Insensitive Semantics~ A Defense of Semantic Minimalism and Speech Act Pluralism

P R A I S E F O R I N S E N S I T I V E S E M A N T I C S

“This book is an ingenious defense of two positions not widely thought to
be compatible: truth-conditional semantics, on the one hand, and seman-
tic minimalism (the view that extra-linguistic context has only a minimal
effect on semantic content) on the other. Cappelen and Lepore’s highly con-
troversial views are already, and will continue to be, at the center of inquiry
into the nature of linguistic communication.”

ASON

TANLEY

, U

NIVERSITY OF

ICHIGAN

“To what extent should a theory of human semantic competence incorpo-
rate an account of context-dependence? Cappelen and Lepore offer an inter-
esting answer; namely, only to the extent that, by virtue of the meanings of
certain linguistic forms, it absolutely has to. Particular examples will be
debated, and we may hope clariﬁed, as the scope and limits of semantic
investigation are drawn more tightly. In the meantime, Cappelen and
Lepore have performed a singular service in bringing together the threads
of the contextualist debate, and in formulating a minimalist alternative to
some current trends.”

AMES

IGGINBOTHAM

, U

NIVERSITY OF

OUTHERN

ALIFORNIA

“This is a pleasingly spare yet instructively sophisticated account of how
Davidsonians can accommodate the massive context sensitivity of language
use. And the authors offer powerful arguments that suggest limits, rarely
respected, on attempts to diagnose this sensitivity in terms of indexicality.
Good stuff.”

AUL

IETROSKI

, U

NIVERSITY OF

ARYLAND

Insensitive Semantics

A Defense of Semantic

Minimalism and Speech

Act Pluralism

Herman Cappelen and Ernie Lepore

BLACKWELL PUBLISHING

350 Main Street, Malden, MA 02148-5020, USA
108 Cowley Road, Oxford OX4 1JF, UK
550 Swanston Street, Carlton, Victoria 3053, Australia

The right of Herman Cappelen and Ernie Lepore to be identiﬁed as the Authors
of this Work has been asserted in accordance with the UK Copyright, Designs,
and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a
retrieval system, or transmitted, in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, except as permitted by the UK
Copyright, Designs, and Patents Act 1988, without the prior permission of the
publisher.

First published 2005 by Blackwell Publishing Ltd

Library of Congress Cataloging-in-Publication Data

Cappelen, Herman.

Insensitive semantics : a defense of semantic minimalism and speech act

pluralism / Herman Cappelen and Ernie Lepore.

p. cm.

Includes bibliographical references and index.
ISBN 1-4051-2674-4 (hardback : alk. paper) – ISBN 1-4051-2675-2 (pbk. : alk.

paper)

1. Semantics.

2. Semantics (Philosophy)

3. Context (Linguistics)

4. Minimalist theory (Linguistics)

5. Speech acts (Linguistics)

I. LePore,

Ernest, 1950–

II. Title.

P325.C285 2005
401

¢.43–dc22

2004015918

A catalogue record for this title is available from the British Library.

Set in 10/12pt Legacy
by SNP Best-set Typesetter Ltd, Hong Kong
Printed and bound in the United Kingdom
by MPG Books Ltd, Bodmin, Cornwall

The publisher’s policy is to use permanent paper from mills that operate a
sustainable forestry policy, and which has been manufactured from pulp
processed using acid-free and elementary chlorine-free practices. Furthermore, the
publisher ensures that the text paper and cover board used have met acceptable
environmental accreditation standards.

For further information on
Blackwell Publishing, visit our website:
www.blackwellpublishing.com

Herman Cappelen dedicates this volume to Kat

Ernie Lepore dedicates it to two old friends – Brian McLaughlin,

who taught him life, and Barry Loewer, who retaught

him philosophy

Contents

vii

Preface

Acknowledgments

1 Overview

Part I: From Moderate to Radical Contextualism

2 Exegesis: The Methodology of Contextualism

3 The Instability of Context Shifting Arguments

4 Diagnosis: Why Context Shifting Arguments are Misused

5 The Instability of Incompleteness Arguments

6 Digressions: Binding and Hidden Indexicals

Part II: Refutation of Radical Contextualism

7 Objections to Radical Contextualism (I): Fails Context

Sensitivity Tests

8 Objections to Radical Contextualism (II): Makes

Communication Impossible

123

9 Objections to Radical Contextualism (III): Internal

Inconsistency

128

Part III: Semantic Minimalism and Speech
Act Pluralism

10 Semantic Minimalism

143

11 Semantics and Metaphysics

155

12 Semantics and Psychology

176

13 Speech Act Pluralism

190

References

209

Index

215

Contents

viii

Preface

From the end of the nineteenth century right up until today, Philosophy of
Language has been plagued by an extensive, and notoriously confusing, lit-
erature on how to draw the distinction between semantic content and non-
semantic content, or, in a terminology we prefer not to use, on how to draw
the distinction between semantic and pragmatic content. This debate, at its
deepest level, is about how to accommodate context sensitivity within a
theory of human communication. It is concerned with the way in which con-
texts of utterance inﬂuence communicative interactions (and, as a corollary,
what a context of utterance is and what it is to be in one). It is impossible to
take a stand on any issue in the philosophy of language without being clear
on these issues because what you consider as evidence for a semantic theory
depends on how these distinctions are ultimately drawn. And it doesn’t stop
there. Epistemologists, metaphysicians, philosophers of mind, ethicists, aes-
theticians, philosophical logicians, psychologists, linguists, anthropolo-
gists, literary critics, cognitive scientists, and perhaps, everyone else, live by
claims about whether this or that expression is context sensitive or not.
More often than not, theorists conclude that a lot more context sensitivity
abounds than one might have thought. All such claims presuppose a general
theory of the role of context in human communication.

Our ambitious goal in this book is to defend a simple and naive view

about context sensitivity, the kind of view you might come up with after
just a few moments’ reﬂection. Our view goes something like this: There are
just a few easily identiﬁable context sensitive expressions in natural lan-
guage. In English, they are familiar words like ‘I,’ ‘you,’ ‘that,’ ‘now,’ etc. In
essence, our view is that there are no deep secrets or hidden surprises behind
that ‘etc.’

To this end, we defend a combination of two views, both of which we

have given fancy names: Semantic Minimalism and Speech Act Pluralism. If

these views are right (and they are), then numerous philosophers and lin-
guists are guilty of some very profound mistakes. Not only that, but if we
are right, then the chief theses of a signiﬁcant number of published articles
and books are based on an internally inconsistent view. In this sense, our
view is deeply critical of the last century of the literature on these issues.
Since we are making these rather bold claims about colleagues’ views, we
try to be very careful in our presentations of the views we criticize. We have
included extensive exegetical sections. As a result, our readers will, we hope,
end up not only with a presentation of the correct view, but also with a clear
understanding of the structure of the last one hundred years of debate
about these issues.

The central opponents in this book are philosophers and linguists who

inﬂate the role of context in semantics. We call such philosophers Contex-
tualists. The common thread that runs throughout our criticism of contex-
tualism is that it fails to account for how we communicate across contexts.
People with different background beliefs, goals, audiences, perceptual
inputs, etc. can understand each other. They can agree or disagree. They can
say, assert, claim, state, investigate, or make fun of the very same claim. No
theory of communication is adequate unless it explains how this is possi-
ble. Contextualists cannot provide such an explanation. The solution pro-
posed in this book is a context insensitive semantics (i.e. the view we call
Semantic Minimalism), combined with Speech Act Pluralism.

Preface

Acknowledgments

Eight years ago we wrote ‘On an Alleged Connection between Semantic
Theory and Indirect Speech.’ Since then we have written a series of related
papers on speech act content, semantic content, and various versions of
contextualism in the philosophy of language. This book supersedes several
of those papers which are yet to reach print – even though they were written
well in advance of this book. Apparently some authors can’t make dead-
lines no matter how long editors are prepared to extend them. However,
this book reﬁnes and uniﬁes these published papers and we hope soon to
be published earlier papers.

Before our collaboration, Lepore had written a series of papers on David-

son with Barry Loewer in the 1980s, a forthcoming two-volume book with
Kirk Ludwig on Davidson’s philosophy of language, and a book and numer-
ous articles with Jerry Fodor on the nature of content and compositional-
ity. These all played crucial roles in forging his commitment to Semantic
Minimalism. Cappelen’s initial encounter with Radical Contextualism
came in an exceptionally thought provoking seminar given by François
Recanati at the University of California at Berkeley in 1994 (these lectures
provided the basis for Recanati’s Literal Meaning). Discussions with Recanati
and other seminar participants (including Stephen Neale and John Searle)
started Cappelen down the road towards Semantic Minimalism. Our
largest joint inﬂuence has been Jason Stanley. He together with Zoltán
Szabó wrote a paper defending contextualism for quantiﬁer domain speci-
ﬁcation. Working through their paper combined with an almost continu-
ous all-out-all-systems-go conversation with Jason jumpstarted this book.
We owe each and every one (read unrestrictedly) of these philosophers a
debt for help and inspiration.

There have been many others. Drafts of these chapters have been pre-

sented on ﬁve continents and to literally dozens of institutions and con-

ferences. We have each taught the volume several times at the undergradu-
ate and graduate levels. We thank the participants of all of these events; in
particular, in no order of signiﬁcance (and with apologies in advance to
anyone we may have forgotten), we thank: Kent Bach, Mark Baker, Anne
Bezuidenhout, Dan Blair, Emma Borg, Manuel García-Carpintiero, Robyn
Carston, David Chalmers, Jennifer Church, Lenny Clapp, Eros Corraza,
Sam Cumming, Martin Davies, Ray Elugardo, Chris Gaulker, Michael
Glanzberg, Peter Godfrey-Smith, Delia Graff, Gill Harman, John
Hawthorne, Jessica Hughes, Kent Johnson, David Kaplan, Chris Kennedy,
Jeff King, Karson Kovakovich, Richard Larson, Sarah Jane Leslie, Peter
Ludlow, Kirk Ludwig, Brian McLaughlin, Europa Malynicz, Robert May,
Michael Nelson, Terry Parsons, Jeff Pellitier, Paul Pietroski, Stefano Predelli,
Marga Reimer, Mark Richard, Stephen Schiffer, Roger Schwartzchild,
Adam Sennett, Elka Shortsleeve, Ted Sider, Rob Stainton, Dan Stoljar,
Matthew Stone, Zoltan Szabó, Ken Taylor, Charles Travis, Brian Weather-
son, Sam Wheeler, Tim Williamson, Douglas Winblad, and of course the
late Donald Davidson.

We would also like to thank Laurien Berkeley and Sarah Dancy for their

help in editing the book, Elka Shortsleeve for indexing it, as well as Jeff
Dean for helping us to get the book to press.

Acknowledgments

xii

C H A P T E R 1

Overview

Try this on some pure, uncontaminated, students: List a few incontrovert-
ibly context sensitive expressions like ‘I,’ ‘you,’ ‘now,’ and ‘that.’ Then ask
them to pick other expressions just like these. They are very good at it. They
consistently choose expressions like ‘yesterday,’ ‘those,’ ‘we,’ and they never
choose expressions like ‘penguin,’ ‘red,’ ‘know,’ or ‘dance.’ And if you ask
them directly whether they think that ‘penguin’ is like the ﬁrst person per-
sonal pronoun ‘I,’ they think you must be joking; when they understand
that you’re serious, they invariably answer ‘no.’ Of course, they might be
wrong. It might be that the more reﬁned intuitions of seasoned linguists
and semanticists reveal that our natural inclinations in these respects are
mistaken. But we don’t think so. We think these strong and clear initial clas-
siﬁcations are correct and that semanticists who ignore them are led astray.

On the ﬁrst page of Kaplan’s classic ‘Demonstratives,’ there is a list of

expressions he calls indexicals. Slightly elaborated, his list goes like this: The
personal pronouns ‘I,’ ‘you,’ ‘he,’ ‘she,’ ‘it’ in their various cases and number
(e.g., singular, plural, nominative, accusative, genitive forms), the demon-
strative pronouns ‘that’ and ‘this’ in their various cases and number, the
adverbs ‘here,’ ‘there,’ ‘now,’ ‘today,’ ‘yesterday,’ ‘tomorrow,’ ‘ago’ (as in ‘He
left two days ago’), ‘hence(forth)’ (as in ‘There will be no talking hence-
forth’), and the adjectives ‘actual’ and ‘present’ (Kaplan 1989a, p. 489).
Words and aspects of words that indicate tense also have their reference so
determined. And there are also the contextuals, which include common
nouns like ‘enemy,’ ‘outsider,’ ‘foreigner,’ ‘alien,’ ‘immigrant,’ ‘friend,’ and
‘native’ as well as common adjectives like ‘foreign,’ ‘local,’ ‘domestic,’
‘national,’ ‘imported,’ and ‘exported’ (cf., Vallée 2003; Nunberg 1992; Con-
doravdi and Gawron 1995; Partee 1989).

To be honest, we have our doubts about so-called contextuals; and it’s probably no acci-

dent that they did not occur on Kaplan’s (1989) original list. We will let you decide for your-
self after you have read our book.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

In what follows, we shall refer to this set of expressions both as the Basic

Set of Context Sensitive Expressions (the Basic Set, for short) and as the set of
genuinely context sensitive expressions.

Why choose those expressions? Why didn’t he put, say, ‘red,’ ‘know,’

‘duck,’ ‘every,’ ‘good,’ or ‘happy’ in this set? Here’s an interesting fact about
Kaplan’s classic paper: He doesn’t give a reason. He never sees the need to
elaborate on, or defend, his choice of examples. In the end, he develops a
sophisticated theory of the semantics of demonstratives and other context
sensitive expressions. But his account presupposes that the domain he is
theorizing about is obvious and already identiﬁed.

One central goal in this book is to defend the uncontaminated intuitions

that underlie Kaplan’s methodology from a wide range of popular objec-
tions. In so doing, we also defend a certain view of the role of context sen-
sitivity in the semantics for natural language. It’s simultaneously a defense
of a certain conception of semantics and of a conception of semantic
content.

This ﬁrst chapter is intended just as an overview of what’s to come.

We don’t really engage in any serious argumentation here; we just quickly
present the views we advertised in our subtitle, i.e., Semantic Minimalism
and Speech Act Pluralism; we describe our central opponents (Radical
and Moderate Contextualists), the kind of arguments used by our adver-
saries; and at the end of the chapter we outline the book’s argumentative
strategy.

Introduction to Semantic Minimalism

At this introductory stage, we’ll just list three particularly important fea-
tures of Semantic Minimalism, all of which will be elaborated on, and
defended, later in the book (see in particular Chapter 10):

1 The most salient feature of Semantic Minimalism is that it recog-

nizes few context sensitive expressions, and, hence, acknowledges a
very limited effect of the context of utterance on the semantic
content of an utterance. The only context sensitive expressions are
the very obvious ones listed above plus or minus a bit. These are
not only obvious, but they pass certain tests for context sensitivity
that we spell out in Chapter 7.

2 It follows that all semantic context sensitivity is grammatically (i.e.,

syntactically or morphemically) triggered.

3 Beyond ﬁxing the semantic value of these obviously context sensi-

tive expressions, the context of utterance has no effect on the
proposition semantically expressed. In this sense, the semantic

Overview

content of a sentence S is the proposition that all utterances of S
express (when we adjust for or keep stable the semantic values of
the obvious context sensitive expressions in S).

Some illustrations: If we keep tense ﬁxed,

any utterance of (1)

(1) Rudolf is a reindeer

is true just in case Rudolf is a reindeer, and expresses the proposition that
Rudolf is a reindeer.

Any utterance of (2)

(2) Rudolf has a red nose

is true just in case Rudolf has a red nose, and expresses the proposition that
Rudolf has a red nose.

Any utterance of (3)

(3) Rudolf is happy

is true just in case Rudolf is happy, and expresses the proposition that Rudolf
is happy.

Any utterance of (4)

(4) Rudolf has had breakfast

is true just in case Rudolf has had breakfast, and expresses the proposition
that Rudolf has had breakfast.

Any utterance of (5)

(5) Rudolf doesn’t know that penguins eat ﬁsh

is true just in case Rudolf doesn’t know that penguins eat ﬁsh, and expresses
the proposition that Rudolf doesn’t know that penguins eat ﬁsh.

If you ﬁnd it surprising that someone would write a book defending con-

clusions so obvious, we have a great deal of sympathy. The problem is that
a wide range of our contemporary colleagues rejects these views (see below).
(It’s probably no exaggeration to say that our views about (1)–(5) are

Overview

As we will do throughout this book.

Semantic Minimalism, as understood in this book, need not take a stand on whether

semantic content is a proposition, or truth conditions, or what have you. Throughout the
book we try to remain neutral by couching the issues both in terms of truth conditions and
in terms of propositions.

currently held only by a small minority of philosophers and linguists, at
least among those who have thought about the surrounding issues.) This
book is our attempt to rebut these inﬂuential objections. A great deal of
that defense focuses on the relationship between speech act content and
semantic content, and in that respect Speech Act Pluralism plays a central
role.

Introduction to Speech Act Pluralism

Here’s one way to summarize Speech Act Pluralism (for fuller presentation
see Chapter 13):

No one thing is said (or asserted, or claimed, or . . .) by any utterance:
rather, indeﬁnitely many propositions are said, asserted, claimed,
stated. What is said (asserted, claimed, etc.) depends on a wide range
of facts other than the proposition semantically expressed. It depends
on a potentially indeﬁnite number of features of the context of utter-
ance and of the context of those who report on (or think about) what
was said by the utterance.

It follows from Speech Act Pluralism that an utterance can assert propo-

sitions that are not even logical implications of the proposition semanti-
cally expressed. Nothing even prevents an utterance from asserting (saying,
claiming, etc.) propositions incompatible with the proposition semantically
expressed by that utterance. From this, it further follows that if you want
to exploit intuitions about speech act content to ﬁx semantic content, then
you have to be extremely careful in so doing. It can be done, and we’ll show
you how, but it’s a subtle and easily corrupted process.

These points are connected to our defense of Semantic Minimalism

because one underlying assumption in many anti-minimalist arguments (in
particular, what we shall call the Context Shifting Arguments) is the idea
that semantic content has to be closely connected to speech act content. If
Speech Act Pluralism is correct, then there is no such tight connection, and
so, this requirement is revealed to be a philosophical prejudice. Another way
to see the connection is this: If there really were (or had to be) a close con-
nection between speech act content and semantic content, then all the data
we think support Speech Act Pluralism would also serve to undermine
Semantic Minimalism. That’s how some of the most clearheaded contex-
tualists argue. Our strategy is to endorse the data they invoke, but under-
mine their assumption that this data has semantic implications.

At this initial stage it’s worth highlighting one more aspect of Speech

Act Pluralism that both has wide ranging implications and sets our view
apart from (all?) other contemporary accounts of context sensitivity. We

Overview

don’t think that everything a speaker says by uttering a sentence in a
context C is determined by features of C. The speaker’s intentions, facts about
the audience, the place and time of utterance, background knowledge that’s
salient in C, the previous conversations salient in C, etc., are insufﬁcient to
ﬁx what the speaker said. According to Speech Act Pluralism, a theory of
speech act content has to take into account the context of those who say
or think about what the speaker said, i.e., the context of those who report
what’s said by the utterance can, in part, determine what was said by that
utterance. (As far as we can tell, we are on our own in defending this view;
we published a paper defending it in 1997 and don’t know of anyone else
who has endorsed it yet.)

Opponents of Semantic Minimalism

As we have already mentioned, a wide range of semantic theorists can advo-
cate Semantic Minimalism. Indeed, those who practice semantics accepting
these kinds of constraints tend to ﬁght ﬁerce internal battles. This book is
not a contribution to such rivalries. It’s about a range of arguments (below
we call them Context Shifting Arguments and Incompleteness Arguments)
which, if sound, would undermine the possibility of semantic theorizing.
Not all of those who employ these arguments realize the logical implica-
tions of doing so. Indeed, one of the points we’ll emphasize below is that
most proponents of these arguments operate under the illusion they can
be a part of ‘business as usual’ semantics. They don’t recognize the dangers
lurking right around the corner as soon as they start down this path.

The two central opponents of Semantic Minimalism we’ll be concerned

with we will call Radical and Moderate Contextualists. What they have in
common is that their positions are based on similar kinds of arguments.
We now outline these positions, and then, the kinds of arguments used by
their proponents.

Central Opponent 1: Radical

Contextualism (RC)

We want to engage two traditions according to which Semantic Minimal-
ism is fundamentally mistaken. One of these goes back to the Wittgenstein
of the Philosophical Investigations, on through Austin, and is today represented
by a wide range of philosophers, some of whom call themselves Relevance
Theorists,

some neo-Wittgensteinians, some Sellarsians. We call them all

Radical Contextualists. These theorists all hold some version or other of the

Overview

Cf., e.g., Sperber and Wilson (1986); Carston (1988, 2002), Recanati (1989, 1993, 2004).

view that every single expression is context sensitive,

and that the pecu-

liarities of members of the Basic Set are of no deep theoretical signiﬁcance.
Slightly more precisely, they adhere to some version of (RC1)–(RC3):

(RC1) No English sentence S ever semantically expresses a proposi-

tion. Any semantic value that Semantic Minimalists assign to
S can be no more than a propositional fragment (or radical), where
the hallmark of a propositional fragment (or radical) is that it
does not determine a set of truth conditions, and hence,
cannot take a truth value.

(RC2) Context sensitivity is ubiquitous in this sense: No expansion of

what we are calling the Basic Set of context sensitive expressions
can salvage Semantic Minimalism, i.e., however the Basic Set is
expanded, the output will never be more than a propositional
fragment; something, therefore, not even truth evaluable.

(RC3) Only an utterance can semantically express a complete propo-

sition, have a truth condition, and so, take a truth value.

Though they are not alone, John Searle and Charles Travis – without
acknowledging each other often (if at all) – are and have been for over thirty
years the chief spokespersons for RC.

. . . the notion of literal meaning of a sentence only has application
relative to a set of background assumptions, and furthermore, these
background assumptions are not and could not all be realized in the
semantic structure of the sentence in the way that presuppositions
and indexically dependent elements of the sentence’s truth condi-
tions are realized in the semantic structure of the sentence. (Searle
1978, p. 210)

What words mean plays a role in ﬁxing when they would be true; but
not an exhaustive one. Meaning leaves room for variation in truth
conditions from one speaking to another. (Travis 1996, p. 451)

. . . in general the meaning of a sentence only has application (it only,
for example, determines a set of truth conditions) against a back-
ground of assumptions and practices that are not representable as a
part of meaning. (Searle 1980, p. 221)

Both of these philosophers allude to Wittgenstein and Austin as their chief
inﬂuences (Travis 1985, p. 187; 1996, p. 451; Searle 1980, p. 229).

