L. Austin. London: Routledge.
Washington, C. (1992). “The Identity Theory of Quotation.” The Journal of
Philosophy 89: 582–605.
Williams, B. (1978). Descartes:  The Project of Pure Inquiry. London:  Penguin
Books.
Williams, M. (2001). Problems of Knowledge: A Critical Introduction to Epistemo-
logy. Oxford: Oxford University Press.
Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.
Wittgenstein, L. (1922). Tractatus Logico-Philosophicus. London: Routledge.
(1953). Philosophical Investigations. Oxford: Blackwell.
(1969). On Certainty. Oxford: Blackwell.
Wright, C. (1985). “Facts and Certainty.” Proceedings of the British Academy 71:
429–472.
4
235

Index

Alston, W. P., 88–89, 89n21, 89n22, 90n24, 91 DeRose, K., 222

Annis, D., 222 Descartes, R., 25, 150n8, 183, 183n27
Anscombe, G. E. M., 63n5 descriptive fallacy, 100, 118, 121–24, 125
Atherton, M., 144n2 Dickerson, A. B., 206n1
Ayer, A. J., 143, 146, 166–71, 174, 180–81, 184, direct reference, 104, 108
218n9 Dodd, D., 217n7
Dougherty, T., 217n7
Bach, K., 86, 89n20, 91, 103 Dretske, F. I., 219, 219n11, 220, 222
Barwise, J., 2n2, 121
Bauer, N., 2, 80, 82–83, 85, 86, 88 Eddington, A., 157
Baz, A., 2 Epstein, W., 149n7
Beardsley, M. C., 17 Etchemendy, J., 2n2
Beauvoir, S. de, 81n5 excuses, 9, 120, 130–35, 140
Benveniste, E., 130
Berkeley, G., 10, 144–59, 164 Fiengo, R., 2, 4–6, 60, 62n2, 77, 77n16
Bird, G., 88n17 Fine, K., 41
Black, M., 17 Fischer, E., 2
Bradley, D., 220n12 Forguson, L. W., 88n17
Brandom, R., 91 free speech, 87, 92, 93
Butler, J., 86n13 Frege, G., 60–64, 73n12, 77n16, 93, 93n30,
122, 126
Camp, E., 17 French, C., 166n7
Cappelen, H., 52n19 Fricker, M., 80, 81, 83–84, 85, 86
Carnap, R., 131, 170 Furth, M., 62n3
Cattell, R., 39n1
Cavell, S., 121, 126, 128, 132, 136, 137, 138, Geach, P., 52
140, 141 Gerrig, R., 52
Chaucer, G., 94 Ginzburg, J., 52n20
Chisholm, R. M., 208–09 Giorgi, A., 54, 54n22, 55
Chomsky, N., 66n9 Goffman, E., 134n4, 135, 136, 138–42
Clark, H. H., 52, 86 Goldman, A. I., 49, 219n11
Clarke, T., 201 Goodall, G., 54
Cohen, S., 222 Graham, K., 88n17, 206
Collingwood, R. G., 196 Green, M., 43n5
Cook Wilson, J., 187–94, 196 Grice, P., 31, 89n20, 91
Cooper, R., 52n20 Grimshaw, J., 43, 47, 51, 55
Crane, T., 166n7 Gustafsson, M., 80n3
Crary, A., 2
Habermas, J., 86n13
Davidson, D., 38, 52, 56 Hanks, P. W., 35, 52
de Gaynesford, M., 6–7, 79, 79n2, 93n29, 95 Harnish, R., 86, 89n20, 91