Overview

There are different ways of characterizing their views: For example, Every sentence is

context sensitive. Or, if the only context sensitivity you take into account is that due to the
expressions in the Basic Set, you won’t get a proposition or anything truth evaluable.

There is a sense in which we have a great deal of respect for RC. RC, we’ll

argue, is the logical consequence of denying Semantic Minimalism. As far
as we can tell, philosophers and linguists who try to modify Semantic
Minimalism only along the edges, by adding a bit of context sensitivity here
and there, fail to see that by so doing they lead themselves directly into the
clutches of RC.

Central Opponent 2:

Moderate Contextualism (MC)

The other opponents of Semantic Minimalism we are calling Moderate
Contextualists. Moderate Contextualists try to steer a middle course
between Semantic Minimalism and Radical Contextualism by minimally
expanding the Basic Set of context sensitive expressions. Slightly more pre-
cisely, Moderate Contextualists endorse some version of (MC1)–(MC3):

(MC1) The expressions in the Basic Set do not exhaust all the sources

of semantic context sensitivity.

(MC2) Many sentences that Semantic Minimalism assigns truth con-

ditions to, and treats as semantically expressing a proposition,
fail to have truth conditions or to semantically express a
proposition; they express only fragmentary propositions.
Such linguistic expressions are described as providing ‘incom-
plete logical forms,’ ‘semantic skeletons,’ ‘semantic scaffold-
ing,’ ‘semantic templates,’ ‘propositional schemas’ (see, e.g.,
Carston 2002; Sperber and Wilson 1986; Recanati 1993, 2004;
Bach 1994a; Taylor 2001). All of these locutions entail that
the expression is not fully propositional; it is incomplete qua
semantic entity; it is not truth evaluable.

(MC3) For the cases in question, only their utterances semantically

express a proposition, and have (interpretive) truth condi-
tions, and so, take a truth value.

Moderate Contextualists don’t typically see themselves as belonging to a
tradition or a group and they wouldn’t classify themselves as Moderate
Contextualists. There are two kinds of Moderate Contextualists: Misguided
Semanticists and Semantic Opportunists.

The Misguided Semanticists come to MC by noticing some data or evi-

dence they think has to be accounted for by a semantic theory (we will
describe data of this kind below). They don’t see how to account for it
except by expanding the Basic Set.

The Semantic Opportunists are sneakier. They are philosophers who

come to semantics with a nonsemantic agenda. They might be concerned
with defending a view in epistemology, ethics, philosophical logic, philos-

Overview

ophy of mind, metaphysics, etc. They have no interest in, or understanding
of, the overall semantic project. They postulate that various expressions are
context sensitive because doing so lends support to a view, usually radical,
they endorse in their respective area.

Paradigm examples are ethicists who claim that ethical terms are context

sensitive; epistemologists who claim that certain epistemic terms are
context sensitive; metaphysicians who claim that vague terms are context
sensitive; philosophical logicians who claim that quantiﬁers or certain
semantic terms are context sensitive, and so on.

For our purposes, what motivates Moderate Contextualists doesn’t really

matter. What does matter is how MC is implemented. Here’s what we have
in mind: Let’s assume, for the sake of argument, that Moderate Contextu-
alists hold that some expression e, not in the Basic Set, is context sensitive.
Remember, they do not think, as Radical Contextualists do, that semantics
is impossible. They therefore face a range of additional questions about how
a semantic theory should accommodate this additional context sensitivity.

If you have evidence that e is context sensitive and you want that incor-

porated into a semantic theory, primarily three basic strategies are available
to you: the Surprise Indexical Strategy, the Hidden Indexical Strategy, and
the Unarticulated Constituent Strategy. Here, in very brief outline, is each
option.

The Surprise Indexical Strategy

The Surprise Indexical Strategy is the most straightforward of the three. If
you opine that an expression e is context sensitive, then add e to the Basic
Set, thereby treating it as an indexical, in the exact same way that ‘I’ and
‘that’ are indexicals. So, some epistemologists, e.g., Lewis, DeRose, or
Cohen, think that knowledge attributions exhibit context sensitivity. This
leads them to treat the verb ‘to know’ as context sensitive. One way to incor-
porate this contextualist view into semantics is to treat ‘know’ as an index-
ical expression in a straightforward manner: The semantic value of ‘know’
changes from one context of utterance to another. As a result, what’s
required for satisfying, say, ‘Lewis knows that penguins eat ﬁsh’ varies from
one context of utterance to another, contingent, say, on rising or falling
epistemic standards. Commitment to epistemological contextualism in this
manner thereby commits one to extending the Basic Set to include ‘know’
in addition to ‘I,’ ‘here,’ etc.

The Hidden Indexical Strategy

The Hidden Indexical Strategy postulates a phonetically unrealized com-
ponent (a covert indexical) at some level of linguistic representation, say, in

Overview

Logical Form. Rather than treating a surface (overt) expression e itself as an
indexical (as the Surprise Indexical Strategy does), the Hidden Indexical
Strategy accounts for alleged context sensitivity by ﬁnding (or postulating)
a ‘hidden’ (i.e., unpronounced or covert) indexical associated with the
expression(s) we hear pronounced. For example, most philosophers and lin-
guists think that sentences with comparative adjectives are context sensi-
tive. They hold that when someone utters, for example, ‘Bill is short,’ there’s
an unpronounced indexical associated with ‘short’ that makes reference to
a comparison class. For any utterance of this sentence, you don’t hear ‘for
an F’ or anything like it; rather, what you hear is just ‘short.’ But in the
underlying logical–syntactic form of the sentence, there’s alleged to be a
(covert) lexical item that refers in context to a comparison class. Again, there
are many ways to achieve this end formally, but the basic idea is to take the
logical form of ‘Bill is short’ to be something along the lines of ‘Bill is short
for an F,’ where ‘F’ can vary from one context of utterance to another.

The Unarticulated Constituent Strategy

The Unarticulated Constituent Strategy ﬁnds context sensitivity in certain
sentences, but does not recommend treating any pronounced or unpro-
nounced component of that sentence as the source of this context sensi-
tivity. According to this view, a propositional component gets added
without being triggered by a syntactic component (pronounced or unpro-
nounced) in the uttered sentence.

For example, consider the sentence ‘It’s

raining.’ Perry (1986) claims that unless the proposition semantically
expressed by an utterance of this sentence included a location, it would not
be ‘complete,’ and so, would not be truth evaluable. However, there’s no
expression in the logical–syntactic form of this sentence that makes refer-
ence to a location. Instead, the location is somehow or other added to the
proposition semantically expressed by an utterance of the sentence without
its being required by any lexical item in the sentence.

Methodological observation: MC and RC are supported by

only two kinds of arguments

Here’s a methodological observation that underlies the entire rest of this
book:

Overview

‘An indexical is like a free variable needing to be assigned a value. . . . the conceptual gaps

in utterances of semantically underdeterminate sentences do not correspond to anything in
the sentences themselves . . . Not being sentence constituents, they enter in not at the lin-
guistic level but at the conceptual level . . .’ (Bach 1994a, p. 133).

Methodological observation. There are two basic kinds of argument
adduced in favor of all versions of RC and MC: Context Shifting Argu-
ments and Incompleteness Arguments.

These two kinds of argument are the central motivation behind all depar-
tures from Semantic Minimalism.

This observation about the literature on (semantic) context sensitivity is

meant to be substantial and controversial. If we are right, then a wide range
of apparently diverse philosophical positions rely solely upon two kinds of
argument. Chapter 2 is devoted to presenting textual evidence in support
of this claim. Here we give a rather brief introduction to what we mean by
Context Shifting Arguments (CSA) and Incompleteness Arguments.

Context Shifting Arguments (and a preview of how they are misused)

Someone in the business of investigating context sensitivity contemplates
and imagines language as used in contexts other than the one she happens
to ﬁnd herself in. She is, after all, interested in the way in which content is
inﬂuenced by variation in the context of utterance; in particular, she tries
to elicit intuitions about whether what is said, or expressed by, or the truth con-
ditions of, an utterance varies in some systematic way with contexts of utter-
ance. To do so, she imagines a range of utterances, u

–u

, of a sentence S.

The resulting data consists of her reports of, and the audience’s own, in-
tuitions about the content of u

–u

. Arguments that appeal to this kind of

evidence we call Context Shifting Arguments.

Here’s a preview of what we’ll argue later: The literature on context sen-

sitivity is plagued by a blatant misuse of this kind of argument. The mistake
is not simply of the kind Grice pointed out, i.e., that theorists have to dis-
tinguish between intuitions about what utterances say and what they impli-
cate. The way we see it, that mistake is superﬁcial and relatively easy to avoid.
Rather, the fundamental mistake in the entire contextualist literature is
this: To properly engage in this sort of thought experiment a theorist has
to locate herself in a particular context. To not make the context of the
thought experiment an essential variable of the experiment is like trying to
measure the speed of objects around you while ignoring your own speed.
You can’t do it. This mistake, we argue, is exactly the one that both Radical
and Moderate Contextualists are guilty of.

If our metaphorical presentation of the problem seems obscure, bear

with us until Chapters 7–9, where full details and clariﬁcation will be
provided.

Overview

Incompleteness Arguments (and a preview of how they are misused)

The second kind of argument in the literature on context sensitivity we call
Incompleteness Arguments. These also require an appeal to intuition, but
an appeal to a kind of metaphysical intuition rather than to a linguistic
one. The goal of an Incompleteness Argument is to establish that the
proposition Semantic Minimalists claim is semantically expressed by some
sentence S is no more than a propositional fragment.

Incompleteness Arguments are always simple (so simple that they might

not deserve the moniker ‘argument’). Typically, all they amount to is a claim
like the following:

Consider the alleged proposition that P that some sentence S seman-
tically expresses. Intuitively, the world can’t just be P simpliciter. The
world is neither P nor not P. There’s no such thing as P’s being the
case simpliciter. And so, there is no such proposition.

So, for example, consider ‘Al is ready.’ Some authors contend that it is just
plain obvious that there isn’t any such thing as Al’s being ready simpliciter.
Likewise, Perry (1986) and Crimmins (1992) argue, for example, that with
a range of weather or temporal reports (containing pleonastic ‘it’s’), as in
‘It’s raining’ and ‘It is 3 p.m.,’ there’s no such thing as raining simpliciter or
as being 3 p.m. simpliciter.

Again, a preview of our central contentions about Incompleteness

Arguments: First, Moderate and Radical Contextualists who use such argu-
ments are typically deeply confused about the relationship between seman-
tics and metaphysics. These arguments are not about language; they are
about various nonlinguistic aspects of the world. Even if they were good
arguments, nothing would follow about the sentences in question, more
generally, no semantic conclusions follow from these arguments even if
they were sound. Second, considered, as they ought to be, as metaphysical
arguments, they are unsound.

Comparison with Other Ways of

Structuring the Debate

The way we have presented the debates about context sensitivity (as a debate
between Semantic Minimalism, MC, and RC) is controversial. It is, for
example, not how all of the participants of these debates think of them.
More speciﬁcally:

Overview

䊉

Moderate and Radical Contextualists do not see themselves as
aligned with each other, differing only with respect to where they
are located on a continuum (the former wanting more of something
that the latter wants not as much of). The Radical Contextualists
see the Moderate Contextualists as ﬁerce opponents, and vice versa.

䊉

The three different versions of MC (Surprise Indexicalists, Hidden
Indexicalists, and proponents of Unarticulated Constituents) do
not think of each other as holding different versions of the same
view. Advocates of each of these views spend a great deal of time
arguing against the other two.

䊉

Within each version of MC, there is disagreement about which
expressions should be added to the Basic Set.

䊉

Even those proponents of MC that agree on some version of MC
and about which expressions should be added to the Basic Set dis-
agree about how the versions should be implemented. For example,
Moderate Contextualists who are Hidden Indexicalists about quan-
tiﬁed noun phrases disagree about each of the following:

䊊

The nature of the semantic value of the hidden indexical (a class
or a property or something else).

䊊

The larger semantic frameworks that this view should be embed-
ded in.

䊊

Various issues concerning how semantic values of the hidden
indexicals are ﬁxed (‘wide’ or ‘narrow’ context).

䊊

Where to place the hidden indexical: attach it to the quantiﬁer,
to the noun phrase, as separate lexical entry or as ‘co-habitating’
with some other expression.

䊉

There’s a lively debate among Radical Contextualists, and many of
those we so classify do not think of themselves as holding versions
of the same view. They spend huge chunks of time arguing with
each other about the differences between ‘enrichment,’ ‘saturation,’
‘free enrichment,’ ‘concept construction,’ and a wide range of other
issues.

Not only does our structuring of the debate lump together philosophers
and linguists who would rather not be lumped together, but it might also
(in part, as a result of this (apparently) crude classiﬁcation) seem to miss
what some think of as the deep and underlying issues. In particular, we have
heard the following suggestions for what these debates are really about:

1 Deep down it is all about compositionality. Roughly, the central issue

in all these debates is whether it is possible to develop a composi-
tional semantics for natural language. The interesting question is
not about context sensitivity as such, but about how it affects
compositionality.

Overview

2 Deep down it is all about whether we need to take speaker’s intentions

into account to ﬁx semantic values. In a terminology often used, it
is all about whether semantics needs to take into account ‘wide’
context in addition to ‘narrow’ context (of the kind Kaplan seemed
to focus on in his paper ‘Demonstratives’).

In sum, we have encountered various charges to the effect that we have
failed to see what these debates are really all about deep down and that our
structure leaves out important distinctions.

That, unsurprisingly, is not now how we see things. We do, of course,

agree that there are many interesting, deep, and subtle issues about context
sensitivity not addressed in this book. In no way do we mean our discus-
sion to be exhaustive. But to leave it at that would be to understate our case.
We organize the various positions as we do because we think so doing elicits
(renders explicit) the fundamental assumptions shared by positions that
conceive of themselves as fundamentally opposed. Four substantive and
controversial views underlie our organization of the debate:

1 All opponents of Semantic Minimalism (be they some version of

MC or some version of RC) share certain important assumptions.

2 These assumptions seem so obvious to opponents of Semantic

Minimalism that they are almost never made explicit and when
they are made explicit they are never convincingly defended.

3 We argue that all of these underlying, shared assumptions should

be rejected.

4 If these shared assumptions are rejected, then:

(a) Most (maybe all) of the arguments against Semantic Minimal-

ism collapse.

(b) The distinction between various versions of MC and RC will

seem unimportant (since these questions don’t even arise
unless one makes certain false assumptions).

to a range of issues independent of any debates internal to MC
or RC.

One underlying assumption (the simplest version of which we call the Mis-
taken Assumption – MA, for short) is spelled out in Chapter 4. In brief, it
is the view that the semantic content of a sentence S is constrained in
certain ways (spelled out in Chapter 4) by what speakers can use S to say
(assert, claim, state, etc.) and that intuitions about what speakers say
(assert, claim, state, etc.) with a sentence S provide evidence for the seman-
tic content of S. This can also be described as the mistake of conﬂating
semantic content and speech act content. (This, we further argue in Chapter 4,

Overview

is what underlies the constant misuse of Context Shifting Arguments in
the philosophy of language, and elsewhere.)

Outline of Argumentative Strategy

In Chapter 2 we document that in a wide range of cases, indeed, in all of
the cases we are aware of, arguments to the effect that an expression e
exhibits semantic context sensitivity are based either on some version of a
Context Shifting Argument or on some version of an Incompleteness Argu-
ment. (Other arguments occasionally presented as arguments for context
sensitivity are shown to be parasitic on these two kinds of argument.)

In Chapters 3–6 we show that any attempt to exploit these two kinds of

argument to expand the Basic Set of context sensitive expressions to one
any larger is susceptible to an instability charge. The charge takes this form:
We consider a range of data, D, presented in favor of expanding the Basic
Set of context sensitive expressions, and show that if this evidence supports
an expansion of the Basic Set, then all expressions are context sensitive, i.e.,
RC is true. In other words, we will establish that any argument for MC
inevitably slips into an argument for RC.

In Chapters 7–9 we show that RC is, ﬁrst, empirically ﬂawed, and worse,

ultimately incoherent. Since MC collapses into RC, it follows that MC also
is both empirically ﬂawed and ultimately incoherent.

In Chapters 10–12 we present and defend Semantic Minimalism.
In Chapter 13 we present and defend Speech Act Pluralism.

Overview

PA R T I

From Moderate to Radical

Contextualism

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

C H A P T E R 2

Exegesis: The Methodology of

Contextualism

Much of this book consists of discussions, criticisms, and reﬁnements of
what we call Context Shifting and Incompleteness Arguments. We focus on these
because they are at the center of all arguments that attempt to estab-
lish that expressions or sentences exhibit semantically relevant context
sensitivity.

While presenting this material in seminars and at conferences we some-

times encounter the following reply: Your objections to these arguments might
be good, but so what? Those aren’t the kinds of arguments contextualists invoke.
What’s really going on is . . . This chapter is meant in part as a response to this
sort of reply and in part as an introduction to the topic for those not already
immersed in the literature.

Context Shifting Arguments

Our view is that sentences are context sensitive just in case they contain an
expression from what we call the Basic Set of Context Sensitive Expressions.
Suppose someone suspects that an expression e, not in that set, is context
sensitive. How could he go about establishing this? One way that philoso-
phers of language do so is to think about (or imagine) various utterances
of sentences containing e. If they have intuitions that a semantically relevant
feature of those utterances varies from context to context, then that, it is
assumed, is evidence e is context sensitive.

For this strategy to work it is of course important that the features one

has intuitions about are semantically relevant. The kinds of features that
contextualists claim to have intuitions about include:

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

䊉

What is said or asserted or claimed or stated by utterances of sentences
containing e.

䊉

The truth conditions of utterances of sentences containing e.

䊉

The proposition expressed by utterances of sentences containing e.

In Chapters 7–9 we argue that intuitions about variability in these

features do not, even prima facie, provide evidence of semantic context
sensitivity. In this chapter we simply document extensive appeal to such
intuitions in the philosophical and linguistics literature. We’ll begin by
discussing a few of the speciﬁc cases and then turn to the more general
case.

Quantiﬁers

We start with quantiﬁer sentences; these are sentences which include quan-
tiﬁer expressions like ‘every bottle,’ ‘no man,’ ‘the table,’ etc. Stanley and
Williamson, for example, appeal to intuitions about context shifting in
defending their claim that quantiﬁer sentences are context sensitive when
they write:

Since there are clearly true utterances of

(1) Every bottle has been put on the shelf.

in ordinary contexts, it follows that utterances of sentences contain-
ing quantiﬁed expressions are evaluated with respect to contextually
restricted domains. (Stanley and Williamson 1995, p. 291)

Their intuition is that the truth values of utterances of (1) can shift from
one context to another even though the locations of all extant bottles
change not at all. This is so, they claim, because the domain of the quanti-
ﬁer ‘every bottle’ itself shifts from context to context of utterance. So, in
this case, it is intuitions about shifting truth values that underlie an infer-
ence to context sensitivity.

Recanati concurs. In discussing an utterance of the quantiﬁer sentence

‘They took everything,’ he writes that he has ‘a feeling that “everything”
ranges over the domain of valuable objects in the house – not everything in
the world’ (Recanati 1996, p. 445). Recanati’s feelings (or intuitions) are
about what quantiﬁers range over; in effect, what their application condi-
tions are, and so, what the truth conditions are for utterances of sentences
in which these expressions occur. If Recanati’s intuitions are correct, it
follows that quantiﬁer sentences admit of context sensitivity.

From Moderate to Radical Contextualism

Lewis, on the other hand, has intuitions about what utterances of quan-

tiﬁer sentences say; he writes: ‘If I say that every glass is empty, so it’s time
for another round, doubtless I and my audience are ignoring most of all
the glasses there are in the whole wide world throughout all of time. They
are outside the domain. They are irrelevant to the truth of what was said’
(Lewis 1996, p. 225). In these cases, Lewis’s intuitions are that utterances of
quantiﬁer sentences can change in what their utterances say, and so, in their
truth conditions – contingent upon what the restricted domain of the
quantiﬁer is.

Schiffer focuses neither on domains of discourse nor on truth values

shifting from context to context, but rather on what is stated by distinct
utterances of quantiﬁer sentences, in particular, by utterances of sentences
with deﬁnite descriptions in them: ‘it is clear that in uttering “The dog is
chewing your hat” the literal speaker is not stating something that entails
that there is exactly one dog in the universe’ (Schiffer 1998, p. 375–6). Neale
likewise draws contextualist conclusions based on his intuitions about
changes in what speakers are asserting with utterances of the same sentence
as context shifts from occasion to occasion.

Suppose I had a dinner party last night. In response to a question as
to how it went, I say to you:

(2) Everyone was sick.

Clearly I do not mean to be asserting that everyone in existence was
sick, just that everyone at the dinner party I had last night was. (Neale
1990, p. 95)

Stanley and Szabó have the intuition that which propositions are semanti-
cally expressed can shift from context to context, so that distinct utter-
ances of the same unambiguous quantiﬁer sentence can convey different
propositions:

Consider the sentence:

(1) Every bottle is empty.

Suppose someone utters (1) in a conversation. It is unlikely that what
she intends to convey is that every bottle in the universe is empty; she
most likely intends to convey that every one of a restricted class of
bottles (say, the bottles in the room where she is, the bottles pur-
chased recently, etc.) is empty. And if the context is right, she can
succeed in communicating such a proposition. Permanent linguistic
features of (1) – its phonological and morphological constituents, its
syntactic structure, the meanings of the lexical items it contains – do
not determine the proposition thereby communicated. They cannot

Exegesis: The Methodology of Contextualism

do so, for these features are the same on every occasion when the sen-
tence is used, but on most of those occasions the speaker would com-
municate a different proposition by the sentence. (Stanley and Szabó
2000a, pp. 219–20)

Though Stanley and Szabó’s claim is also one about context shifting, their
intuitions are about context shifts in the proposition semantically
expressed by distinct utterances of (2). (This of course may elicit a change
in truth value as well.)

We ﬁnd a similar idea in Gaulker:

Imagine a goatherd in the Peruvian Andes whose community has long
been isolated from the rest of the world. One evening all the people
of the village are gathered for a traditional celebration and there
appears in the sky a remarkable bright falling star. Everyone looks up
into the sky and sees it. As a result our goatherd forms a belief that
he attempts to convey in the words that translate thus: ‘Everyone saw
the falling star’. Call this the goatherd’s ﬁrst utterance.

Sometime later, our goatherd is out in the hills accompanying a

philosophical friend. Bored with tending goats, the philosopher asks
the goatherd, ‘Do you think there might be people like us on the other
side of these distant mountain tops?’ For the ﬁrst time our goatherd
contemplates the question and forms the opinion that, yes, very prob-
ably, there are other people over there . . . To convey this thought, he
chooses the words that translate thus: ‘Not everyone in the universe
is a member of our community’. Call this the goatherd’s second
utterance.

A charitable interpretation would say that by means of his ﬁrst

utterance the goatherd intended to convey the proposition that every-
one in the goatherd’s community saw the falling star, and that by
means of his second utterance the goatherd intended to convey the
proposition that not everyone in the universe is a member of the
goatherd’s community. (Gaulker 1997, pp. 17–19)

In sum, each author invokes an intuition or feeling about quantiﬁer sen-

tences in use; each employs intuitions or feelings about distinct utterances
of quantiﬁer sentences; namely, that there is a shift in evaluations of these
utterances across distinct contexts. What shifts is either

䊉

the application condition of the quantiﬁers and so the truth con-
ditions of the utterances of sentences in which they occur, and so,
possibly the truth values of these utterances, or

From Moderate to Radical Contextualism

䊉

which propositions speakers semantically expressed with their
utterances, or

䊉

what they are asserting (stating, afﬁrming, conveying) with these
utterances.

It may turn out that these different sets of intuitions all co-vary and have
a common cause – namely, the context sensitivity of quantiﬁer expressions.
But that would need to be argued for. Still, each is some version of a Context
Shifting Argument.

Comparative Adjectives

Although commitment to contextualism about quantiﬁer expressions is
relatively commonplace, with appeals to intuitions about context shifting
being the chief defense, commitment to contextualism about comparative
adjectives is virtually universal. Once again appeal to intuitions about
context shifting is the chief contextualist defense. Some authors assume it
is truth conditions that vary:

Consider the class of gradable adjectives, those which take the com-
parative and superlative; ‘rich’, ‘urgent’, ‘dangerous’, ‘tall’, and
‘square’ are examples. Almost everyone agrees that these are contex-
tually sensitive, in the sense that context provides the adjective a para-
meter necessary for it to determine (even) a (vague) extension. . . . It
is, I think, beyond serious dispute that the truth conditions of ‘Mary
is rich’ vary across contexts, as vary the interests, focus, and so on of
participants in a conversation. (Richard 2004, pp. 218–19; cf., also,
Higginbotham 1985, pp. 563–5; Parsons 1990, pp. 42–4)

Other authors hold that it is what’s said that varies:

Consider predicative uses of a comparative adjective, such as:

(26) That building is small.
(27) That basketball player is short.
(28) That ﬂea is small.

On one natural reading of (26), the building in question is not being
said to be small for an object in general (whatever that may mean).
Rather, the building is being said to be small for a building. Similarly,
on a natural reading of (27), the basketball player in question is not
being said to be short for a person, but only for a basketball player.
(Stanley 2002b, p. 377)

Exegesis: The Methodology of Contextualism

According to Richard, what shifts from context to context is truth

conditions; according to Stanley, what shifts from context to context is
what speakers say when they use sentences with comparative adjectives,
even though the heights and sizes of the individual in question remain
constant.

Clapp’s intuitions about context shifting for comparative adjectives

concern their truth values:

An utterance of (4) (‘Bradley is tall’) that occurred in a discussion con-
cerning the physical characteristics of presidential candidates would
be true iff Bradley is tall for a presidential candidate, while an utter-
ance of (4) that occurred in a discourse concerning great centers in
the NBA would be true iff Bradley is tall for a great center in the NBA.
Thus, the truth conditions of (4) depend upon what contrast class is
invoked by the sentence. (Clapp 2002, p. 237)

Propositional Attitude Ascriptions

Another fragment for which context shifting intuitions are presented in
defense of contextualism concerns propositional attitude ascriptions. For
example, Crimmins and Perry present an inﬂuential theory about belief
reports according to which such reports are context sensitive. Belief reports
have ‘notions’ as unarticulated constituents and these unarticulated con-
stituents vary based on what is contextually salient. Three claims are at the
center of their theory:

1 That the notions vary across conversational contexts.
2 That this variation corresponds to differences in communicated

content because notions are unarticulated constituents of belief
reports.

3 That these variations in communicated content can help solve clas-

sical puzzles involving belief reports.

Here’s a summary of their view:

We take a belief report to be an utterance u of a belief sentence, of the
form:

A believes that S

where A is a singular term and S is a sentence. . . . The claim made by
the belief report is that the agent a has a belief with the content p,
involving notions n

. . . n

(in a certain way). . . . We shall say in such

From Moderate to Radical Contextualism

cases that the notions that the belief report is about are provided by
the utterance and its context. (Crimmins and Perry 1989, pp. 263–4)

According to Crimmins and Perry there is no expression in the logical form
of the sentence corresponding to these contextually provided notions:

On our account, the complex relation invoked in belief reports is a
four-place relation: an agent believes a proposition at a time relative
to a sequence of notions. But there is no argument place in the
‘believes’ predicate for the sequence of notions. The notions are unar-
ticulated constituents of the content of the report. (Crimmins and
Perry 1989, pp. 264–5)

Belief reports, according to this view, ‘call for a propositional constituent
that meets, say, certain conditions of relevance and salience’ (Crimmins and
Perry 1989, p. 266; emphasis our own).

Our interest is in the role of CSAs in their argument. Here’s a way to

think about the way they defend their view: They take a range of puzzles
involving belief reports and show that those puzzles can be solved on the
assumption that the relevant notions vary from one context of utterance to another.

Here’s an illustration involving Kripke’s puzzling Pierre case:

In the Pierre case, the sentence (3) [‘Pierre believes that London is
pretty’] gets used in two reports, ﬁrst in a discussion of Pierre’s initial
acquaintance with London through stories, then in a discussion
about Pierre’s thoughts about his adopted home. Call these reports
U3 and U

¢3. Pierre actually has two notions of London, one relevant

to each discussion; call the ﬁrst n and the second n

¢. The notion n

meets the condition C of being a notion germane to the discussion
of Pierre’s reaction to these stories; the notion n

¢ meets the condition

¢ of being a notion germane to the discussion of Pierre’s new home.

(Crimmins and Perry 1989, p. 273)

Here is one way in which this variation in contextually relevant notions can
contribute to the analysis of the problematic belief reports:

The speaker of the former report is claiming that Pierre has a belief
involving some notion germane to the current conversation about the
stories, with the content that London is pretty. The speaker of the
latter report requires that the belief involve some notion relevant to
the conversation about Pierre’s new home. (Crimmins and Perry 1989,
p. 273)

Exegesis: The Methodology of Contextualism

1 For elaboration of this point, see Crimmins and Perry (1989, pp. 272–3).

At the center of this argument one ﬁnds an appeal to the intuition that the
claim made by a belief report is determined by the contextually salient or
‘germane’ notion, and hence, varies depending on which notions are salient
or germane. The rest of Crimmins and Perry (1989) and Crimmins (1992)
provides a plethora of further illustrations of appeals to context shifting
intuitions.

Here’s another variation on this kind of argument, this one from Clapp:

We are observing Jerry who is tasting the cookies from a plate with a
tag that reads ‘Ms. O’Connor’ . . . suppose that we know that Jerry
does not realize that the baker of the cookies, Ms O’Connor, just is
his acquaintance Marie . . . I thus say, ‘Ha! Poor Jerry does not know
that “O’Connor” is Marie’s last name, so he doesn’t know that those
are Marie’s cookies!’ If you were to utter (5) [Jerry believes that Marie
baked the cookies] immediately following my statement, your utter-
ance would be false . . . But . . . suppose that we are at a cookie baking
contest, and suppose that Jerry, whom we know to have no prior
acquaintance with Ms. Marie O’Connor, observes her at a distance
placing her cookies on a plate. Seeing Jerry observe Marie putting her
cookies on a plate, I utter (5) to you. In this context my utterance of
(5) is (probably) true. . . . So (5) is clearly context sensitive . . . (Clapp
2002, p. 238)

Clapp’s intuition or feeling is that distinct utterances of the same attitude
ascription can disagree in truth value. The natural way for him to explain
his intuitions is to infer that psychological attitude ascriptions are context
sensitive.

Counterfactual Conditionals

The general view about counterfactual conditionals is that they

are not categorically true or false but only relative to a set of implicit
background assumptions. Utterances of both of the following could
be true if different background assumptions are held ﬁxed.

(15) If Lincoln hadn’t gone to the theatre, he wouldn’t have been

assassinated.

(16) If Lincoln hadn’t gone to the theatre, he would have been

assassinated anyway.

This suggests that these conditionals do not express complete
propositions as they stand. (Bach 1994a, pp. 128–9)

From Moderate to Radical Contextualism

An inﬂuential version of this view is developed by Lewis. According to Lewis,
the truth conditions for counterfactuals vary from one context of utterance
to another and they do so because they appeal to the similarity relation
between worlds. Whether two worlds are relevantly similar depends on the
context of utterance. As a result counterfactuals are context sensitive. Here’s
a passage in which Lewis describes the alleged variability of the similar-
ity relation (and, by implication, the variability of truth conditions for
counterfactuals):

All this is not special to the comparative similarity of worlds that
appears in my analysis of counterfactuals. It is the same sort of vague-
ness that arises if I say that Seattle resembles San Francisco more
closely than it resembles Los Angeles. Does it? That depends on
whether we attach more importance to the surrounding landscape,
the architecture, the dominant industries, the political temper, the
state of the arts, the climate, the public transportation system, the
form of the city government, or what. Possible worlds are bigger than
cities (sometimes) and are capable of differing in a greater variety of
respects. . . . Still, any problems posed by my use of comparative simi-
larity differ only in degree, not in kind, from problems about simi-
larity that we would be stuck with no matter what we did about
counterfactuals. (Lewis 1973, p. 92)

The respects of similarity we attach importance to vary between contexts,
but counterfactuals don’t allow just any kind of variability:

There is a rough consensus about the importance of respects of com-
parison, and hence about comparative similarity. Our standards of
importance and similarity do vary; but mostly within a certain range,
narrow by comparison with the range of variation permitted by the
formal constraints in my deﬁnition of a system of spheres. (Lewis
1973, pp. 93–4)

Here is how this applies to a speciﬁc example. Lewis writes about Quine’s
pair of counterfactual conditionals

If Caesar had been in command [in Korea] he would have used the
atom bomb
If Caesar had been in command he would have used catapults

as follows:

In dealing with Quine’s opposed counterfactuals about Caesar,
context must of course be consulted somehow . . . I could . . . call on

Exegesis: The Methodology of Contextualism

context . . . to resolve part of the vagueness of comparative similarity
in a way favourable to the truth of one counterfactual or the other.
In one context, we may attach great importance to similarities and
differences in respect of Caesar’s character and in respect of regular-
ities concerning the knowledge of weapons common to commanders
in Korea. In another context we may attach less importance to these
similarities and differences, and more importance to similarities and
differences in respect of Caesar’s own knowledge of weapons. The ﬁrst
context resolves the vagueness of comparative similarity in such a way
that some worlds with a modernized Caesar in command come out
closer to our world than any with an unmodernized Caesar. It thereby
makes the ﬁrst counterfactual true. The second context resolves the
vagueness in the opposite direction, making the second counterfac-
tual true. (Lewis 1973, p. 67)

Two kinds of context shifting intuitions are at the center of Lewis’s
argument:

1 The intuition that the truth conditions for counterfactuals depend

on the topic of conversation, the assumed background knowledge,
and more generally, salient features of the context of utterance.

2 The intuition that this variability can be captured by, and is

reﬂected in, the variability in similarity judgments. That variability
is, again, justiﬁed by appeals to intuitions about how the truth con-
ditions of a sentence of the form ‘A is similar to B’ vary between
contexts.

Knowledge Attributions

An area of philosophy where contextualism has really taken hold is in epis-
temology. Epistemic contextualism is often invoked to solve traditional
epistemic puzzles/paradoxes, e.g., the Lottery Paradox, and Gettier and
Skeptical problems. These contextualists all appeal to intuitions that speak-
ers have about context shifts about knowledge attributions. Here’s an
example from DeRose:

From Moderate to Radical Contextualism

2 A related kind of argument is used to show that conditionals are context sensitive. The
basic idea goes back to Ramsey and is expressed in the following passage: ‘In general we can
say with Mill that “If P then Q” means that Q is inferable from P, that is of course, from P
together with certain facts and laws not stated but in some way indicated by the context ’ (Ramsey
1978, p. 247; emphasis our own).

Bank Case A: My wife and I are driving home on a Friday afternoon.
We plan to stop at the bank on the way home to deposit our pay-
checks. But as we drive past the bank, we notice that the lines are very
long, as they often are on Friday afternoons. Although we generally
like to deposit our paychecks as soon as possible, it is not especially
important in this case that they be deposited right away, so I suggest
that we drive straight home and deposit our paychecks on Saturday
morning. My wife says, ‘Maybe the bank won’t be open tomorrow.
Lots of banks are closed on Saturdays.’ I reply, ‘No, I know it’ll be
open. I was just there two weeks ago on Saturday. It’s open until
noon.’

Bank Case B: My wife and I drive past the bank on a Friday afternoon,
as in Case A, and notice the long lines. I again suggest that we deposit
our paychecks on Saturday morning, explaining that I was at the bank
on Saturday morning only two weeks ago and discovered that it was
open until noon. But in this case, we have just written a very large and
important check. If our paychecks are not deposited into our check-
ing account before Monday morning, the important check we wrote
will bounce, leaving us in a very bad situation. And, of course, the
bank is not open on Sunday. My wife reminds me of these facts. She
then says, ‘Banks do change their hours. Do you know the bank will
be open tomorrow?’ Remaining as conﬁdent as I was before that the
bank will be open then, still, I reply, ‘Well, no. I’d better go in and
make sure.’ (DeRose 1992, pp. 920–1)

DeRose comments that the ‘contexts of my utterance in the two cases make
it easier for a knowledge attribution to be true in Case A than in Case B’
(DeRose 1992, p. 920). He writes about contextualism in general that:

Once the standards have been so raised, we correctly sense that we only
could falsely claim to know such things as that we have hands . . . [and]
as soon as we ﬁnd ourselves in more ordinary conversational contexts, it will
not only be true for us to claim to know the very things that the skeptic
now denies we know, but it will also be wrong for us to deny that we
know these things (DeRose 1995, p. 185; emphasis our own).

In these cases DeRose is appealing to intuitions about the truth values of
claims we are making with distinct utterances of the same unambiguous
knowledge attribution in distinct contexts of utterances, even though all of
the nonepistemological facts remain constant. Along the same lines, he
concludes, ‘the truth conditions of sentences of the form “S knows that p”
or “S does not know that p” vary in certain ways according to the context

Exegesis: The Methodology of Contextualism

in which they are uttered’ (DeRose 1992, p. 914; cf., also, 1995, sect. 8;
Casteneda 1980).

Other authors, e.g., Cohen, argue that what shifts is the threshold

required for justiﬁcation, where for any given subject his belief that p is
assigned some absolute degree d of justiﬁcation for p and what shifts is
whether d sufﬁces for justiﬁcation. Cohen writes, in consequence, that:

Contextualism is the view that . . . the truth-values of sentences con-
taining ‘know’, and its cognates depend on contextually determined
standards. Because of this, sentences of the form ‘S knows P’ can, at
one time, have different truth-values in different contexts. Now when
I say ‘contexts’, I mean ‘contexts of ascription’. So the truth-value of
a sentence containing the knowledge predicate can vary depending on
things like purposes, intentions, expectations, presuppositions, etc.,
of the speakers who utter these sentences. (Cohen 1999, p. 57)

Suppose one speaker says about a subject S and a proposition P, ‘S
knows that P.’ At the very same time, another speaker says of the very
same subject and proposition, ‘S does not know P.’ Must one of the
two be speaking falsely? According to the view I will call ‘contextual-
ism’, both speakers can be speaking the truth. (Cohen 1999, p. 57)

Mary and John are at the L.A. airport contemplating taking a certain
ﬂight to New York. They want to know whether the ﬂight has a layover
in Chicago. They overhear someone ask a passenger Smith if he knows
whether the ﬂight stops in Chicago. Smith looks at the ﬂight itiner-
ary he got from the travel agent and responds, ‘Yes I know – it does
stop in Chicago.’ It turns out that Mary and John have a very impor-
tant business contact to make at the Chicago airport. Mary says, ‘How
reliable is that itinerary? It could contain a misprint. They could have
changed the schedule at the last minute.’ Mary and John agree that
Smith doesn’t really know that the plane will stop in Chicago. They
decide to check with the airline agent. . . . neither standard is simply
correct or simply incorrect. Rather, context determines which stan-

From Moderate to Radical Contextualism

3 Not all the contributors to the epistemic contextualism debate agree about what they are
claiming when they say that knowledge attributions are context sensitive. Some suggest that
‘know’ is a kind of indexical (Cohen 1988); others that the expression is vague and that context
effects different precisiﬁcations (Heller 1999); others are even less committal and rest with
claims about what counts as true knowledge attributions depends on context (DeRose 1995),
or on what possibilities get ignored (Lewis 1996). We intend our objections to epistemic con-
textualism in this book to apply to all such views, and so, will not focus on the details of any
particular account.

dard is correct. Since the standards for knowledge ascriptions can vary
across context, each claim, Smith’s as well as Mary and John’s, can be
correct in the context in which it was made. When Smith says ‘I know
. . . ’, what he says is true given the weaker standard operating in that
context. When Mary and John say ‘Smith does not know . . . ’, what
they say is true given the stricter standard operating in their context.
And there is no context independent correct standard. (Cohen 1999,
pp. 58–9)

Again we see that it’s the truth values (or truth conditions; Cohen 1991,
p. 23) of utterances of knowledge ascriptions that are alleged to shift from
one context of use to another; or what’s said by these utterances (Cohen
1999, p. 57).

Moral Attributions

Context shifting also is invoked in ethics, for example, in talking about
‘good.’ Unger speaks about contextual variability in judgments about
whether something is permissible:

In many cases, the truth value (or the acceptability) of a judgment
about whether a person’s behavior is morally permissible depends on
the context in which the judgment is made. (Unger 1995, p. 2)

According to Unger’s intuitions, the truth values of moral judgments can
shift from context to context.

Dreier agrees with him when he writes:

For one thing, the content of a moral claim or belief is, on my view,
relative to a context. For another (and this is really just a consequence
of the ﬁrst) two people in different contexts may utter ‘x is good’ and
‘x is not good’ and both speak truly. (Dreier 1990, p. 7)

And more of the same, he writes:

speaker relativism is the theory that the content of (what is expressed
by) a sentence containing a moral term varies with (is a function of)
the context in which it is used. (Dreier 1990, p. 6)

In this latter quotation, Dreier is talking about what’s expressed varying
from context to context.

Exegesis: The Methodology of Contextualism

Weather Reports

According to numerous authors, distinct utterances of weather reports
vary, e.g., in what they say, as in:

Fred hears Mary say [‘It’s raining’]; he doesn’t know whether she is
talking about the location where they are; or some other location –
perhaps the location of the person to whom she is talking on the
phone. So, in a sense, he doesn’t know what she has said. (Perry 1998,
p. 7)

[‘It’s raining’] is used to say different things on different occasions of utter-
ance. A speaker S who utters [‘It’s raining’] says that it is raining where
S is (or at some other contextually salient location) at the time of
utterance (or at some other contextually salient time). (Pagin forth-
coming, p. 3; emphasis our own)

Other Sorts of Expressions

In addition to his contextualist commitment about quantiﬁer expressions
and counterfactual conditionals, Lewis advocates contextualism for a
rather wide range of expressions. For example, for words about geometri-
cal shape, he writes:

An adequate grammar must tell us that truth-in-English depends not
only on what words are said and on the facts, but also on features of
the situation in which the words are said. . . . If the words are ‘France
is hexagonal’ of course the shape of France matters but so do the
aspects of previous discourse that raise or lower the standards of pre-
cision. Truth in English has been achieved if the last thing said before
was ‘Italy is sort of boot shaped’ but not if the last thing said before
was ‘Shapes in geometry are ever so much simpler than shapes in
geography.’ (Lewis 1998, p. 24; see also Austin 1962, p. 143)

Here Lewis is appealing to intuitions about context shifting with regard to
shifts in the truth values of distinct utterances of ‘France is hexagonal.’ It’s
his intuition that an utterance of this sentence can shift in truth value from
true to false (with no physical facts changing) simply because of what other
sentences were uttered before it.

Along the same lines, Bezuidenhout writes:

Suppose that George has a paper route, which he covers every day on
his bicycle. As he rides past a customer’s house, he tosses their paper

From Moderate to Radical Contextualism

towards their house, aiming for their porch. Sometimes he is suc-
cessful but often he is not. In such a context, when a speaker utters (11)
[(11) ‘George managed to porch the newspaper yesterday’] she will be
understood to have said that George was successful yesterday in tossing
the newspaper onto the porch. On the other hand, suppose that it is
George’s job to bundle up each week’s newspapers and put the bundle
out on the porch, where someone from the recycling company will
pick it up. George isn’t very reliable, and some weeks he forgets to do
his job. In this context when a speaker utters (11) she will be understood to
have said that George managed to remember to put the newspaper
bundle out on the porch for recycling yesterday. (Bezuidenhout 2002,
p. 115; emphasis our own)

It is Bezuidenhout’s intuition that what gets said by distinct utterances of
(11) can shift from context to context. Based on this intuition she infers
the contextualist thesis that (11) is itself context sensitive.

Travis also infers contextualism based on intuitions about what’s being

said varying with distinct utterances of the same unambiguous sentence.