235
6

236 Index
Hart, H. L. A., 119, 132, 134 knowing-that as a form of
hate speech, 80n2 knowing-how, 210–13
Hatfield, G., 149n7 knowledge claims and performativity, 207–09
Helmholtz, H., 150n8 knowledge claims, truth conditions of, 178–79
Hetherington, S., 13–14, 206, 210, 211n4, 213n5, knowledge practicalism, 210–16
216n6, 217n7, 217n8, 221n14 knowledge-related illocutionary acts, 209–10
Hill, G., 94 luminosity, 192
Hinton, J. M., 198 Oxford Realism, 12, 187–96, 201, 202,
Hornsby, J., 79n1, 81n5, 81n6, 86n12, 88n16, 97 203, 205
Hume, D., 124, 126, 127, 130 relevant alternatives account of knowledge,
176–77, 219–22
illocutionary acts, 3–9, 34–51, 60–78, 79–95, scepticism, 183–84, 197–203
97–118, 119–42, 207–16 Kratzer, A., 2n2
and epistemology, 176–77, 206–16 Kukla, R., 88n18, 91
exercitive acts and sentence types, 71–77
exercitive acts, four varieties of, 68–71 Lance, M., 88n18, 91
expositive acts, 209–10, 214–16 Langton, R., 2, 6, 79, 80n4, 86n13, 87, 87n15,
illocutionary act ascriptions, semantics 88n16
of, 35–42 Laugier, S., 2, 9, 119, 121, 134n4, 141
illocutionary vs. locutionary act Lawlor, K., 2, 10–11, 165, 175n16, 183n24, 184n28,
ascriptions, 42–50 206n2, 218, 219, 221n13
vs. illocutionary products, 36–42 Leite, A., 183, 198–201, 204
misfires and abuses, 89, 111–12, 129–31 Lemmon, E. J., 102
and normativity, 91–92, 119–42 Lepore, E., 52n19
promising, 37–38, 64n6, 66, 90n25, 99, 101n2, Lewis, D., 102, 221n13, 222
111, 114, 119, 123, 125–28, 129–31, 207–08 Locatelli, R., 165n4
uptake-dependent vs. uptake-free illocutions, Locke, J., 150n8
6–7, 87–93 locutionary acts and products, 3–4, 34–35, 41–59
verdictive acts, 209–10, 214–16 locutionary act ascriptions, semantics
illocutionary force and grammar, 4–6, 60, of, 41–51
64–78, 101 locutionary act products and the semantics of
illocutionary force indicators, 64–67 quotation, 51–59
imperative sentences, 67–74 locutionary vs. illocutionary act
indicative sentences, 74–76 products, 45–46
interrogative sentences, 77 rhetic act products as objects of
implicature, 16, 30–31 reference, 46–50
Ingarden, R., 42n3 rhetic act products vs. propositions, 45–46
Longworth, G., 12–13, 186
Jacobson, D., 88n17
Jubien, M., 35, 52 MacKinnon, C., 79n1
judgment stroke, 60–64 Maddy, P., 2
Maier, E., 51n17
Kalderon, M., 180n22, 194 Maitra, I., 80, 81–82, 83, 84–85, 89n20
Kaplan, D., 104 Marion, M., 194
Kaplan, M., 183n27, 203–04 Martin, M. G. F., 159, 159n14, 161, 161n15,
Katz, J. J., 65 174n14
knowledge, theory of, 11–14, 176–84, 186–222 Matushansky, O., 57n24
concessive knowledge attributions, 217 McDowell, J., 13, 181n23, 201–03
contextualism, 221n13, 222 metaphor, 2, 17, 19–21, 22–23
disjunctivism, 181–83 Mikkola, M., 80n4
dogmatism, 195–97 Millikan, R. G., 89n19
doxastic responsibility, 13, 187–96, 201–03 Moltmann, F., 3–4, 34, 35, 37, 38, 39, 40, 41,
epistemic closure, 183–84, 203–05 46n11, 47, 52, 54, 57
fallibilism, 177, 182, 216–22 Moore, G. E., 28, 187
knowing actions, 210, 213–16 Munro, P., 43
237