Consider the English ‘Ice ﬂoats’ . . . Now suppose, as may be, that ice
sinks in certain substances – glycerine, perhaps, or mineral oil or
ethanol. . . . Many typical speakings of ‘Ice ﬂoats’ rightly understood, are
not shown false by such things, since so understood, they do not say things to
be any way things are not if ice so behaves. For some speakings, though,
some or all of the above does matter. Sam and Pia, e.g., may be won-
dering what will happen if they drop an ice cube in the bowl of glyc-
erine before them. ‘Oh, it will just bob around’ Sam assures Pia, ‘After
all, ice ﬂoats’. If the cube sinks, then what Sam said in ‘Ice ﬂoats’ is
false. . . . So there is more than one thing to be said in saying ‘Ice ﬂoats’
where those words mean what they do mean in English; more than one
thing, that is, each of which is what sometimes would be said in so
speaking . . . (Travis 1994, p. 172; emphasis our own)

Radical Contextualism

So far we have been assuming that contextualism in its various forms is
limited to a small class of words that provoke contextualist intuitions of
various sorts, e.g., about the uses of quantiﬁers, counterfactual condition-
als, comparative adjectives, psychological, epistemic, and moral attribu-
tions. Some authors, however, set no such limits on intuitions about
context shifting. These authors believe that every single expression in every
single sentence in the language is subject to context shifting (all read unre-

Exegesis: The Methodology of Contextualism

strictedly!). And they invoke the exact same range of intuitions to defend
their widespread contextualism. This generalization of the argument was
central to much ordinary language philosophy in the twentieth century. It
is succinctly summarized by Austin as follows:

If you just take a bunch of sentences . . . impeccably formulated in
some language or other, there can be no question of sorting them out
into those that are true and those that are false; for . . . 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 circumstances
in which it is uttered. Sentences are not as such either true or false.
(Austin 1962, pp. 110–11)

Versions of the same view are found in Travis and Searle (repeated here from
Chapter 1):

What words mean plays a role in ﬁxing when they would be true; but
not an exhaustive one. Meaning leaves room for variation in truth
conditions from one speaking to another. (Travis 1996, p. 451)

The literal meaning of a sentence only determines a set of truth
conditions given a set of background practices and assumptions.
Relative to one set of practices and assumptions, a sentence may
determine one set of truth conditions; relative to another set of
practices and assumptions, another set; and if some sets of assump-
tions and practices are given, the literal meaning of a sentence may
not determine a deﬁnite set of truth conditions at all. (Searle 1980, p.
227)

Other prominent supporters of this view include Relevance theorists
(such as Dan Sperber, Deirdre Wilson, and Robyn Carston) and François
Recanati. Recanati, for example, says:

Contrary to what formal semanticists tend to assume, the (intuitive)
truth-conditions of our utterances are not compositionally deter-
mined by the meanings of words and their syntactic arrangement, in
a strict bottom-up manner. They are shaped by contextual expecta-
tions and world-knowledge to a very large extent. That is true of all

From Moderate to Radical Contextualism

utterances, however ‘literal’ they are (in the ordinary sense). (Recanati
2004, p. 92)

It is not surprising, of course, that these kinds of appeals to intuitions
about contextual variability should be at the center of discussion of context
sensitivity. Context sensitivity is, after all, variability between contexts. In
order to establish such variability one has to think about and compare what
happens to the same sentence in different contexts of utterance. What is
not equally obvious is:

(a) What the relevant variability is (what is said, claimed, expressed,

or what have you).

(b) What kinds of comparisons constitute solid semantic evidence.

(a) and (b) are discussed further in Chapters 7–9.

We now turn to the second, closely related form of argument for

contextualism – Incompleteness.

Incompleteness Arguments

Context Shifting Arguments and what we are calling Incompleteness Argu-
ments are not unrelated. If a sentence really is context sensitive, not only
may it shift in what is said or expressed by an utterance of it, it also makes
no sense to ask what it says, or expresses, independent of context. For
example, it makes no sense to ask of the bona ﬁde context sensitive sen-
tence ‘I am American’ whether it is true or false, nor does it make any sense
to ask of it whether it says or expresses anything independent of a (felici-
tous) use.

In the case of incompleteness (unlike that of context shifting) the intui-

tions invoked are ones in which a speaker is called upon to ask whether she
thinks a sentence says or semantically expresses anything, or has conditions
of truth, and so, a truth value, independent of any context. For example,
Taylor writes about sentence (3)

(3) It’s raining

that it

is missing no syntactically mandatory sentential constituent,
nonetheless, it is semantically incomplete. The semantic incompleteness
is manifest to us as a felt inability to evaluate the truth value of an

Exegesis: The Methodology of Contextualism

utterance of (3) in the absence of a contextually provided location
(or range of locations). This felt need for a contextually provided
location has its source, I claim, in our tacit cognition of the
syntactically unexpressed argument place of the verb ‘to rain’. (Taylor
2001, p. 53)

So in this case, (3) is claimed to be context sensitive because unless context
provides a location the sentence is felt to lack a truth value. Perry concurs
when he writes that:

in order to assign a truth value to my son’s statement [3] . . . I needed
a place. (Perry 1993, p. 206)

These claims about incompleteness, or as it is sometimes called ‘semantic
underdetermination,’ have been advocated for a range of sentences. Bach
writes about sentence (1):

(1) Steel isn’t strong enough

that

(1), though syntactically well formed, [is] semantically or conceptu-
ally incomplete, in the sense that something must be added to the
sentence for it to express a complete and determinate proposition. With (1)
we need to know strong enough for what (it does not express the weak
proposition that steel is strong enough for something or other) . . .
(Bach 1994b, p. 269)

The idea is that (1) is (semantically) incomplete – as Bach says, it does not
express a complete and determinate proposition. His ‘argument’ for the
conclusion is that only after it is speciﬁed ‘strong enough for what’ can a
determinate proposition be expressed. A sentence like (1),

even after disambiguation and reference ﬁxing, does not by virtue of
linguistic meaning express a complete proposition. When a sentence
is in this way semantically underdeterminate, understanding its utter-
ance requires a process of completion to produce a full proposition.
(Bach 1994a, p. 125)

Context is supposed to supply this information. The speaker ‘intends the
hearer to read something into the utterance, to regard it as if it contained
certain conceptual materials that are not in fact there’ (Bach 1994a, p. 126).

Sperber and Wilson, much like Bach, proclaim about sentence (20),

From Moderate to Radical Contextualism

(20) Peter’s bat is gray,

that

‘Peter’s bat’ might refer to the bat owned by Peter, the bat chosen by
Peter, the bat killed by Peter . . . and so on indeﬁnitely. . . . It seems . . .
that the semantic interpretation of a sentence with a genitive from
which ambiguities and referential indeterminacies have been elimi-
nated is still something less than fully propositional. Contextual
information is needed to resolve what should be seen as the seman-
tic incompleteness, rather than the ambiguity, of the genitive.
(Sperber and Wilson 1986, p. 188)

According to Sperber and Wilson, it ‘seems’ that sentences with genitive
(possessive) constructions fail to express any proposition whatsoever unless
context provides the required information.

Bezuidenhout concurs:

Let us take an example of a sentence that most would agree does
involve some incompleteness. Suppose for instance I utter the sen-
tence ‘There is no beer left.’ The quantiﬁer phrase ‘no beer’ is incom-
plete and in context it must be completed, either by restricting the
domain or by adding ellipsed material (depending on your favorite
view of the matter). Suppose that in context I am talking about what
beer is left in my refrigerator. Even so my utterance is open to multiple
possible understandings depending on what else is assumed in the
context. For instance, if I am having a party at my house, I might utter
the sentence in question trying to convey that there is no beer left in
the refrigerator for my guests to drink . . . But the context could be
rather different. Several bottles of beer have exploded in my refrig-
erator, spraying the inside of the refrigerator with beer. I have
been mopping up puddles of beer. When my husband asks me how
things are going I reply ‘There is no beer left.’ He understands me to say
that there are no more puddles of beer inside the refrigerator.
(Bezuidenhout 2002, pp. 112–13)

Her argument is that ‘most would agree’ that the sentence ‘There is no beer
left’ does not express a proposition, or has truth conditions, out of context.
We’re supposed to see that, note, because we recognize that in different con-
texts we recognize different propositions being expressed by the same sen-
tence. In this regard we see how context shifting and incompleteness go
hand in hand.

Exegesis: The Methodology of Contextualism

Bach extends the incompleteness charge to propositional attitude ascrip-

tions. He writes:

sentences used to make the belief reports, though semantically equiv-
alent, are also semantically incomplete. That is, they do not express com-
plete propositions, and to that extent they are like such sentences as

(5) Fred is ready.

and

(6) Jerry has ﬁnished.

Though syntactically well-formed (compare (6) with the virtually syn-
onymous but ungrammatical ‘Jerry has completed’), these sentences
are semantically incomplete because of a missing argument . . . Like
words such as ‘big’ and ‘short,’ a belief-predicate does not have a
context-independent condition of satisfaction, so that a sentence con-
taining it does not have a context-independent truth condition. A
belief-predicate does not express, independently of context, a unique
belief-property. So, for example, there is no unique property of believ-
ing that Batman is a wimp. (Bach 1997, p. 228)

Why, for example, are sentences (5) and (6) supposed to be (semantically)
incomplete? According to Bach, ‘these sentences are semantically incom-
plete because of a missing argument.’ What’s the argument that they lack
an argument?

Bach clearly thinks that the Incompleteness Argument can be extended

indeﬁnitely for belief reports. He writes:

Consider the following variation on the original version of the
Paderewski case. Suppose that Peter hears a recording of Paderewski
playing Rachmaninov in Carnegie Hall. Peter likes what he hears.
Then Peter hears a recording of Paderewski playing with a jazz combo
at the Apollo Theatre. This time he hates what he hears. It is clear to
us that Peter does not realize he has heard the same pianist twice. But
here it won’t do any good to say that Peter disbelieves that Paderewski
the pianist had musical talent, because we could also have truly said
that he believes that Paderewski the pianist had musical talent. We
could say that Peter disbelieves that Paderewski the jazz pianist had
musical talent and say too that he believes that Paderewski the classi-
cal pianist had musical talent. But this ploy won’t ultimately work
either. Suppose Peter hears a recording of an atrocious performance
of Paderewski playing Mozart. It is clear to us that Peter does not
realize that he has heard the same pianist a second time. We could say
that Peter disbelieves that Paderewski the classical pianist had musical
talent, but this would not distinguish what he disbelieves from what

From Moderate to Radical Contextualism

he believes. We would need to say that Peter disbelieves that
Paderewski the classical pianist playing Mozart had musical talent,
and that Peter believes that Paderewski the classical pianist playing
Rachmaninov had musical talent. Well, you get the idea. (Bach 1997,
pp. 230–1)

What Bach is intending to establish here is that the context sensitivity of
belief reports cannot be eliminated by inserting additional material into the
‘that’-clause. The problem is not one of insufﬁcient detail.

Bach is claim-

ing that you can add all the detail you want but the problem doesn’t go
away. No matter how much material is inserted into it, a ‘that’-clause does
not determine belief content but merely narrows it down. So, it seems that
no belief report is inherently capable of specifying a belief fully.

Here’s an example of an author who runs CSA and Incompleteness

together in the same example:

Typically, the proposition that a sentence expresses depends not only
on the meaning of the constituent words and their grammatical com-
position but also on the context in which the sentence is uttered. . . .

Domain of discourse: ‘Everyone is present.’ If the domain of dis-

course is students still enrolled in the course, then the proposition
expressed will be the proposition that every student still enrolled in
the course is present. If the domain of discourse is students who have
been attending recently, then the proposition expressed will be the
proposition that every student who has been attending recently is
present . . . Suppose that a teacher enters a classroom, looks around
and declares, ‘Everyone is present.’ Taken out of context, this sentence
does not express any particular proposition, because, taken out of
context, there is no particular domain of discourse relative to which
we may interpret ‘everyone’ . . . Nonetheless, the sentence, as a sen-
tence of English, carries a certain potential for expressing proposi-
tions and this potential is, in one sense, its meaning. . . . So sentences
may fail to express a proposition all by themselves, but may nonethe-
less express a proposition in a context. When a speaker utters a sen-
tence in some context, we may describe the proposition that the
sentence expresses in that context as what is said, or what the speaker
says. (Gaulker 2002, pp. 11–13)

Exegesis: The Methodology of Contextualism

4 Bach (and other Moderate Contextualists) doesn’t feel this way about all cases of incom-
pleteness. About the sentence ‘Steel isn’t strong enough’ he writes: ‘the speaker could have
made the additional conceptual material explicit by including the corresponding lexical mate-
rial in his utterance’ (Bach 1994a, p. 127). This will be important to us when we turn to crit-
icism of Incompleteness Arguments.

Carston (2002) is in large part a defense of the most general version of

this kind of incompleteness claim. She summarizes her views on this issue
as follows:

Underdeterminacy is an essential feature of the relation between lin-
guistic expression and the propositions (thoughts) they are used to
express; generally, for any given proposition/thought, there is no sen-
tence which fully encodes it . . . Underdeterminacy is universal and no
sentence ever fully encodes the thought or proposition it is used to
express. (Carston 2002, p. 29)

Only limitations of space and time prevent us from pursuing and elabo-

rating further examples. We hope, though, that the above is more than suf-
ﬁcient to justify our claim that Context Shifting Arguments and
Incompleteness Arguments are at the center of the contemporary debate
about contextualism in philosophy and linguistics.

From Moderate to Radical Contextualism

C H A P T E R 3

The Instability of Context

Shifting Arguments

Moderate Contextualism (MC) is not a stable position. A consistent (and
sufﬁciently imaginative) Moderate Contextualist must endorse Radical
Contextualism (RC). The kind of evidence that supports MC leads directly
to RC. The kinds of arguments used to support MC lead directly to RC.
Someone who starts down the path of contextualism, but wants to stop
short of RC, can do so only arbitrarily. This arbitrariness, we will argue, is
a form of inconsistency. We aim to establish this in the next three chapters.
Here’s the central claim of this chapter:

If Context Shifting Arguments of the kind described in Chapter 2

sufﬁce to show that MC is true (i.e., if they sufﬁce to show that there
is a context sensitive expression or locution e not in the Basic Set of
context sensitive expressions), then RC follows.

We cannot emphasize enough the conditional nature of this claim. Don’t
forget, we do not think the arguments for MC are any good. Indeed, we’re
going to argue MC is false. However, for the sake of argument, we will for
the time being place our convictions to one side and run an entirely inter-
nal argument against MC. We will show that, given the standards of evi-
dence implicit in the arguments adduced by Moderate Contextualists, RC
follows. This is signiﬁcant, since most Moderate Contextualists do not
endorse RC.

We pointed out in the last two chapters that a CSA involves a complex

imaginative activity. It requires conjuring up interesting scenarios, making
these scenarios vivid (to yourself and to your audience), and then in some
way empathizing in the most literal fashion with the participants in these

1 See Chapter 7 for a very special kind of CSA we like.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

imagined scenarios: it requires, so to speak, placing yourself imaginatively
into the shoes of a participant in these alternative scenarios. In that light,
here’s another way to articulate the central claim of this part: Moderate Con-
textualists are unimaginative Radical Contextualists.

Context Shifting

In the last chapter we documented how Moderate Contextualists argue for
the context sensitivity of some expression e (not in the Basic Set) by solic-
iting intuitions that what is said, stated, asserted, or claimed by an utter-
ance u of an (unambiguous) sentence S containing e need not be the same
as what is said, etc., by another utterance u

¢ of S (and that this difference is

not due to any other context sensitive expression in S).

These intuitions and feelings about the context shiftiness of various

kinds of content (what’s said, asserted, and expressed) are triggered by
imaginatively varying the context of utterance for S, i.e., the sentence which
contains the alleged context sensitive e. The claim we will establish ﬁrst is:

(GEN) With sufﬁcient ingenuity, a CSA can be provided for any sen-

tence whatsoever, and consequently, for any expression.

After presenting evidence in favor of GEN and discussing various possible
Moderate Contextualist responses, in Chapter 4 we will offer a diagnosis of
what’s gone wrong with MC. We will suggest that the widespread abuse of
CSAs in current philosophical debates is based on a mistaken (or ill-
advised) assumption about the relationship between semantic content and
speech act content. In short:

Diagnosis. The ubiquitous abuse of Context Shifting Arguments in
current philosophical debates (not just in semantics, but in all areas
of philosophy) is based on a confused assumption about the rela-
tionship between semantic content and speech act content.

If this assumption were true, then the kinds of CSAs appealed to by Mod-
erate Contextualists would be sound. That leaves proponents of such argu-
ments with a choice: either give up the assumption about the relationship
between semantic content and speech act content, or endorse RC.

Preliminaries

Since the arguments we are running in this chapter and Chapter 5 are inter-
nal, namely, arguments of the form: ‘If this MC argument is any good, then

From Moderate to Radical Contextualism

RC follows,’ we need a paradigmatic MC argument to compare our new
CSAs against. We will use a CSA involving quantiﬁer domain restriction.
Such CSAs, as even a casual glance at the literature will conﬁrm, are para-
digmatic of the CSAs that entice Moderate Contextualists. Consider
sentence (1):

(1) There are no French girls.

The Moderate Contextualist’s argument for the context sensitivity of (1)
can, for heuristic purposes, be presented in three stages:

S1. The Moderate Contextualist asks us to imagine a particular utter-

ance u of (1), say, one in which someone is searching a particular classroom,
Room 401, where it is shared knowledge that his aim is to ﬁnd a French
girl. Properly elaborated, it is not difﬁcult to trigger the following two kinds
of connected intuitions.

First, the intuition that the speaker said something true as long as there

are no French girls in Room 401. What that allegedly shows is that the truth
conditions for u do not require that the universe be devoid of French girls.
That intuition can be bolstered by noticing how natural it would be to
respond to u by ‘Right,’ or ‘That’s true,’ or ‘You’re right,’ if you think there
are no French girls in Room 401. In other words, what u expresses could be
true even though the universe as a whole includes many French girls. The
proposition expressed by u has a restricted domain of quantiﬁcation. More
speciﬁcally, u is true just in case there are no French girls in Room 401. By
uttering u, the speaker only said, asserted, claimed, and stated that there are
no French girls to be found in Room 401.

S2. By going on to describe another context of utterance, the Moderate

Contextualist can trigger another intuition about a distinct utterance u

¢ of

(1) which has a domain restriction for its quantiﬁer other than that deter-
mined by u. If u

¢ is uttered by the same speaker, but one who is now scurry-

ing about in New York City desperately seeking French girls, the Moderate
Contextualist predicts that the general intuition will be that the speaker
asserted, claimed, etc. that there are no French girls in New York City, and that u
is true if there were no French girls in New York City (even though Room
401 situated in New Jersey may be crowded with French girls).

Again, there are several ways to see this. Notice for example that if you

believe there are no French girls in New York City, it’s very natural to
respond to u

¢ by uttering ‘Right,’ or ‘That’s true,’ or ‘You’re right.’

S3. According to the Moderate Contextualist, these two intuitions

provide strong evidence that the quantiﬁer in (1) is context sensitive. More
generally, that sentences containing quantiﬁers exhibit context sensitivity
with respect to domain restrictions.

Instability of Context Shifting Arguments

Our immediate aim is to convince you that if the argument sketched in

S1–S3 impresses you, then you ought to be equally impressed by analogous
arguments that involve exacting the same sort of intuitions for utterances
of any English sentence whatsoever.

Our claim is an empirical one. It’s a claim about what kinds of intuitions

can be triggered by thinking about language use under various circum-
stances. The only way to establish it (or its negation, for that matter) is
through an empirical investigation of the kinds of intuitions (or feelings)
that can be triggered by telling the kinds of imaginative stories that Mod-
erate Contextualists are so fond of.

As far as we can tell, the only way to conduct such an investigation is to

pick a large sample of sentences; try to construct CSA for them; and see
what intuitions these stories trigger. We need to make sure that the sample
is representative and that the sentences in the sample do not lend them-
selves to ‘local’ contextualist solutions.

This component about local contextualist solutions is important. For

any case that a Radical Contextualist might devise, a Moderate Contex-
tualist might try to accommodate it by adding a new expression to the Basic
Set of context sensitive expressions. The strategy for the Radical Contex-
tualist, therefore, is to choose a wide enough and sufﬁciently disparate
range of cases so that localized solutions will seem increasingly artiﬁcial
and desperate.

One ﬁnal but crucial preliminary: We have to ensure that the relevant

features of the selected samples cannot be explained away by appeal to
factors such as:

䊉

ambiguity

䊉

syntactic ellipsis

䊉

polysemy

䊉

nonliterality

䊉

vagueness.

Since our goal is to show that certain kinds of intuitions can be triggered
for an arbitrary English sentence, for these intuitions to have the relevant
signiﬁcance (i.e., to provide evidence of semantic context sensitivity), it is
crucial that they are not triggered by irrelevant factors. If, for example, the
intuition that sentences containing e change truth conditions across con-
texts of utterance can be explained by e’s being ambiguous, polysemous,
used metaphorically (in one case but not the other), etc., then our examples
would be irrelevant. They could be dismissed by Moderate Contextualists
as disanalogous to their own favorite cases.

The range of cases we’ve chosen to present, we hope, is obviously not

explicable by any of these irrelevant factors. As far as we know, no one has
suggested that the examples we will now utilize can be so explained.

From Moderate to Radical Contextualism

In Defense of GEN

Our arguments for GEN are not particularly original. Radical Contex-
tualists such as Searle, Sperber, Wilson, Travis, Moravcsik, Recanati,
Carston, and Bezuidenhout have done excellent jobs of showing how the
sorts of intuition Moderate Contextualists invoke can be generated for arbi-
trary English sentences. First, we’ll rehearse some of their efforts. Then we
will mimic their arguments with respect to novel cases.

Some of our favorite examples from Radical Contextualists

Imagine two scenarios in which (2) is uttered, with its meaning ﬁxed. (The
example is adapted from both Travis and Searle.)

(2) Smith weighs 80 kg.

In one scenario the intuition is supposed to be that an utterance of (2) is
true, while in the other the equally powerful intuition is supposed to be
that its utterance is false:

Scenario One. Smith has been dieting for the last eight weeks. He steps
on the scale one morning, naked, before breakfast (but after having
gone to the bathroom), and it registers 80 kg. A friend at work who
wants to let Smith’s co-workers in on his achievement can use (2) to
say something true. Notice it doesn’t matter at all that Smith is, at
that time, dressed, wearing a heavy overcoat, and has just consumed
an enormous lunch.

Scenario Two. Smith is exactly as in Scenario One. However, the
speaker’s circumstances (and purposes) have changed. At the time of
this utterance of (2) (suppose the same time as in Scenario One),
Smith is about to enter an elevator with a capacity of no more than
an extra 80 kg. An utterance of (2) in these circumstances could be
both fatal and false. Note that what the scale registers when Smith is
naked in the morning is in this context irrelevant.

Our question is whether the kind of intuitive evidence that allegedly

showed (1) to be context sensitive can be triggered with respect to (2). Ask
yourself: Did the speaker in Scenario One succeed in saying something
true? Clearly, says the Radical Contextualist, the answer is ‘yes.’ Did the
speaker in Scenario Two succeed in saying something true? ‘No,’ says the
Radical Contextualist. So, we have two imagined contexts in which two
(simultaneous) utterances of the same sentence type with the same

Instability of Context Shifting Arguments

meaning, whose referring terms are assigned the same referents, are alleged
to differ intuitively in their truth conditions due to ‘nonsemantic’ differ-
ences surrounding their respective contexts of utterance. If it follows from
like intuitions that distinct utterances of (1) disagree in truth value, why
doesn’t it also follow from the intuitions surrounding Scenarios One and Two that the
truth conditions of utterances of (2) are generated in individual contexts of utterance?

Here’s an example from Bezuidenhout:

We’re at a county fair picking through a barrel of assorted apples. My
son says ‘Here’s a red one,’ and what he says is true if the apple is
indeed red. But what counts as being red in this context? For apples,
being red generally means having a red skin, which is different from
what we normally mean by calling a watermelon, or a leaf, or a star,
or hair, red. But even when it is an apple that is in question, other
understandings of what it is to call it ‘red’ are possible, given suitable
circumstances. For instance, suppose now that we’re sorting through
a barrel of apples to ﬁnd those that have been afﬂicted with a horri-
ble fungal disease. This fungus grows out from the core and stains
the ﬂesh of the apple red. My son slices each apple open and puts the
good ones in a cooking pot. The bad one he hands to me. Cutting
open an apple he remarks: ‘Here’s a red one.’ What he says is true if
the apple has red ﬂesh, even if it also happens to be a Granny Smith
apple. (Bezuidenhout 2002, p. 107)

Again, distinct utterances of a single sentence ‘Here’s a red one’ are alleged
to disagree in truth value (ignoring the contribution of ‘here’). What could
possibly render this CSA for RC unsound while leaving perfectly intact
analogous cases of CSA for MC?

By now, you’ve probably picked up sufﬁciently well on the methodology

to be able almost effortlessly to manufacture your own CSAs for any expres-
sion whatsoever; here are some of our own modest efforts.

Analogous cases

Take an arbitrary sentence, say, (3) (if you think it’s not arbitrary, choose
your own; try to stump GEN).

(3) John went to the gym.

We will now devise three distinct contexts of utterance for (3). Each, we
claim, triggers a distinct intuition about what that utterance of (3) says or
expresses, or its truth conditions. We’ll let you be the judge.

From Moderate to Radical Contextualism

Context of Utterance 1. We are engaged in a conversation about where
John takes his nightly walk. It’s common knowledge in our crowd that
his only form of exercise is walking, and that he does it only late at
night when the streets are empty and everything, including the gym,
is closed. Your utterance of (3) in this context expresses the claim that
John walked to the vicinity of the gym. If John did go to the vicinity
of the gym, then your utterance of (3) is true; if not, it’s false.

Context of Utterance 2. You are participating in a discussion about John’s
dieting and exercise routines. His trainer asks you to account for John’s
exercise today. It would be a lie for you to reply with an utterance of (3)
in this context if all John had done was walk to the vicinity of the gym.
That is to say, even if the exact same circumstances obtained vis-à-vis
John as in the Context of Utterance 1, your utterance of (3) in this
context is false. The truth of this utterance of (3) requires John to be
engaged in certain kinds of activities inside the gym.

Context of Utterance 3. You are at a meeting of a construction company
planning on putting down a new hardwood ﬂoor at a local gym. The
boss asks who went over to the gym to supervise the construction of a
bathroom in its basement. You reply with (3). Clearly, again, what you
said is not true if John simply walked to the vicinity of the gym. What
you intend to say is that John has gone over to the gym to supervise the
work on the bathroom. The sort of activity he engaged in while at the gym
is again intuitively part of the truth conditions of your utterance.

For our next example, consider (4) and two attending contexts of

utterance.

(4) Jill didn’t have ﬁsh for dinner.

Context of Utterance 1. Jill went to a restaurant, ordered salmon as her
main course, but never touched the salmon; she just ate the vegeta-
bles on her plate surrounding the ﬁsh. Suppose further that Jill came
down with food poisoning later that night after returning home. A
doctor, in an effort to ascertain the cause of her illness, enquires as to
what Jill ate at the restaurant earlier that night. He wants to know
what she consumed for dinner. In this context, an utterance u of (4)
would be true. In uttering u, the speaker has said something true.
Intuitively, he said, asserted, claimed, and stated that Jill didn’t
consume ﬁsh for dinner. Even more speciﬁcally, that she didn’t put
any ﬁsh into her mouth and ingest it.

Context of Utterance 2. Jill, as before, went to the same restaurant,
ordered the salmon, and again didn’t eat it, but did eat the vegetables.

Instability of Context Shifting Arguments

In this case a debate arises over the bill. Again, someone utters (4). In
this case, the utterance of (4) is intuitively false. It makes no difference
whatsoever what Jill ingested. What matters is what she ordered and
there is no doubt about her having had the ﬁsh for dinner. (We don’t
deny that a Moderate Contextualist doesn’t have all sorts of room to
stabilize his position, e.g., positing ambiguities; more on this below.)

For another example, consider sentence (5):

(5) Justine destroyed those shoes.

Context of Utterance 1. Suppose that in this context of utterance, the
conversational focus is on the aesthetic features of a pair of Marie’s
shoes. Justine has just spray painted Marie’s shoes bright yellow, and
poured oil on top of them. Getting dressed for a dinner Marie utters
(5). Did she say something true? Yes. Notice that an utterance of (5)
would express something true even though the shoes are perfectly
functional qua footwear. What was said in this case is that the aes-
thetic features of her shoes have been destroyed.

Context of Utterance 2. In this context, the conversational focus is on the
solidity and warmth of footwear, and its aesthetic features are of no
importance. We’re on an expedition in the jungle in Kenya, and some
of our fellow travelers’ shoes have fallen apart. Justine has spray
painted Marie’s shoes bright yellow so that they’ll be easier to identify
from a distance, and poured oil all over them to protect them from the
elements. Would an utterance of (5) in this context be true? No. In fact,
Justine has improved the shoes. She has made them more functional.
What would be said, in this context, by an utterance of (5) is that
Marie’s footwear has fallen apart, that they are no longer functional.

For a fourth example consider sentence (6) and its ensuing contexts of

utterance:

(6) That’s a dangerous dog.

Context of Utterance 1. Suppose an utterance u of (6) in this context is
used to say that the demonstrated dog bites people: u is true only if
the dog is aggressive and initiates acts that put people in danger.
Notice that u would not be true if the dog is, for example, kind and
loving, but unhealthy.

Context of Utterance 2. Suppose (6) is used in a different context to say
that the dog has a viral disease that can spread to humans: an utter-
ance u of (6) in this context doesn’t require the dog to exhibit any

From Moderate to Radical Contextualism

aggressive behavior whatsoever. u can be true even though the dog
never bit anyone; u can be true even though the dog is kind. What’s
said by u in this context is that being in the proximity of the dog can
have detrimental consequences.

Context of Utterance 3. Suppose a group of people carry on in the most
peculiar fashion. One of their strange practices is to toss dead dogs
at each other: some of these dogs are small and cause no harm upon
impact, but some are rather hefty, and being hit by one can cause
severe damage. An utterance u of (6) in this context could be true even
though the demonstrated dog is neither sick nor aggressive. u would
be true, just by virtue of the dead dog being very heavy and stiff, and
hence, potentially harmful should it land on anyone.

Our next example is a variant of an example of Travis’s.

(7) Justine is a philosopher.

Context of Utterance 1. We ﬁnd ourselves in a conversation about
Justine’s character. Some of us think Justine a rather unreﬂective
person with virtually no commitments to any position one way or the
other. In this context, an utterance u of (7) might be deemed false even
though Justine, as a matter of fact, let us suppose, is working as a pro-
fessional philosopher at some university or other.

Context of Utterance 2. A group of us are sitting around late at night at
an American Philosophical Association meeting gossiping about the
professions of our friends (and enemies). Here an utterance u of (7) is
a way of signaling that Justine is indeed a professional philosopher.
u could be true even though she lacks any philosophical character
whatsoever.

So far we have been appealing to examples involving relatively simple

sentences and the various contexts of utterance have differed in goals, moti-
vations, purposes, and shared assumptions. To a certain extent these kinds
of examples are artiﬁcial. There’s a kind of neatness about them not often
found in ordinary conversations. That’s an advantage because it makes it
easier to ensure that the desired intuitive effects aren’t triggered by irrele-
vant factors (such as ellipsis, polysemy, etc.). We’ll end, however, with a more
complex case.

Imminent Threats and the ‘Bush Doctrine’

In an interview with Tim Russert on MSNBC’s Meet the Press, President
George Bush said the following:

Instability of Context Shifting Arguments

I don’t think America can stand by and hope for the best from a
madman, and I believe it is essential – I believe it is essential – that
when we see a threat, we deal with those threats before they become
imminent. It’s too late if they become imminent.

These remarks were, at the time, interpreted in the light of the known back-
ground facts that Bush was the president of the United States, that he had
started a war in Iraq, and that one of the reasons he did was because he
thought preventive wars were justiﬁed if the United States faced an immi-
nent threat. In that context, the utterance was interpreted as having sig-
niﬁcant foreign policy implications. In particular, it was interpreted to
mean that Bush no longer thought that a state being an ‘imminent threat’
was a necessary condition for a pre-emptive war. He seems to be saying that
all that’s needed to justify a pre-emptive war is evidence that a country
might become an imminent threat.

Of course, these background facts are not encoded in the sentences

uttered. Indeed, it is easy to imagine these sentences uttered in a completely
different context without the same implications or interpretations. Imagine
Bush uttering these sentences while working as a nurse in an asylum. The
conversation topic concerns how to treat aggressive patients. Some of the
nurses think they should be locked up, and some think they should be
sent out medicated. If Bush uttered these sentences in such a context,
it should be obvious that his utterance wouldn’t be interpreted as having
far-reaching implications for US foreign policy.

We hope these various cases sufﬁce to convince you that the intuitions

surrounding the MC cases are on no surer ground than those we have just
provoked. And so we hope we have convinced you of GEN.

Our conclusion from all of this data is that if you start down the slope

of MC, there’s no stopping short of RC. There’s nothing about the context
shiftiness of quantiﬁers, for example, that distinguishes them from any
other sort of expression. If you think the CSAs presented in favor of context
sensitivity are convincing, then you have no reason not to become a Radical
Contextualist.

Attempts to Stabilize MC

We’re expecting Moderate Contextualists to respond to our defense of GEN
in at least one of the following ways.

From Moderate to Radical Contextualism

First attempt to stabilize CSAs: Introspection

Moderate Contextualists might claim that the intuitions we triggered with
our various cases are different in feel and introspective quality from those trig-
gered, e.g., for domain restrictions on quantiﬁers. For example, a philoso-
pher might respond with the following introspective report: When she
imagines an utterance of ‘There are no French girls’ the feeling that some-
thing true is said (even though there are lots of French girls in the universe)
is qualitatively different from the feeling that something true is said by an
utterance of ‘John went to the gym’ (even though the person in question
just walked in the vicinity of the gym). There might be a qualitative differ-
ence in the introspective qualities or a difference in strength.

If you think this response sounds silly, we assure you that we have

encountered distinguished and intelligent philosophers (and linguists)
who have made exactly this response to our defense of GEN.

Our reply. This reply, we suppose, signiﬁes one of those points where philo-
sophical argumentation simply comes to a halt. We can only report that
intense introspection does not register any principled difference in feel
between the intuition that some utterance of ‘There are no French girls’
expresses the proposition that there are no French girls in Room 401 and the
intuition that an utterance of ‘John went to the gym’ expresses the
proposition that he walked to the vicinity of the gym. If you insist that when you
introspect, these two intuitions feel quite different to you, we have two
replies:

(a) Consider the possibility that your feeling is biased, that it’s gen-

erated by your conviction that MC is true: the intuitions that are
alleged to establish your theory have ended up being shaped by
your commitment to MC itself.

(b) You at least owe us an explanation for why we, and every nonbi-

ased person we have tried these cases on, fail to register any sig-
niﬁcant difference. If a Moderate Contextualist is serious about
basing her theory on this kind of data, then she at least owes us
an explanation for why these intuitions aren’t widely shared.

Second attempt to stabilize CSAs: Optimistic–energetic

search for localized solutions

An optimistic and energetic Moderate Contextualist will see these various
cases as providing a challenge, an opportunity to generate new publications

Instability of Context Shifting Arguments

and exhibit formal ingenuity. They’ll try to ﬁnd what we called above ‘local
ﬁxes.’ Their hope is this: Moderate Contextualists can accommodate all of
our cases and any other case we might throw at them simply by adding a
tad of indexicality here and a tad there, using any one of three strategies
available to Moderate Contextualists: surprising indexicals, hidden indexicals,
and unarticulated constituents.

Our reply. We chose these cases because, as far as we know, there are no local
ﬁxes for them. As it stands, no MC account of these cases is currently avail-
able. That, of course, makes the suggested solution highly speculative. We
don’t in fact see any point in even considering this sort of rejoinder until a
suggested local ﬁx, and a strategy for generating them for all of the cases,
are forthcoming. For any one such ﬁx we can imagine, we suspect a range
of objections we present in Chapter 6 to apply. But absent a speciﬁc sug-
gestion we cannot go into details.

Third attempt to stabilize CSAs: Strongly dismissive reaction:

These reports are false

The intuitions we are triggering are intuitions about what speakers say (or
assert, or claim . . . ). Our examples therefore assume that the indirect
reports in which we report on these intuitions (i.e., reports of the form ‘She
said that . . . ’) are true. We take our practice of indirect reporting at face
value and assume that the speakers have said what we have the reporters
saying that they have said. One might deny this. One way to do so would
be to treat these reports as merely appropriate or reasonable in the contexts
described, but not as true. Instead, some pragmatic story can be invoked
to account for why we occasionally ﬁnd false reports still reasonable or
appropriate.

Our reply. This reply, however, is not open to any of our opponents. Our
primary opponents use CSAs, i.e., they are philosophers who take it to be
signiﬁcant that what’s said by an utterance, say, for example, of ‘There are
no French girls,’ is, in some particular context, that there are no French girls
in Room 401. They assume that this is a true indirect report. It is this
assumption, we claim, that ultimately leads to MC (and RC). If it’s obvious
to them that the quantiﬁer case is an example of a true indirect report, by
what standard do they determine that our cases are false? How do they dis-
tinguish their favorite cases from our cases? It seems that the only way to
draw this distinction is not by appeal to intuitions about what people say,
but through some sort of theoretical processing for these intuitions. We,
however, know of no nonquestion-begging theoretical assumptions that

From Moderate to Radical Contextualism

would distinguish their favorite cases from our own. Hence, our opponents
are in no position to make this reply against us.

Fourth attempt to stabilize CSAs: These reports would be false if

modiﬁed by ‘strictly speaking,’ ‘literally,’ or ‘really’

Sometimes we encounter the response that the problematic reports, though
true, would be false if ‘said’ were modiﬁed by ‘literally’ or ‘strictly speaking’
(e.g., Bach 1994b). On this view, a special class of indirect reports of the
form ‘A literally (strictly) speaking said that p’ creates a connection between
indirect speech and semantic theory relevantly similar to the connection
we’ve been denying.

Our reply. As a claim about how the expressions ‘literally’ and/or ‘strictly
speaking’ function in our language, this reaction fails. Consider Jill’s
utterance of ‘I had ﬁsh’ in response to a waiter’s question during a discus-
sion of a restaurant bill. Our intuition is that she said that she ordered ﬁsh,
and that it makes no difference to the truth value of what she said what
she actually put in her mouth. But suppose someone asks ‘Did she really
say that?’ or ‘Is that literally what she said?’ or ‘Is that strictly speaking what
she said?’ There’s a perfectly natural sense in which the answer to all these
questions is a simple ‘yes.’ She really did say that. That’s literally what she
said. She said that, strictly speaking.

As far as we can determine, the only way to take this question so that its

answer is ‘no’ is as a question about direct quotation. So understood, the
only correct answer is ‘She said “I had ﬁsh.” ’ Of course, so understood this
question is irrelevant to our concerns.

We think this response generalizes, but we will leave the generalization

as an exercise for the reader.

In short, actual usage of the modiﬁers ‘literally,’ ‘really,’ and ‘strictly

speaking’ does not stabilize MC. If anything, they can be used to ask
whether an indirect quote is also a direct one. But that of course is irre-
levant to our concerns here.

Fifth attempt to stabilize CSAs: CSAs are necessary, but not sufﬁcient

Some Moderate Contextualists might be inclined to respond by saying that
the presence of intuitions of the kind triggered by CSAs is just a necessary
condition for saying that an expression (or sentence) exhibits semantic
context sensitivity. (Heal (1997) makes something like this response.) In
addition to this evidence something else is needed. What that something

Instability of Context Shifting Arguments

else might be will vary among different Moderate Contextualists. The two
options we think most likely to be defended are, ﬁrst, that in addition to a
CSA for S, it should be possible to present an Incompleteness Argument
for S. We respond to this in Chapter 5. The second most likely candidate is
that there should be some kind of syntactic evidence added to the CSA. In
Chapter 5 we’ll argue that Incompleteness Arguments are as unstable as
CSAs are, so that additional constraints will not prevent the Moderate Con-
textualist from slipping into RC. In Chapter 6 we’ll argue that the kind of
syntactic evidence typically appealed to in defense of MC is irrelevant with
respect to whether a sentence is or is not semantically context sensitive.

From Moderate to Radical Contextualism

C H A P T E R 4

Diagnosis: Why Context Shifting

Arguments are Misused

Here’s an interesting fact: Many sensible philosophers reject any associa-
tion with RC. They would consider any set of premises from which RC
follows to be a reductio of those premises. These are perfectly modest
philosophers, without radical revolutionary aspirations, i.e., with no
sympathies for RC. Nonetheless, they freely make use of CSA locally (e.g.,
for quantiﬁer domain restrictions, comparative adjectives, possessives,
the pleonastic ‘it’ of weather and temporal reports; and so on). It’s no exag-
geration to say they are hooked on Context Shifting Arguments. They
wouldn’t know how to underwrite an inference to semantic context sensi-
tivity without them. But CSAs, taken seriously, lead to RC. This raises an
interesting question: How did a collection of perfectly pleasant and sensi-
ble philosophers end up addicted to a form of argumentation that, when
thought through carefully, leads to a crazy view like RC? (Of course, we have
not yet established that RC is crazy, but we will in Chapters 7–9. So, for
now, just take our word for it.)

Assumption Underlying the Appeal to CSAs

There’s an assumption underlying the use of CSAs that in its vaguest and
weakest form can be formulated as what we have labeled without prejudice
the Mistaken Assumption (MA, for short):

(MA) A theory of semantic content is adequate just in case it

accounts for all or most of the intuitions speakers have about
speech act content, i.e., intuitions about what speakers say,
assert, claim, and state by uttering sentences.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

Several variations on MA can be identiﬁed in the literature. We’ll focus on
MA*:

(MA*) If CSAs of the kind presented so far trigger the intuition that

proposition p is said, claimed, stated, or asserted by an utter-
ance u of sentence S in language L, then an adequate seman-
tic theory for L should assign p as the semantic content of u.

Why think that MA* is made (often tacitly) by those who freely use CSAs,
and more generally, by all Moderate (and Radical) Contextualists? The most
obvious answer is: It’s the only way we can make sense of their extensive
(mis)use of CSAs. If you did not believe in some version of MA, why would
you care about the intuitions that speakers have that an utterance of ‘There
are no French girls’ can be used to say (assert, claim, state) that there are
no French girls in Room 401? Why would a semanticist hold that piece of
information relevant? Why, more generally, would a semanticist think any
of the intuitions appealed to in CSAs are relevant?

But there is no need for us to extrapolate only from the use of CSAs to

ﬁnd a commitment to MA; it’s found explicitly in a surprisingly diverse
group of philosophers. As an illustration, we will show how central MA is
in the works of three philosophers as different in philosophical outlook
and commitments as Donald Davidson, David Kaplan, and Charles Travis.

Kaplan and MA

According to Kaplan, a semantic theory must be anchored in speaker intu-
itions about ‘what-was-said.’ Here are two representative passages from his
justly famous ‘Demonstratives’:

What is said in using a given indexical in different contexts may be
different. Thus if I say, today, I was insulted yesterday and you utter
the same words tomorrow, what is said is different . . . Let’s call this
ﬁrst kind of meaning – what is said – content. (1989a, p. 500)

If I may wax metaphysical in order to ﬁx an image, let us think of the
vehicles of evaluation – the what-was-said in a given context – as
propositions. (1989a, p. 494)

In his ‘Afterthoughts,’ Kaplan writes:

The idea of content – the what-is-said on a particular occasion – is
central to my account. (1989b, p. 568)

From Moderate to Radical Contextualism

An appeal to what is said by an utterance of a sentence is not unimportant.
Kaplan, and in fact all contemporary intensionalists, develop elaborate
technical frameworks within which they attempt to capture the notion of
what is said. Were this notion entirely theoretical, it would be unclear
exactly which phenomenon the technical apparatus was attempting to
explicate or clarify or even describe.

Davidson and MA

Davidson’s commitment to MA is found not so much in what he says about
the aims of semantics, but in what he says about indirect speech. Accord-
ing to Davidson, truth conditions for (1) are provided by (2) (Davidson
1968):

(1) Galileo said that the earth moves.
(2) (

$u)(Ugu & SSu,that): [The earth moves.]

The quantiﬁer ranges over utterances; ‘Ugu’ holds just in case Galileo
uttered u; ‘SS(u,that)’ holds when u and the utterance demonstrated by the
occurrence of ‘that’ samesay each other, where the demonstratum of ‘that’ is
an ensuing utterance of the bracketed sentence.

What few remarks Davidson makes about samesaying can be (and

usually are) read as endorsing MA. Passages like the following encourage
commentators to infer Davidson imposes MA as an adequacy condition on
semantic theory.