Index 237
Neale, S., 15, 17, 18, 21–22 Reinach, A., 119, 121–28, 130–31
negation, 64–65, 75–76 Richards, I. A., 127
Rickless, S., 150n9
Ogden, C. K., 127 Rizzi, L., 65
Owens, D., 90n25 Robinson, H., 165n2
Ross, J., 65
Pagin, P., 51 Russell, B., 28, 46n11, 127
perception, metaphysics and epistemology of, Ryle, G., 119, 158, 214, 216n6
9–11, 143–85 Rysiew, P., 217n7
de facto conclusive grounds and perceptual
knowledge, 180–82 Saka, P., 52, 56
de facto incorrigibility and perceptual Saul, J., 80n4
knowledge, 170, 177–80 Schwartz, R., 2, 9–10, 143, 144n2, 147n6,
disjunctive theories of perception, 149n7, 149n7, 153n12, 159n13, 159n14,
160–63, 181–83 163n17
immediate vs. indirect perception, 147–50 Searle, J. R., 4, 5, 15, 17, 18, 19–21, 37, 43n5,
object of perception, 159–62, 173–75 44, 45n10, 45n7, 45n9, 46, 48, 49, 64–65,
ontological pluralism and the object of 77, 86, 86n13, 103, 111–12, 119, 121, 130,
perception, 154–58 144n3, 171
perceptual illusions, 144–47 Shakespeare, W., 94
perceptual illusions vs. hallucinations, 162 silencing, 6–7, 79–95
scepticism and perceptual knowledge, 150, illocutionary vs. communicative conceptions
151, 154, 180–82 of, 80–93
sensation and perception, 148–50 situation semantics, 2n2, 172–73
sense data, 10, 143, 159, 165, 166–71, Snowdon, P., 165n1, 211n4
174n14, 181 Soames, S., 35, 52, 102
veridical perception, 154–59 Soteriou, M., 163, 191
performative utterances, 7–8, 68, 70, 96–118, Stanley, J., 217n7
119, 120–25, 128–31, 132–33, 207–09, 222 Strawson, P. F., 15, 17, 18–19, 28, 86, 89n19,
performative verbs, 68, 99, 107 89n20, 90n24
perlocutionary acts and effects, 64n6, 74–76, 77, Stroud, B., 183–84, 201
80, 81–82, 81n5, 84–85, 128
Perry, P., 2n2 tense, 5, 73–74, 98–99
Postal, P. M., 65 Thau, M., 159n14, 162n16, 165n2
Potts, C., 58n25 Travis, C., 2, 25–28, 126, 138, 144n3, 148, 159n14,
Price, H. H., 166, 168, 170 179–80, 190, 194
Prichard, H. A., 101n2, 123n1, 187–94, 196 truth and falsehood, 2–3, 2n2, 7–8, 15–33, 41,
Pritchard, D., 181n23 61, 71–74, 96–118, 122–24, 129–31, 154–56,
propositional attitudes, 127 172–73, 176–79
and attitudinal objects, 36–42 vs. clarity and unclarity, 71–72
Putnam, H., 137, 159n14, 165n3 context-dependence of, 11, 25–26, 154–56,
172–73, 176–79
Quine, W. V. O., 102–03, 155 vs. felicity and infelicity, 129–31
quotation, semantics of, 3, 4, 34, 35–36, 51–59 as involving descriptive and demonstrative
direct quotation, 36, 43–45, 53–55 conventions, 71–74
mixed quotation, 58–59 in relation to imperatives, 71–74
pure quotation, 36, 52, 52n18, 53–58 in relation to performatives, 96–118
as spectral concepts, 23–25
‘real’, reality, and realism theories of, 15–33
in relation to action, 135–42 truthmaker semantics, 41
in relation to perception, 150–59 Twardowski, K., 4, 35, 38, 42
Recanati, F., 52n18, 119, 120, 127, 128, 130
Reichenbach, H., 52 Ulrich, W., 37
Reid, T., 130
Reimer, M., 2–3, 15, 17 Vanderveken, D., 111
8

238 Index
Waldron, J., 80n2 Williams, M., 222
Warnock, G. J., 88n17, 144n2, 171, 219n10 Williamson, T., 189, 190, 192, 213
Washington, C., 52, 56 Wittgenstein, L., 63n5, 122, 126, 137, 138,
Westerståhl, D., 51 142, 222
Williams, B., 191 Wright, C., 222