The ‘that’ refers to the second and the ﬁrst utterance is true iff an
utterance of Galileo’s was the same in content (‘translates’) the utterance
to which that ‘that’ refers. (1976, p. 177; our own emphasis)

Platts, taking his cue from these passages, writes, ‘indirect discourse
requires that the reporter reproduce the meaning of the original utterance:
the content-sentence employed by the reporter should be a correct transla-
tion of the original speaker’s utterance’ (Platts 1979, p. 126). Likewise,
Burge writes, ‘the point of indirect discourse might be fairly taken to be to
introduce and produce an utterance that gives the content of the original
speaker’s utterance’ (Burge 1986, p. 196). More explicitly, Larson and
Ludlow maintain ‘it is arguable that one of the main charges of any seman-
tic theory is to give . . . an account of the content of a given utterance:
semantic theories should characterize what is said in uttering a given sen-
tence S . . . In a truth conditional theory . . . content is ostensibly captured
through the truth-conditions that are assigned’ (Larson and Ludlow 1993,
p. 334). And, lastly, McDowell writes:

Why Context Shifting Arguments are Misused

The basis of the truth-conditional conception of meaning, as I see it,
is the following thought: to specify what would be asserted [i.e., said],
in the assertoric utterance of a sentence apt for such use, is to specify
a condition under which the sentence (as thus uttered) would be true.
(McDowell 1987, p. 60)

Travis and MA

Travis is a sort of hero to us. His view is completely wrong, but of all those
who are wrong due to an acceptance of MA, Travis at least sees the impli-
cations clearly and is not afraid of what he sees. He never blinks. He is
explicit in his endorsement of MA.

If there is nothing which is said in a sentence, in particular, nothing
which is said to be so, then, one would think, the sentence as such,
even under the best of circumstances, could not be true. That not
being its business, one would also expect that there could not be a
(substantive) condition for its truth. . . . Given that a variety of dis-
tinct things might be said to be so in some speaking of the sentence
(at a ﬁxed time), no one of these, and so it seems, nothing, is what we
could sensibly require to be so if the sentence, as such is to be true.
(Travis 1985, p. 188)

Underlying this argument is the assumption that a sentence cannot have
truth conditions unless these are reﬂected in what is said by the sentence.
It assumes that a sentence’s truth conditions must be reﬂected in the truth
conditions of what Travis calls sayings.

Solution: Give Up Blind Adherence to MA

So far we have shown:

1 A wide range of philosophers explicitly endorse some version of

MA.

2 The truth of these versions of MA would support the use of CSAs

of the kind presented and discussed in Chapters 1–3. MA would
give us some reason to think that the intuitions triggered by these
examples have semantic signiﬁcance.

3 If you accept these versions of MA, and thereby, the relevance and

evidential import of those examples, then you are led directly into
the clutches of RC.

From Moderate to Radical Contextualism

From here on in, you have a choice of moving in either of two directions:
either endorse RC or reject every version of MA that renders the intuitions
triggered by the CSAs semantically relevant. Put succinctly: liberate seman-
tics from MA or endorse RC.

A central goal in this book is to present an argument for the ﬁrst option.

If we are right, then the choice is bogus. It is not genuine because, as we will
argue in Chapters 7–9, RC is not only empirically inadequate; it is also inco-
herent. So, the only option remaining is to sacriﬁce suspect versions of MA.

This, we should mention, is a twist on strategies we have used when dis-

cussing MA in earlier work. In Cappelen and Lepore (1997, 1998) we argued
as follows. We assumed that if we could show that RC followed from some
version of MA, then that would constitute a reductio of that version of MA.
We just took it to be plain obvious that we addressed a group of philoso-
phers who would agree with us that such an argument constituted a reduc-
tio, i.e., who agreed that RC is an entirely unacceptable position. In this book
we suspect that assumption. We go one step further and show why RC
should be rejected.

Are There Any Connections Between Semantic Content

and Speech Act Content?

We are not claiming there are no interesting or informative connections
between intuitions about speech act content and semantic content. In
fact, in Chapter 7 we will present tests for context sensitivity and those
tests rely on such intuitions. However, they are very ﬁne grained. The
intuitions about speech act content we think semanticists should rely on
are subtle and exclude all of those intuitions appealed to by MC and RC in
their arguments against Semantic Minimalism. The ﬁne-grainedness we
have in mind is not related to the Gricean strategy of sharply distinguish-
ing what is said from what is implicated. We’ll argue later that that dis-
tinction is superﬁcial and of no signiﬁcance in trying to ﬁnd semantic
content. Since our views will get spelled out in great detail in Chapter 7
we will only give a rough outline of the main idea here. We present two
strategies.

One way to exploit intuitions about what speakers say by uttering S to locate

the semantic content of S is to identify tests that help the theorist focus on the
speech act content that a wide range of utterances of S have in common. A
second strategy we use is to see whether the expression in question behaves
like a classical context sensitive expression (such as ‘here’ and ‘now’) in
certain respects.

Why Context Shifting Arguments are Misused

Semantics, Speech Act Content, and MA

If we succeed in convincing you that there’s no close and immediate con-
nection between semantic content and speech act content, the picture that
emerges, in the end, is this: There are languages. Languages have words.
Words combine into complex expressions and sentences. The semantic
values of words contribute to the semantic values of the complex expres-
sions and sentences of which they are a part. Semantics is about how best
to specify the semantic value of the lexical items and their contribution to
the semantic values of complex expressions and sentences in which they
occur.

On the other hand, when we think about and describe what people say,

i.e., when our aim is to represent or articulate what’s said by an utterance,
we aim to characterize a speaker’s act (that utterance), and in so doing our
aim is to determine something about a particular act in a particular context
C in order to pass it along on to a particular audience situated in a (perhaps
a very) different (sort of) context C*.

If you share this general idea of what semantics is about, and if you agree

with our data about what speakers say by uttering sentences, then there is
no motivation for imposing MA* or some related version of MA as an a
priori constraint on a semantic theory.

Semanticists disagree on what the central semantic features are (truth

conditions, intensions, extensions, propositions, functions from worlds to
truth values, prototypes, stereotypes, situations, or whatever), but they do
tend to agree that semantics is a discipline that aims to characterize sys-
tematically certain features of linguistic expressions and to do so in a way
that captures general truths about languages, and not just truths about par-
ticular speakers in speciﬁc contexts. Characterizations of what speakers
say (claim, state, assert, etc.), on the other hand, are devices used for
characterizing acts (utterances) performed by other speakers. In so doing,
reporters are interested neither in systematicity nor in generality; they aim
to convey something about a particular act in a particular context C to a
particular audience situated in a different context C*.

Reporters draw on information about the speciﬁc intentions of, knowl-

edge about, and the history of, a reported speaker in C and (maybe) similar
features of an audience in C*. These are features one does not want to solicit
when the aim is systematic and general. If this is so, then it would be bizarre
and unmotivated to impose as a requirement on a semantic theory that it
be connected to speech act content in a manner as described by MA*.

From Moderate to Radical Contextualism

C H A P T E R 5

The Instability of Incompleteness

Arguments

This chapter mimics Chapter 3 inasmuch as its chief aim is to establish a
conditional claim, namely:

If Incompleteness Arguments of the kind described in Chapters 1–2
sufﬁce to show that MC is true (i.e., if they sufﬁce to show that an
expression e not in the Basic Set is still context sensitive), it follows
that RC is true.

Again, it is important to keep in mind the conditional nature of this claim.
We do not think Incompleteness Arguments are any good. We will, however,
for the sake of argument, assume that these arguments sufﬁce to establish
MC. And then we will show that if this is so, then RC follows.

The Structure of Incompleteness

Arguments

A typical Incompleteness Argument, as we think of it, comes in two stages.

Stage 1. A solicitation of an intuition to the effect that the proposi-
tion semantically expressed by an utterance of a sentence S (accord-
ing to Semantic Minimalism) is incomplete, i.e., it is not the kind of
thing that can take a truth value.

Stage 2. A solicitation of an intuition to the effect that utterances of
S have a truth value, i.e., that they can express propositions, and
hence, do have truth conditions, and so, can take a truth value.

Conclusion. Something unaccounted for by Semantic Minimalism
must be added in the context of the utterance in order for a complete
proposition to be semantically expressed.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

On its own, this kind of argument doesn’t establish that whatever is added
changes between contexts of utterance. But if the kind of intuitions
appealed to in this argument is supplemented with various context shift-
ing intuitions of the kind we discussed in Chapter 3, then, our opponents
claim, you have good evidence of context sensitivity. The argument, in
summary form, is this:

There’s no proposition expressed if you don’t add something; there
clearly is a proposition there, and moreover, a different one in differ-
ent contexts of utterance. This is a reason for thinking the sentences
in question are context sensitive.

One way to think of this is as a version of the ﬁfth attempt in Chapter 3 to
reply to the instability objection to CSAs, namely, that CSAs, though nec-
essary, are not sufﬁcient to establish context sensitivity. The Moderate Con-
textualist might say: ‘You’re right. CSAs alone don’t establish context
sensitivity, but when combined with the initial intuition that there’s no
proposition there, nothing truth evaluable, without adding something,
then you have sufﬁcient evidence to conclude that the sentences in ques-
tion are context sensitive.’

For the remainder of this chapter we will focus on the ﬁrst stage of these

arguments, i.e., on the solicitation of the intuitions of incompleteness. Our
argumentative strategy is to show that whatever arguments a Moderate
Contextualist might proffer for the view that a certain sentence S is in-
complete, generalize; i.e., if those are sufﬁcient reasons for holding that
S expresses an incomplete proposition, then all sentences do. Then, in
Chapter 10, we will argue that none of the Moderate Contextualist’s
favorite cases are incomplete. So, keep in mind, we do not endorse the
premise from which we are arguing in this chapter. We are using it simply
to present additional evidence for one of our central theses: MC is an unsta-
ble position.

Some Typical Instances of the First Stage

of Incompleteness Arguments

Kent Bach, a very sensible philosopher, has, as far as we know, no revolu-
tionary tendencies. He surely does not sympathize with RC. Nonetheless,
he provides a paradigmatic instance of a Moderate Contextualist who
makes extensive use of Incompleteness Arguments (1994a,b). He asks us to
consider sentences like (1):

(1) Tipper is ready.

From Moderate to Radical Contextualism

His intuition is that if (1) is not completed by adding a propositional com-
ponent that speciﬁes what she is ready for, then all we have is a proposi-
tional fragment; something that does not express a complete proposition;
something that does not have truth conditions; something that cannot
have a truth value. The proposition that Tipper is ready is not, as such, a com-
plete proposition.

For some further typical illustrations along these lines, consider sen-

tences (2)–(4) and their alleged (minimalistic) semantic contents (2*)–(4*):

(2) Steel isn’t strong enough.
(3) Peter’s book is gray.
(4) It’s raining.

(2*) ‘Steel isn’t strong enough’ expresses the proposition that steel

isn’t strong enough and is true iff steel isn’t strong enough.

(3*) ‘Peter’s book is gray’ expresses the proposition that Peter’s book is

gray and is true iff Peter’s book is gray.

(4*) ‘It’s raining’ expresses the proposition that it is raining and is true

iff it’s raining.

These alleged truth conditions are deemed deﬁcient. According to Bach,
(2*) doesn’t express genuine truth conditions. There’s no proposition of
the sort that steel isn’t strong enough. The italicized expression in the previous
sentence just doesn’t refer to something truth evaluable. There’s no such
thing as steel not being strong enough simpliciter. And so on for all of the
other examples.

Criteria for In/completeness

We are going to argue that if this is true, then it is true about any English
sentence. For us to be able to run this kind of argument we need to know
why Bach and other Moderate Contextualists think there’s no such thing
as these minimal propositions. What makes them incomplete? What are
the criteria by which one proposition is deemed incomplete and another
complete?

The only answer we have been able to discern, beyond just appeal to

brute intuitions, is this: (2*) fails to ascribe truth conditions to (2) because
it doesn’t answer questions such as, Strong enough for what? (3*) fails to
ascribe truth conditions to (3) because it doesn’t tell us what the relevant
relationship is between Peter and the book; and (4*) fails to ascribe truth
conditions to (4) because it doesn’t tell us where it’s raining. Had these sen-
tences truth conditions, there would be answers to these questions forth-

Instability of Incompleteness Arguments

coming and these answers would supply part of these sentences’ truth
conditions.

The Instability of Incompleteness

We’ll run our argument using three sets of examples. First we will focus on
the completions Moderate Contextualists propose for (1)–(4) and show that
those completions are equally susceptible to Incompleteness Arguments.
Then we will look at some examples from Travis, an ally in this context since
he, as a Radical Contextualist, is also in the business of showing that there’s
nothing special about (1)–(4). Finally, we will appeal to some of the exam-
ples we presented earlier.

First set of instability illustrations

Suppose we propose, as in fact Moderate Contextualists do, completions of
(1)–(4) along the lines of (1b)–(4b) (see, e.g., Bach 1994b, pp. 128ff.; Carston
1988, p. 167; Sperber and Wilson 1986, p. 188; Recanati 1993, p. 235; Perry
1986, p. 206; Crimmins 1992, p. 17).

(1b) Tipper is ready for the exam.
(2b) Steel isn’t strong enough to support the roof.
(3b) The book owned by Peter is gray.
(4b) It’s raining in Palo Alto.

Our central question is: What exactly distinguishes (1)–(4) from these sug-
gested completions (1b)–(4b)? For, why should we suppose that (1b)–(4b) are
complete, while (1)–(4) are not? What’s the principled difference? None, as far
as we can tell. One can trigger exactly the same kind of incompleteness intu-
itions for (1b)–(4b) as one can for (1)–(4). Consider (1b*)–(4b*):

(1b*) ‘Tipper is ready for the exam’ is true iff Tipper is ready for the

exam.

(2b*) ‘Steel isn’t strong enough to support the roof ’ is true iff steel

isn’t strong enough to support the roof.

From Moderate to Radical Contextualism

1 If you ﬁnd this alleged criterion for incompleteness vague and unsatisfactory, we sympa-
thize. We have simply not been able to ﬁnd any sharp deﬁnition of ‘in/complete’ in this entire
literature. These arguments are essential to many contextualists, but are universally presented
simply by appeal to intuitions, without any formal criterion for in/completeness. Our strat-
egy in what follows is to run an internal objection appealing to (what seems to us to be) the
same kind of intuitions.

(3b*) ‘The book owned by Peter is gray’ is true iff the book owned by

Peter is gray.

(4b*) ‘It’s raining in Palo Alto’ is true iff it’s raining in Palo Alto.

One can argue (as we did in the case of (1*)–(4*)) that (2b*) fails to specify
truth conditions for (2b) (we leave the others as an exercise) because it
doesn’t settle for how long the support must last. Do a few seconds sufﬁce?
More than three days? Many years? Why mustn’t (2b*) also settle whether
(2b) is false if steel fails to support the roof when placed in temperatures
over 390°? Then there is the question of why the amount of steel needed
to support the roof mustn’t be decided in order to settle whether (2b) is
true. Would (2b) be true if one tenth of a square inch of steel wouldn’t
sufﬁce to support the roof?

Nothing in the meanings of the words in (2b) (or their composition)

answers these questions; certainly, nothing in (2b*). If the sort of reasons
that led Moderate Contextualists to conclude that (2*) is incomplete are
any good, why don’t they extend to (2b*) as well? MC has provided us with
no reason to believe that (2b*) gets it just right. If (2*) is incomplete because
it doesn’t answer the question ‘Isn’t strong enough for what?’, then why
isn’t (2b*) also incomplete because it doesn’t answer, inter alia, the question
‘Isn’t strong enough to support the roof for how long?’

So, if the reason why (1*)–(4*) are incomplete is because they fail to

answer these kinds of ‘would it be true if . . .’ -questions, why should we
hold that this alleged incompleteness is limited to some (more or less) cir-
cumscribed subset of English sentences? What Radical Contextualists see
clearly, and are prepared to bravely pursue to the bitter end, is that the same
sort of incompleteness is to be found in every English sentence.

Second set of instability illustrations

Travis uses the same kind of argument as Bach, but he doesn’t think it
applies only to a subset of English sentences. He thinks this kind of argu-
ment can be used to establish that no English sentence has truth condi-
tions or expresses a complete proposition (even when you ﬁx the semantic
values of expressions in the Basic Set, disambiguate, etc.) without large scale
contextual additions. Consider (5) and (6) and their semantic content spec-
iﬁed in (5*) and (6*):

(5) This kettle is black.
(6) Smith weighs 80 kg.

(5*) ‘That kettle is black’ expresses the proposition that [that kettle] is

black and is true iff the demonstrated kettle is black.

Instability of Incompleteness Arguments

(6*) ‘Smith weighs 80 kg’ expresses the proposition that Smith weighs

80 kg and is true iff Smith weighs 80 kg.

According to the incompleteness charge, (5*) fails to specify truth condi-
tions for (5) because it fails to answer any of the following sorts of ques-
tions. (What follows are essentially quotations from Travis (1985, p. 197);
similar ones can be found in Searle (1978, pp. 208, 215; 1980, p. 224).)
Would (5) be true of some demonstrated kettle if it is

(5a) made of normal aluminum but soot covered?
(5b) made of normal aluminum but painted?
(5c) made of cast iron but glowing from heat?
(5d) enameled white on the outside but saturated in black light?
(5e) made of cast iron with a lot of brown grease stains on the

outside?

Since (5*) doesn’t tell us whether (5) is true when the demonstrated kettle
is washed, black on its inside, black all the way through, painted, illumi-
nated or not, etc., RC concludes that it fails to specify (‘settle,’ ‘decide,’
‘determine’) conditions under which (5) is true. Likewise, (6*) doesn’t tell
us whether (6) would be true were Smith to weigh

(6a) 80 kg when stripped in the morning
(6b) 80 kg when dressed normally after lunch
(6c) 80 kg after being force fed 4 liters of water
(6d) 80 kg four hours after having ingested a powerful diuretic
(6e) 80 kg after lunch adorned in heavy outer clothing.

Since (6*) is mute on these (and endless related) questions, RC concludes
that it fails to specify truth conditions for (6).

Third set of instability illustrations

The examples of CSAs illustrate our point in the same way as Travis exam-
ples do. Consider sentences (7)–(10) from Chapter 3 (renumbered here):

(7) John went to the gym.
(8) Jill didn’t have ﬁsh for dinner.
(9) Justine destroyed those shoes.

(10) That’s a dangerous dog.
(11) Justine is a philosopher.

For each of these we can ask questions of exactly the kind that’s alleged to
bring out the incompleteness of (1)–(4). For (7) we can ask: Went to the gym

From Moderate to Radical Contextualism

how? Walked to the vicinity? Did something in the gym? Did what in the
gym? For how long? What if he went into the gym but was sleepwalking?
Etc. We don’t know how to evaluate (7) without settling these questions,
but nothing in (7)’s disquotational truth conditions would answer these
questions. We hope it is obvious how to generalize this point.

Two Attempts to Block the Instability

Argument

The challenge to the Moderate Contextualist is to show how their favorite
cases of incompleteness are fundamentally different from any arbitrary
English sentence. We will now look at two attempts to block the slippery
slide into RC from MC based on incompleteness. (We should mention that
both of these imagined replies seem extraordinarily weak to us, and we
discuss them simply because they are the only ones we have encountered
(the ﬁrst in conversation, the second in a published paper) and we’ve been
unable to think of any better alternatives.)

The psychological reply

A number of commentators have suggested that they can cite a principled
reason both for blocking the slide from MC to RC (i.e., an argument against
our slippery slope argument), and for why we can stop just at expansions
(1b)–(4b). One attempt to block the slide invokes the intentions that
accompany utterances (or, at least those intentions that typically accom-
pany utterances). These contextualists say something like this:

If a kind of completion, C, is not something usually intended by and
communicated by utterances of S, then it’s no argument against S’s
completeness that it fails to explicitly specify (or determine a value
for) C.

Utterances of (1) are usually accompanied by the intention to commu-

nicate what Tipper is ready for and that’s why this is a required completion.
It is only when utterances of S are typically accompanied by the intention
to communicate a completion, C, that C is required for completeness.

Our reply is simply to point out that an incompleteness claim, as we have

construed it, is a metaphysical claim. It is not a psychological claim and it
is not a claim about what speakers usually do. It is a claim about what
propositions exist. It is, for example, the claim that there is no such thing
as the proposition that Jane has had enough. The thing picked out by the ital-
icized part of the previous sentence is not a complete, genuine, real, etc.
proposition, according to those who invoke Incompleteness Arguments.

Instability of Incompleteness Arguments

The reply we are now considering has to do with what propositions are
usually or regularly expressed by the utterance of various sentences. Such
claims, even if true, would not sufﬁce to bolster the relevant metaphysical
claim. It simply doesn’t follow from the fact that we do not usually use ‘It’s
raining’ to express the proposition that it is raining that there is no such
proposition. It is perfectly compatible to hold both that a sentence S isn’t
usually used to express an (alleged) proposition that p, and that p is a real
(complete) proposition.

Taylor’s Reply to the Instability Argument

Ken Taylor tries to respond to our challenge. He thinks ‘It’s raining’ is
incomplete, because it doesn’t specify a location for the rain. But he does
not think that ‘She is dancing’ is incomplete, even though that sentence
also fails to specify a location, the location where the dancing takes place.
Why the difference? Why does a sentence about raining require a location
to be speciﬁed in order to semantically express a proposition, while a state-
ment about dancing does not? Taylor writes:

The answer, I think, has to do with how ‘to dance’ and ‘to rain’ relate
to the places where raining and dancings happen. ‘To dance’ does not
mark the place where a dance happens as the undergoer of the dance.
The theme or undergoer of dancing is the dancer herself. The place
where a dancing ‘takes place’ is merely the place where the dancer
dances. When Laura is dancing in a place, it is not the place that under-
goes the dancing. This, I think, is what explains why despite the fact that
one cannot dance without dancing somewhere or other, a sentence
containing ‘to dance’ can be semantically complete, even if the place
where dancing happens is not contextually provided. That a dancing
must take place somewhere or other is a (mutually known) meta-
physical fact about the universe – a fact that supervenes on the nature
of dancing and the structure of space-time. But that metaphysical fact
is not explicitly reﬂected in the subsyntactic structure of the lexicon.
. . . Things are quite otherwise with the verb ‘to rain’. I take the verb
itself, and its subsyntactic lexical structure, to be the source of the felt
need for the contextual provision of a place or range of places where
a raining happens. Facts about the subsyntactic lexical structure of
the verb directly entail that nothing fully propositionally determinate
has been expressed by an utterance of a sentence like [4] unless a place
is contextually provided. (Taylor 2001, p. 60)

Question: How do we ﬁnd out when a verb has a subject as its ‘theme’ in
the ‘subsyntactic structure’? Taylor responds:

From Moderate to Radical Contextualism

Though subconstituents need not be expressed as sentence-level con-
stituents, they make their presence felt by demanding to be assigned a
contextually supplied value. Thus, though

[4] It’s raining

is missing no syntactic mandatory sentential constituent, nonethe-
less, it is semantically incomplete. The semantic incompleteness is
manifest to us as a felt inability to evaluate the truth value of an utter-
ance of [4] in the absence of a contextually provided location (or range
of locations). This felt need for a contextually provided location has its
source, I claim, in our tacit cognition of the syntactically unexpressed
argument place of the verb ‘to rain’. (Taylor 2001, p. 61; our own
emphasis)

Our reply. There are two central locutions in this passage from Taylor: he
talks about subconstituents ‘demanding to be assigned a contextually
supplied value’ and a ‘felt inability to evaluate the truth value’ of an incom-
plete proposition. We imagine his reply to the instability argument to go
like this:

In (1)–(4) there’s a demand made, and a felt inability to truth evaluate

unless a contextually supplied value is added. In (1*)–(4*) there’s no such
demand made and no such felt inability. The underlying assumption, if this
is to be a reply to the slippery slope argument, must be this: It’s only when
the demand is made and you feel the inability that you’re faced with an
incomplete proposition. That’s how the instability argument is blocked.

If we have understood this right, Taylor’s reply is in effect a version of

the psychological reply. The ﬁrst question we have is this: Why should psy-
chological facts about how people feel have any bearing whatsoever on the
metaphysical question of whether a proposition exists (or is complete)? If
it’s just about the presence or absence of the relevant feelings, this is irrel-
evant. Taylor apparently thinks these feelings reﬂect metaphysical facts, but
we don’t know why he thinks that.

But even if we put that issue aside, the reply misses the mark because it

is possible to trigger these feelings with respect to any sentence whatsoever,
at least if we are allowed to talk to people about the relevant feelings. We’re
not really devoted to the idea of being philosophical therapists, but here is
our meager effort: Take one of Travis’s examples, e.g., ‘Smith weighs 80 kg.’
You might not initially feel that this sentence requires contextual supple-
mentation, not in the same way as ‘She’s ready’ and ‘It’s raining’ do.
However, here’s a prediction: If you read a few of Travis’s little stories, you
will eventually have that feeling. You’ll feel something like this: Smith weigh-
ing 80 kg when? Under what conditions are the weighings done? Do we undress Smith?

Instability of Incompleteness Arguments

We don’t know how to determine that he weighs 80 kg unless the answers
to those questions are (somehow) added to the truth conditions and the
proposition expressed. Or take ‘That’s a red apple.’ On the face of it, you
might, again, not feel strongly that it’s in need of supplementation. But
think about Bezuidenhout’s example and you might end up thinking ‘Red
where? In what light? Scrubbed? Brushed? Painted?’ etc.

In other words, even if we grant Taylor that these feelings are relevant,

there’s no reason to think they are triggered with respect to just a limited
subset of English sentences. What the generalized CSAs show is that the
kinds of considerations that generate these intuitions are not limited to a
speciﬁc subset of sentences.

Conclusion

Here’s a brief summary of the last three chapters. In Chapter 3 we tried to
show that if Context Shifting Arguments sufﬁce to show that MC is true,
then RC follows immediately. In this chapter we tried to establish that if
Incompleteness Arguments sufﬁce to show that MC is true, it also follows
that RC is true. In Chapter 4 we offered a brief diagnosis for why philoso-
phers and linguists have been led to exploit Context Shifting Arguments in
drawing conclusions about semantic context sensitivity. Before moving on
to a direct attack on RC in Part II, we discuss one more attempt to bolster
various versions of MC without slipping into RC.

From Moderate to Radical Contextualism

C H A P T E R 6

Digressions: Binding and

Hidden Indexicals

In this chapter we respond to someone who reacts to Chapters 2–5
as follows: ‘Hold it, guys! No one should hold that the kind of intuition
mongering described in Chapter 2 is sufﬁcient to establish semantic
context sensitivity. You are perfectly right about that. However, when that
kind of evidence is combined with certain kinds of syntactic evidence,
you then have a strong case for semantic context sensitivity. More speciﬁ-
cally: if you have syntactic evidence that there’s a hidden argument at some
level of linguistic representation, e.g., in LF, and you also have strong
attending context shifting intuitions, then you have made your case for this
hidden argument being context sensitive; you are in effect forced by the
conﬂuence of the semantic and syntactic evidence to postulate a hidden
indexical.’

The idea that syntactic evidence is required for the postulation of a

hidden linguistic expression is reasonable. But the idea that syntactic evi-
dence can show that what’s hidden is an indexical is, we think, mistaken.

Our discussion of these issues is structured as follows:

1 We ﬁrst present arguments that have come to be known as ‘Binding

Arguments.’

2 We then show that these arguments alone fail to establish seman-

tic context sensitivity. They are relevant to issues of semantic
context sensitivity only when combined with either Context Shift-
ing Arguments or Incompleteness Arguments. These syntactic
arguments are relevant only to the question of where to locate
context sensitivity in a sentence that is already established to be
context sensitive. Since Context Shifting and Incompleteness Argu-
ments fail to establish the controversial cases (or so we will argue),
Binding Arguments are largely irrelevant to our concerns.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

Though there is much more to be said about Binding Arguments, that’s
really all we need to say about them for the overall purposes of this book.

However, as a service to those with an independent interest in these

issues, or those not entirely convinced by our objections against CSAs and
Incompleteness Arguments, we will add two afterthoughts:

3 First afterthought. Binding Arguments overgenerate: You can con-

struct Binding Arguments to the effect that there are hidden argu-
ment places everywhere (for example, Binding Arguments can be
exploited to establish that there are lots of hidden argument places
in mathematical statements like ‘2

+ 2 = 4’). We take this to be a

reductio of Binding Arguments.

4 Second afterthought. Postulating these hidden argument places

has deeply counterintuitive consequences even for those convinced
that there’s semantic context sensitivity in the relevant cases.

Syntactic Evidence for Hidden Arguments:

Binding

We begin with alleged syntactic evidence for hidden argument places.
Normal utterances of (1) are taken to be about a restricted class of failures,
perhaps, e.g., students in a speciﬁc class.

(1) Many students failed.

One explanation for how this restriction is effected is that quantiﬁer expres-
sions harbor an unpronounced indexical item whose semantic function is
to index a quantiﬁer domain restriction in a context of use. Stanley and
Szabó advocate hidden indexicals; they insist that syntactic evidence must
be adduced for any posited domain variable; and they cite as evidence the
fact that these posited domain variables interact in binding relations with
other quantiﬁer expressions. In sentences like (1), they claim, one can bind
its hidden constituent, enabling its domain to vary according to the values
introduced by a variable-binding operator (Stanley and Szabó 2000a, p. 243;
cf. also Stanley 2000). They infer this constituent must be present in (1).
So, for example, in (2)

(2) In every class, many students failed,

the domain associated with ‘many students’ varies as a function of the
values introduced by ‘every class.’ (2) means (on one of its readings), accord-
ing to Stanley and Szabó, (2*):

From Moderate to Radical Contextualism

(2*) [Every (x): class (x)][many ( y): student in x ( y)](failed y).

Assuming binding to be a syntactic phenomenon,

such examples would

seem to provide evidence for a variable somewhere in the syntactic struc-
ture of quantiﬁed noun phrases.

(Stanley and Szabó go as far as to suggest

that without positing a hidden domain variable, it is not clear that sen-
tences like (2) express ‘coherent propositions at all’ (Stanley and Szabó
2000a, p. 243).)

Stanley and Szabó generalize their idea by associating with each nominal

an indexical, which when unbound behaves like a free variable to which a
semantic value must be contextually assigned. So construed, (1) is inter-
preted along the lines of (1*),

(1*) Many students (i) failed,

where ‘i’ is a hidden indexical (in (1)) that, in a context of use, picks out a
set (or property) which functions to restrict the extension of ‘student,’ and
thereby restricts the domain of ‘many’ further than ‘student’ does by itself.
(Their actual account is more complex, but nothing we say requires us to
go into the extra detail here.)

The Binding Argument is alleged to generalize. To see how, consider

sentence (3):

(3) It’s sunny.

Notice that [3] can be embedded in a larger sentence:

[4] Everywhere Sally goes, it is sunny.

Intuitively, what [4] says, or at least a natural reading of [4], is that for
every place that Sally goes, it is sunny at that place. So we should repre-
sent the logical form [4] something along the following lines:

[4*] For all places, x, if Sally goes to x, then it is sunny at x.

The quantiﬁer phrase ‘Everywhere Sally goes’ is binding a place vari-
able in the logical form of ‘It is sunny’ – otherwise there would be
nothing for the quantiﬁer phrase to bind. (Nelson 2001, pp. 27–8; see
also Stanley 2000, pp. 415–17)

Digressions: Binding and Hidden Indexicals

1 We think they might be best accounted for pragmatically, but that’s a topic for another
occasion (see Cappelen and Lepore (1997, 2002a,b); see also Farkas (1997), who denies that
the data require a syntactic treatment but are instead best accounted for semantically).
2 For the fuller development of this theory, see Stanley and Szabó (2000a).

Stanley, in fact, is optimistic that he can establish that all nominal expres-
sions harbor context sensitive indices:

Extra-linguistic context appears to have a profound effect on the
determination of what is expressed by the use of linguistic expres-
sions. For a bewildering range of very different linguistic construc-
tions, adhering to relatively straightforward linguistic intuition about
what is expressed leads us to the conclusion that facts about the non-
linguistic context play many different roles in determining what is
said. Furthermore, that so many different constructions betray this
sort of sensitivity to extra-linguistic context understandably leads to
pessimism about rescuing the straightforward intuitions while
preserving any sort of systematicity in the theory of meaning. A pre-
sumption motivating the pessimistic inclination is that, if we accept
the ordinary intuitions, what appear to be very different ways in which
context affects semantic content in fact are different ways in which
context affects linguistic content. Pessimism is a natural reaction to
those who adopt this presumption, because if appearance is a good
guide to the facts in this domain, then there are just too many ways
in which context affects semantic content to preserve systematicity.
One common and natural reaction to these facts is, therefore, to deny
the semantic signiﬁcance of the ordinary intuitions, thereby relegat-
ing the project of explaining the apparent effects of extra-linguistic
context on semantic content to a domain of inquiry outside the
theory of meaning proper. So doing removes the threat context poses
to the systematicity of semantic explanation, but at the cost of reduc-
ing the interest of the semantic project. In this paper, I explore a dif-
ferent reaction to the situation. My purpose is to undermine the
presumption that what appear to be very different effects of context
on semantic content are very different effects. My challenge is of
necessity rather limited, since it is too implausible to trace all effects
of extra-linguistic context on semantic content to the very same
source. Rather, I will take, as a case study, three superﬁcially very dif-
ferent effects of context on semantic content, and show that they are
due to the very same mechanism, what I call Nominal Restriction. I
thereby hope to provide convincing evidence of the promise of the
project of reducing all apparent effects of context on semantic
content to a small number of sources. (Stanley 2002b, p. 364)

The Irrelevance of the Binding Argument

Our ﬁrst reaction to this argument is exceedingly simple. Let’s suppose, for
the sake of argument, that the Binding Argument is sound. What does it

From Moderate to Radical Contextualism

show? Well, it shows that the logical–syntactic form of ‘Every penguin has
a tail’ is something like ‘Every penguin

(d)

has a tail.’ That argument alone

doesn’t show that the hidden argument place d is semantically context sen-
sitive. It doesn’t show that different utterances of ‘Every penguin has a tail’
have different truth conditions, and so, can disagree in truth value. All it
would show is that there is a hidden argument place there. Notice, the fol-
lowing two claims are logically compatible:

Hidden domain variable. Every noun phrase in a quantiﬁed noun phrase
cohabits with a domain variable in LF.

Contextual insensitivity of quantiﬁer domain restrictions. Sentences con-
taining quantiﬁed noun phrases are semantically stable (with respect
to their domains).

Even if you have shown that there’s a hidden argument place in a sentence,
you still need an additional argument to establish that that sentence
exhibits semantic context sensitivity (or any other sort of context sensitiv-
ity for that matter).

There’s a signiﬁcant logical gap between the claim that there’s a hidden

syntactic unit in S and the claim that S is semantically context sensitive.
You get from hidden syntactic entities to semantic context sensitivity using
the kind of arguments we presented in Chapter 2.

We should point out that this is not a view that’s a surprise to propo-

nents of Binding Arguments. Recall, we quoted one major proponent of the
Binding Argument, namely, Stanley, extensively in Chapter 2. He’s certainly
aware of the fact that he needs additional arguments.

Some readers might opt now to skip the rest of this chapter. For this

chapter, as we mentioned earlier, should be of interest only to those who,
for some reason or other, are unconvinced by our objections to the supple-
mentary arguments (i.e., those arguments that are supposed to combine
with the Binding Arguments, in particular, the CSAs and Incompleteness
Arguments), or to those researchers with a special interest in binding per se.

In the remainder of this chapter we present four ‘internal’ objections to

the Binding Argument. We show that if you think the ‘classical’ versions of
this argument are sound, then there are innumerable ‘hidden’ argument
places in, e.g., the arithmetical sentence ‘2

+ 2 = 4.’ We take establishing that

this sentence has indeﬁnitely many ‘hidden’ argument places to be a reduc-
tio of the Binding Argument. We also show that postulating these hidden
argument places has deeply counterintuitive implications: it predicts that
there should be anaphoric relations where there are none; and it predicts
that certain sentences, indeed, some of them false, express a priori and/or
necessary truths and falsehoods.

Digressions: Binding and Hidden Indexicals

First Afterthought: Reductio of Binding

Argument

A confused mathematical anthropologist (call her Sally) trying to ﬁnd out
if mathematical truths are universal utters (5) as a summary of her ﬁnd-
ings:

(5) Everywhere I go, 2

+ 2 = 4.

Here’s the Binding Argument applied to (5):

Intuitively, (5) says that for every place Sally goes, 2

+ 2 = 4 at that

place. So we should present the logical form of (5) along the follow-
ing lines:

(5*) For all places, x, if Sally goes to x, then 2

+ 2 = 4 at x.

The quantiﬁer phrase ‘Everywhere Sally goes’ is binding a place vari-
able in the logical form of ‘2

+ 2 = 4’ – otherwise, there would be

nothing for the quantiﬁer phrase to bind. This establishes that the
logical form of the sentence ‘2

+ 2 = 4’ has a freely occurring place

variable.

Since there is obviously no variable ranging over locations in ‘2

+ 2 = 4,’ this

is a reductio of the Binding Argument.

We would like to leave the argument here; it is close to indisputable that

arithmetical statements lack hidden indexicals referring to places. However,
since no bullet is unbiteable, two brief remarks are in order.

First, since both speakers and audiences are blissfully unaware of any

reference to a location in utterances of ‘2

+ 2 = 4,’ the referent of a hidden

indexical would have to be ﬁxed in a manner entirely unconnected with
speaker intentions. Such a reference ﬁxing mechanism would be unique.
We are owed a substantive story about how it is achieved.

Second, those who recognize a place index in ‘2

+ 2 = 4’ are vulnerable

to a most slippery slope, for consider (5**):

(5**) No matter where Sally goes, no matter when she goes there,

+ 2 = 4.

Based on the Binding Argument should the logical form of (5**) be (5

+)?

+) For all places x, for all times y, if Sally goes to x at time y, then

+ 2 = 4 at x at y.

No one should want to conclude that the complex quantiﬁer expression
‘No matter where I go, no matter when I go there’ binds two hidden vari-

From Moderate to Radical Contextualism

ables in ‘2

+ 2 = 4.’ Such examples illustrate that a blind endorsement

of the Binding Argument might ultimately require positing indeﬁnitely
many dedicated variables in every single sentence. Other examples that
might help you see the point, if, for reasons beyond us, you haven’t already,
are:

Wherever I kiss her, she smiles [hidden argument place for location
in ‘kiss’].

Whenever I kiss her with my socks on, she smiles [hidden argument
place for what we wear].

Notice that our unboundedness charge has an analog in an earlier debate
about event verbs and adverbial modiﬁcation. Recall, according to one early
proposal, ascribing (6*) as (6)’s logical form was supposed to explain why
(7) entails (6).

(6) Mary kissed John.

(6*) Mary kissed John in some place.

(7) Mary kissed John in the park.

We increase the adicity of an event verb like ‘kiss’ in order to accommodate
an inferential relation. However, following this strategy consistently
would require each event verb to harbor indeﬁnitely many ‘hidden’
places in order to accommodate inferential data among (6)–(10), etc. (see
Davidson 1967).

(8) Mary kissed John in the park after midnight.
(9) Mary kissed John in the park after midnight behind his left ear.

(10) Mary kissed John in the park after midnight behind his left ear

on August 24, 1999.

So, for example, in order to explain how (10) logically implies (6), should
we treat (6)’s logical form as (11)?

(11) Mary kissed John in some place at some time behind some place on

some date.

Following this strategy would render the adicity of an ordinary verb like
‘kiss’ indeﬁnitely large. How would anyone ever learn his language (Larson
and Segal 1995, p. 468)?

Likewise, the seeming unboundedness of the Binding Argument requires

too many indexicals. So, the Binding Argument fails to be decisive for the

Digressions: Binding and Hidden Indexicals

existence of hidden indexicals.

We turn now to further constraints on

indexicality.

Second Afterthought: Three Additional Objections

to the Postulation of Hidden Indexicals

Positing hidden linguistic expressions incurs certain obligations.
With indexicals there are at least two:

on the syntactic side, a posited

indexical should enter into anaphoric relationships; on the semantic–
epistemological side, it should generate certain kinds of a priori truths
and falsehoods; and it should not generate certain necessary truths and
falsehoods. We discuss these in turn.

Anaphora

Overt indexicals can participate in anaphoric relationships. In (12) and (13)
the antecedents of ‘it’ and ‘himself ’ are the indexicals ‘that’ and ‘he.’

(12) That’s a table but it is not a book.
(13) He’s a senator who likes himself.

Since hidden indexicals are just the same indexicals, they too should be
capable of entering into anaphoric relationships. So, if (1) harbors a hidden
reference to a restricted domain, (14) should be intelligible, with ‘it’
anaphoric.

?(14) Many students failed, and it is a big domain.

That (14) makes no sense (even though (14*) below does) is evidence
against a hidden indexical in (1).

From Moderate to Radical Contextualism

3 What does the Binding Argument show? The data it invokes are interesting, and require
explanation. The facts are these: Sentences like (1), (3), and ‘2

+ 2 = 4’ lack a place variable (or

any other hidden variable of the sort Stanley and Szabó posit). However, these sentences are
still grammatical, and so their initial quantiﬁers are nonvacuous. Without positing hidden
indexicals of the sort Stanley and Szabó favor, how can we explain their grammaticality? It
goes beyond the scope of this book to provide an answer to this grammatical question;
we refer the interested reader to Cappelen and Lepore (2002b), where we sketch a possible
answer.
4 Hidden indexicals fail other tests, e.g., so-called weak cross-violations, but we’ll not pursue
this criticism. Cf., Blair (2004) and Hawthorne (2003).

(14*) Many students in this (domain) failed, and it is a big domain.

For another illustration, consider Stanley and Szabó’s view that (15),

roughly, has the form (15*),

(15) Tigers are mammals

(15*) Tigers (i) are mammals,

where ‘i’ indexes, in a context of use, a domain. But, then, (15**) should make
as much sense as (15

+) does, where ‘it’ is alleged to be anaphoric on ‘i.’

(15**) Tigers are mammals, and it is a big domain.

(15

+) Tigers in this (domain) are mammals, and it is a big domain.

The impossibility of reading ‘it’ anaphorically on the alleged indexical is
evidence against a hidden indexical in (15).

We are not committed to the view that every alleged covert element goes

our way in this respect. For example, it is not easy to make reference to the
covert subject of ‘please’ in ‘John is easy to please.’ But even if it’s hard or
impossible to get anaphora on controlled ‘PRO’ because the potential con-
structions have their own ‘PRO’ controlled by a matrix subject, that in itself
might be good reason to treat controlled ‘PRO’ as a special case. Our point
is that either you do get anaphora, or there is an independent explanation
for why not; minimally proponents of unpronounced indexicals owe us an
independent explanation.

Digressions: Binding and Hidden Indexicals

5 Stanley (2002b) claims (p. 368) that the reason for the unavailability of anaphoric link is
due to the fact that the domain variable cohabits the node with the noun. This case is alleged
to be similar to that of incorporation, as the contrast between (a) and (b) illustrates:

(a) John owns a bicycle. He rides it daily.
(b) John is a bicycle-owner. He rides it daily.

Of course, one needs to motivate claims about incorporation, if the latter notion is taken in
its technical sense. So, we need to hear Stanley’s argument that the index ‘cohabits’ a node
with the noun, as opposed to merely being a sister of the noun. This can’t be stipulated, since
one can’t stipulate that there is syntax that behaves as though it isn’t there. Also, incorpora-
tion typically involves some kind of movement, driven for some independent reason. We know
that compounding of the ‘bicycle-owner’ variety blocks modiﬁcation of various sorts; and even
if we don’t know why this is the case, we have independent grounds for treating ‘bicycle-owner’
as a compound. But while we see that appeal to incorporation blunts the worry Stanley faces,
we don’t see the independent reason for thinking that the (alleged) syntax of indices is rele-
vantly like the familiar cases. Thanks to Paul Pietroski for walking us through this subtle
debate.

A Priori Truths

According to Kaplan, it is an essential feature of an indexical that its lin-
guistic meaning can be used to generate certain kinds of a priori truths.

Intuitively, (6) [‘I’m here now’] is deeply, and in some sense, which we
will shortly make precise, universally true. One need only understand
the meaning of (6) to know that it cannot be uttered false. No such
guarantees apply to (7) [‘D.K. is in Portland on 3/26/1977’]. A Logic of
Indexicals which does not reﬂect this intuitive difference between (6)
and (7) has bypassed something essential to the logic of indexicals.
(Kaplan 1989a, p. 509; our own emphasis)

These ‘universal’ truths are generated as follows: Kaplan identiﬁes the
linguistic meaning of an expression with its character, which is a function
that delivers the expression’s content at each context. So, the character of
the ﬁrst person indexical ‘I’ is a function on contexts whose value at
any context is the agent of that context. Suppose the character of an index-
ical D speciﬁes that its referent in a context, U, is whatever object satisﬁes
conditions C in U. Then an a priori truth will be expressed by an utterance
of

D satisﬁes conditions C.

This kind of sentence cannot be uttered falsely. For instance, no utterance
of (16) or (17) is false; and anyone who understands ‘I’ and ‘you’ will
recognize this, that is, it constitutes a priori knowledge.

(16) I am the person who utters this sentence.
(17) You are the person addressed by this utterance.

However, no utterance of (16) or (17) is necessarily true; whatever proposi-
tion an utterance of (16) expresses is false in any context, say, where the
utterer does not exist. Since any speaker could fail to exist, this proposition
is contingent. Kaplan infers that his semantics for indexical expressions
provides examples of the contingent a priori.

We are now positioned to state our worry about hidden indexicals. In all

of these cases none of these essential features is manifested.

According to Stanley and Szabó (2000a), (1) properly construed means

the same as (1*). Since ‘i’ is an expression it has a character, i.e., a linguis-
tic meaning. Given what Stanley and Szabó (2000a) say we assume they
intend the character of ‘i’ to be something like ‘the contextually salient
domain.’ We (or they) might be wrong about this. But this indexical, accord-

From Moderate to Radical Contextualism

ing to Kaplan, must have some character or other; call whatever it is F. Then
(18) (or (18*)) ought to be such that their every utterance is true and this
is knowable a priori.

(18) Everyone is in the contextually salient domain.

(18*) Everyone is in F.

Likewise, speakers should know a priori that every utterance of (19)–(20) is
false.

(19) Some ducks are only in nonsalient domains.
(20) At least one little duck is not in a salient domain.

However, not every utterance of (18) is true, and not every utterance of (19)
or (20) is false. (We doubt any utterance of (19) or (20) is false.) Hence, no
one has any such a priori knowledge.

If we are right, the analogy between hidden indexicals referring to con-

textually salient domains and ordinary overt indexicals breaks down. But if
Kaplan is right about the semantics and epistemology of indexicals, it
follows that Stanley and Szabó (2000a) must be wrong. Minimally, they
need to defend their departure.

Our goal is not to conclusively establish that all hidden indexicals fail

these two above constraints, but it is, however, fairly simple to construct
similar arguments for every example mentioned in the introduction. As a
brief illustration, consider comparative adjectives. Suppose, as is commonly
supposed (Ludlow 1989; Stanley 2002b), sentences with comparative adjec-
tives contain a hidden reference to a comparison class and that for each
context of utterance it is a contextually salient comparison class that’s ref-
erenced. It should then, ﬁrst, be possible to refer anaphorically to these
classes, as in (21) and, second, to generate certain kinds of a priori truths
or falsehoods, as in (22).

(21) She’s tall, and it has many 5-year-old members.
(22) She’s tall, but not compared to a salient class.

However, we can’t get ‘it’ in (21) to refer to a comparison class, and (22)

doesn’t seem a priori false to us.

Nonexistent Interpretations?

On the hidden indexical proposal, an utterance of (23) can be used to
express a truth, because its context of utterance can effect a domain restric-

Digressions: Binding and Hidden Indexicals

tion on an underspeciﬁed quantiﬁer: context restricts the range of its quan-
tiﬁer ‘every.’

(23) Every table is covered with books.

Suppose that in using (23) a speaker succeeds in restricting the domain of
the quantiﬁer ‘every’ to tables in domain D, and thereby expresses what she
would have had she instead uttered (23

¢).

(23

¢) Every table in domain D is covered with books.

What, then, shall we conclude about (24)?

(24) Every table is in domain D.

(24) is as likely to be used to express a truth as (23). Since (23) can express
what (23

¢) can in the same context, why can’t we infer that (24) can express

in a context what (24

¢) would?

(24

¢) Every table in domain D is in domain D.

The hidden indexical account is a semantic one about what can be said (or
expressed) by sentences with underspeciﬁed quantiﬁers and so complain-
ing that (24

¢) looks less informative than (24) needn’t be a problem. After

all, a present utterance of ‘I am here now’ by Lepore and ‘Lepore is at
Rutgers University on November 21, 2003’ aren’t equally informative, but
on at least one respectable semantic story, these sentences can be used to
say (or express) the same proposition. If the same proposition can be
expressed by nonsynonymous sentences, where one, in some sense, is ana-
lytic and the other is not, then perhaps that sort of explanation extends to
(24) and (24

¢).

A genuine challenge to the hidden indexical account must establish that,

once appropriately contextually relativized, (24) and (24

¢) do indeed express

distinct propositions. Here are considerations intended to support this chal-
lenge.

(24) does not have a true necessary reading, yet (24

¢) does. Minimally, it’s

worth pointing out that embracing the hidden indexical account requires
accommodating such necessary interpretations.

Much along the same lines, but worse, consider the sentence ‘Every table

isn’t in domain D.’ It doesn’t seem to have a reading under which it
expresses a necessarily false proposition, but the hidden indexical account
predicts that it should, since, assigning its negation narrow scope, any utter-
ance of ‘Every table in domain D isn’t in domain D’ expresses a necessarily

From Moderate to Radical Contextualism

false proposition. Represented as ‘[Every x: table in D(x)]NOT(x is in D),’
to be true, it would have to be true of every table in D that it is false that it
is in D.

Though not knockout punches, the accretion of these counterintuitions

succeeds in clarifying commitments inherited from adopting the hidden
indexical account.

Each of these various cases appeals to speaker intu-

itions, but these intuitions seem no less ﬁrm than whatever other intuitions
were supposed to have motivated the hidden indexical account in the ﬁrst
place.

Screening Off

A hidden indexicalist might protest that these alleged troublesome inter-
pretations are never available. Perhaps a policy of screening off is in place,
thereby excluding any contextually determined domain that is explicitly ref-
erenced in a predicate (or elsewhere). So, for example, take domain D – if
explicitly referenced by an utterance of (24), it is screened off as a candidate
domain restriction.

(24) Every table is in domain D.

This constraint is supposed to preempt contextual domain restrictions

that would render some seemingly contingent sentences as expressing nec-
essarily true or false propositions. Any domain referenced, or expressed,
explicitly is thereby rendered contextually irrelevant, at least for the pur-
poses of restricting an underspeciﬁed quantiﬁer. If this ploy can be made
to work, then what’s expressed with (24) cannot be what would be expressed
with (24

¢), but, perhaps, instead what would be expressed with (25).

(24

¢) Every table in domain D is in domain D.

(25) Every table in domain E is in domain D.

Contextually supplementing a token of ‘every table’ by indexing, say, a

domain E secures uniqueness without rendering an utterance of (24), rela-
tive to the same assumptions, as expressing a necessity.

Digressions: Binding and Hidden Indexicals

6 Of course, our intuition is also that we cannot express with an utterance of ‘Every table is
in domain D’ a necessarily false proposition. But what prevents, on semantic grounds alone,
a restriction from issuing in a proposition that expresses what ‘Every table in domain E is in
domain D’ would, with nonintersecting domains indexed? That is to say, what prevents the
restricted domain of the utterance from being the most contextually salient domain?

Whatever can be said in favor of screening off surely reﬂects nothing

more than handy wisdom about the pragmatics of sound interpretation,
and nothing about semantics. For how can semantics prohibit an explicitly
referenced domain from also being the most contextually salient domain
restriction?

With uses of quantiﬁer expressions, as ‘Every table in domain

D,’ an explicitly referenced domain is typically also contextually salient.
Indeed, why can’t a speaker stipulate beforehand that she wants a certain
domain (and that domain alone) to be the most contextually salient
domain of our conversation, and then proceed with (24)? Even in such cir-
cumstances, we presume, what she expresses is not, in any sense, rendered
necessary.

Furthermore, even if some sort of screening off strategy could be made

to work in the simple cases we have been discussing – though we don’t see
how – it wouldn’t help to avoid counterintuitive consequences for more
complicated cases.

Contexts may or may not be accurately representable as sequences of

items which context sensitive expressions can take as semantic values
(

·speaker, addressee, time, place, topic of discourse, perceptually salient

objects, etc.

Ò),

but we presume it’s not controversial that in any given

context at most ﬁnitely many domains are salient. Let C be an ordering

·a

, . . . , a

Ò of every salient domain, and then try to denote a table with a use

of a sentence of form (26),

(26) Every table

j(i),

where ‘i’ indexes a domain of C, and ‘

j’ speciﬁes the predicate of which ‘i’

is a constituent. ‘

j(i),’ for example, might be ‘is domain D,’ etc. None of

these envisaged instances of (26) seems to express a necessary truth (or false-
hood) in C, yet counterparts of form (27) can:

(27) Every table (i

)

(as would ‘Every table in domain D is in domain D,’ ‘Every table in domain
E is in domain D,’ and so on). Suppose, because ‘i

’ occurs in the predicate

in (26), it follows that distinct contextually salient domains must be
indexed by ‘i

’ and ‘i

’ in (27). But consider a new sentence with enough dis-

joined predicates such that each item in C can be picked out by a distinct
index, as in (28).

From Moderate to Radical Contextualism

7 Or worse, hearkening back to the last footnote, the screening off strategy would somehow
have to be made to work so as to prohibit internal incoherence, as with ‘Every table in E is in
D’ (where E and D are nonintersecting domains).
8 Montague (1974); Kaplan (1989); and Lewis (1970).

(28) Every table

) or

) or . . .

According to the hidden indexical proposals, an utterance of (28) expresses
in C what (28

¢) would.

(28

¢) Every table (i

) is

) or

) or . . .

But then (28), which expresses a seemingly contingent claim about the con-
textually salient setting of what, if anything, ‘the table’ denotes, has been
transformed into a sentence which in that same contextual setting has a
(nearly) necessary reading, without a possibility of further screening off.
That we can devise such sentences might convince you something is fun-
damentally wrong with the hidden indexical account.

Conclusion

As we noted at the beginning of this chapter, whether the Binding Argu-
ment is sound or not (in any given case) in no way can establish the pres-
ence of semantic context sensitivity. Just the same, the argument has
attracted enough attention that we thought it worthwhile to refute. Having
done so, we return to the chief topics of this book, the adequacy of Radical
Contextualism. After having established in Chapters 3–5 that Moderate
Contextualism is unstable, a view that inevitably slips into Radical Con-
textualism, in the next few chapters we will discuss and argue for, ﬁrst, its
empirical inadequacy and then its internal incoherence.

Digressions: Binding and Hidden Indexicals

PA R T I I

Refutation of Radical

Contextualism

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

C H A P T E R 7

Objections to Radical Contextualism

(I): Fails Context Sensitivity Tests

Intuitions are the contextualist’s bread and butter. The literature is chock
full of tales about various people who for various reasons under diverse cir-
cumstances utter the same sentence with different imagined audiences but
shared environments. We, as readers, are invited to tap into our intuitions
about these stories; and reﬂection on our intuitions is supposed to convince
us of the truth of RC. There’s a sense, therefore, in which RC is an empiri-
cal thesis, based as it is on a variety of contingent features about human
psychology, in particular, based on the contingent fact that we happen to
have certain intuitions.

The central objection that Radical Contextualists

run against Semantic Minimalism is that it disrespects these intuitions.

Our response is that Semantic Minimalism is both sufﬁciently attentive

and adequately respectful of our actual linguistic practices. RC, on the con-
trary, we’ll argue, is the true culprit; it is radically empirically inadequate.
The Radical Contextualist’s seduction works only on someone whose focus
is on an astonishingly limited range of communicative acts. As soon as one
tries to accommodate a wider range of data, RC runs into insurmountable
empirical obstacles.

To establish our critical point, we will focus on three obvious (but over-

looked – at least overlooked by Radical Contextualists) tests of context sen-
sitivity. These three tests all have the form: An expression e is context sensitive
only if competent speakers have certain intuitions about uses of certain sorts of sen-

1 We don’t mean anything particularly loaded by our use of ‘empirical.’ We mean only that
it is a contingent fact that we have the said intuitions. We need to check that we have them.
In this manner the push for RC is a form of experimentation. For those who want to call intu-
ition mongering nonempirical, feel free to do so; we have no attachment to the word ‘empir-
ical’ or any more general philosophical assumptions about empiricality. Our critical points
stick regardless of whether they are categorized as empirical or not.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

tences containing e. These tests appeal to fundamental features of linguistic
communication, each incompatible with RC.

One more preliminary: MC is a view with much greater currency than

RC. Hardly any contemporary philosopher rejects MC. Therefore, it is
crucial for us to register that we could just as well have run our objections
in this chapter directly against MC. We opted not to do so, because, as
already established in Chapters 3 and 5, MC collapses into RC. If, however,
for some reason you think we failed, simply apply the objections in this
chapter directly to MC itself.

One ﬁnal preliminary: At the end of this chapter we have included two

important appendices. In the ﬁrst, ‘Contextual Salience Absorption,’ we
respond to a reply we speculate contextualists might make to the entire line
of reasoning in this chapter (we’ve put in an appendix since we’ve never
actually seen this response in print). In the second, ‘Diagnosis: Monsters
and Use–Mention Fallacies,’ we present an equally speculative account of
why contextualists tend to ignore, overlook, dismiss, etc. the kinds of tests
we appeal to in this chapter.

The rest of this chapter has three parts structured around the three tests.

We present each test, show why we think RC doesn’t pass it, and respond
to some potential replies.

Test 1: An Expression is Context Sensitive Only if it

Typically Blocks Inter-Contextual Disquotational

Indirect Reports

‘Inter-Contextual Disquotational Indirect Reports’ is just an ugly term for
the following fairly obvious phenomenon:

Take an utterance u of a sentence S by speaker A in context C. An
Inter-Contextual Disquotational Indirect Report of u is an utterance
u

¢ in a context C¢ (where C¢ π C) of ‘A said that S.’

Put intuitively, we suggest using such reports to test for context sensitivity
as follows: If the occurrence of an expression e in a sentence tends to block
disquotational indirect reports (i.e., render such reports false), then you
have evidence that e is context sensitive. Take the ﬁrst person pronoun ‘I.’
Sentences containing ‘I’ cannot be disquotationally indirectly reported
(except by self-reporters); utterances of ‘now’ cannot be disquotationally

Refutation of Radical Contextualism

2 Of course, uttered as a report of u (if you want to make that explicit in the report just add
‘by uttering u’ after ‘S’).

reported (except by simultaneous reporters); utterances containing the
demonstrative expression ‘that’ cannot be disquotationally reported (except
by co-demonstrating reporters), and so on for the other members of the
Basic Set.

This provides the basis for the following test. Suppose you suspect, or at

least want to ascertain whether, e is context sensitive. Take an utterance u
of a sentence S containing e in context C. Let C

¢ be a context relevantly dif-

ferent from C (i.e., different according to the standards signiﬁcant accord-
ing to contextualists about e). If there’s a true disquotational* indirect
report of u in C

¢, then that’s evidence S is context insensitive. (To be ‘dis-

quotational*’ just means you can adjust the semantic values of components
of S that are generally recognized as context sensitive, i.e., we just test for
the controversial components.)

If this exercise sounds confusing, it shouldn’t. Take an obviously context

sensitive expression, e.g., take ‘tomorrow.’ Consider an utterance by Rupert
on Tuesday of ‘John will go to Paris tomorrow.’ If someone tries to report on
Wednesday what Rupert said with his utterance on Tuesday with an utter-
ance of ‘Rupert said that John will go to Paris tomorrow,’ his report is false
because the expression ‘tomorrow’ fails to pick out what it picked out in
Rupert’s original utterance. The presence of ‘tomorrow’ in the disquota-
tional report ﬁgures prominently in an explanation of why the report is false.

Real context sensitive expressions block Inter-Contextual Disquotational* Indirect
Reports. By deﬁnition, for e to be context sensitive is for e to shift its seman-
tic value from one context of utterance to another. So, if e is context sensi-
tive and Rupert uses e in context C, and Lepore uses it in context C

¢, and

the relevant contextual features change, then it will be just an accident if
their uses of e end up with the same semantic value. In particular, if Lepore
ﬁnds himself in a context other than Rupert’s and wants to utter a sentence
that matches the semantic content of Rupert’s utterance of a sentence with
e, he can’t use e, i.e., he can’t report Rupert’s utterance disquotationally.

All the expressions in the Basic Set block Inter-Contextual Disquota-

tional Indirect Reports. This can easily be veriﬁed; a couple of additional
illustrations should sufﬁce:

Illustration 1
Utterance made by George Bush, June 3, 2003: ‘I wasn’t ready
yesterday.’

Indirect report by Lepore, June 5, 2003: ‘Bush said that I wasn’t ready
yesterday.’

Lepore’s report is false because his use of ‘I’ and ‘yesterday’ fail to pick out
the person and the day Bush picked out using those same words. These

Objections to Radical Contextualism (I)

words are such that they don’t support Inter-Contextual Disquotational*
Indirect Reports.

Illustration 2
Utterance made by Lepore: ‘You should wear that’ (where Justine is
the contextually salient audience and a blue hat is being demon-
strated).

Indirect report by Cappelen: ‘Lepore said that you should wear that’
in a context where Ludlow is the contextually salient audience and a
pair of sandals are being demonstrated.

Again, this report fails because Cappelen’s use of the words ‘you’ and ‘that’
pick out something other than what those same words picked out in
Lepore’s utterance. These words are such that they don’t support Inter-
Contextual Disquotational* Indirect Reports.

More generally, none of the expressions in the Basic Set supports Inter-

Contextual Disquotational* Indirect Reports.

Objection 1 to RC: RC postulates context sensitivity for expressions that do

not pass the Inter-Contextual Disquotational Indirect Report Test

We’re in a particular context now; we’re in a café on 5th Street in New York
City between Avenues A and B; it’s one of those hot and muggy New York
summer evenings, the lights are out, and we’re drinking iced tea (look, we’re
not the ones who think any of this is relevant; we’re just trying to humor
Radical Contextualists by providing ‘relevant’ data). Call our context 5stC.
In 5stC we’re engaging in one of the Radical Contextualist’s favorite pas-
times: we’re cooking up thought experiments involving distinct utterances
of a single sentence under diverse circumstances.

We’re thinking about different utterances of ‘John is ready.’

We’re imag-

ining the following two contexts of utterance of (1):

(1) John is ready.

Context of Utterance C1. In a conversation about exam preparation,
someone raises the question of whether John is well prepared. Nina
utters (1).

Context of Utterance C2. Three people are about to leave an apartment;
they are getting dressed for heavy rain. Nina utters (1).

Refutation of Radical Contextualism

3 We’re using this example because it provides the best case possible for contextualism. If
they can’t even show that ‘John is ready’ is context sensitive, we doubt that they have a better
chance with any other sentence.

Here’s a pretty obvious fact: whichever of these contexts of utterance we
consider, no matter how dissimilar you think they are from each other, each
of your utterances can still be reported disquotationally. To ensure this is utterly
obvious, we’ll right now engage in actual speech acts; the indented sen-
tences below represent actual utterances by us in 5stC. These are acts in
which we are describing the two utterances of (1) by Nina; in (1.1) we report
on her utterance in C1, in (1.2) her utterance in C2:

(1.1) Nina said that John is ready.
(1.2) Nina said that John is ready.
(1.3) In both C1 and C2, Nina said that John is ready.

Two features of our exercise are particularly signiﬁcant. First, it’s our intu-
ition that all three of these reports are true. Second, we can say that all three
reports are true in this context, i.e., in 5stC. Sitting here in a café on 5th
Street between Avenues A and B in New York City, sipping on our ice tea,
we can accurately disquotationally report what Nina said in contexts
C1–C2. Note that these contexts are not only different from each other, they
are also each radically different from 5stC. And this isn’t some weird fact
we have pulled out of our hats to refute RC. It’s completely trivial, obvious,
and ubiquitous. Here’s a bold conjecture: For any utterance of (1), we can
utter ‘The speaker said that John is ready’ and utter something true.

(1.1)–(1.3) illustrate two fundamental problems for contextualism:

1 According to RC, the two utterances of (1) assert (say, claim) radi-

cally different propositions. What each says depends on features
speciﬁc to their contexts of utterance. How, then, is it that we are
now able to use in 5stC an utterance of ‘She said that John was
ready’ to describe accurately and literally what she said in those dif-
ferent contexts? That shouldn’t be possible if RC is right. For, if RC
is right, the proposition expressed by an utterance of (1) (as it occurs
in the complement clauses of (1.1)–(1.3)) in 5stC should be shaped
by features speciﬁc to 5stC, and we have no reason to think that
these features match the contextual features relevant in the origi-
nal utterance. More speciﬁcally, we have no reason to think those
change between uttering (1.1) and (1.2) so that the content changes
appropriately. In other words, contextualism cannot account for
this most fundamental feature of linguistic communication.

2 According to RC, the two utterances of (1) that we report on express

radically different propositions. If so, (1.3) should be impossible.
The complement clause of (1.3) cannot express more than a single
proposition. If (1.3) is true, then both the imagined utterances of
(1) said (or expressed) the proposition expressed by (1) as it occurs
in the complement clause of that report.

Objections to Radical Contextualism (I)

It is worth pointing out here that this data indicates that a single propo-

sition is expressed by all these utterances. We’ll later suggest that this might
be the proposition which, according to Semantic Minimalism, is semanti-
cally expressed. More on that possibility in Part III of this book.

We hope it’s obvious that our point has nothing to do speciﬁcally with

features peculiar to the examples involving utterances of (1). Just to clinch
this point, we’ll provide an additional example. (Feel free to skip it if it’s
obvious to you how to generalize from (1).)

Bezuidenhout’s writings on RC provide an excellent source of data; one

of her many examples, one we used in Chapter 3, involves her and her son
discussing red apples. (By the way, just reread the previous sentence and
you’ve got the point we’re about to make.) Here’s a slightly revised version
of her example:

Context of Utterance C1. We’re at a county fair sorting through a barrel
of apples. The apples are sorted into different bags according to the
color of their skin. Some have green skin; others have red skin. Anne
utters:

(2) The apple is red.

Context of Utterance C2. We’re sorting through a barrel of apples to
identify and discard those afﬂicted with a horrible fungal disease.
This fungus grows out from the core and stains the ﬂesh of the apple
red. One of us is slicing apples open, placing the good ones into a
cooking pot. The bad ones are tossed. Cutting open an apple Anne
again utters (2).

We are still on 5th Street. We’re thinking about what Anne said in C1

and in C2. Here’s what we think about C1:

(2.1) Anne said that the apple was red.

This is what we think about C2:

(2.2) Anne said that the apple was red.

Come to think about it:

(2.3) Both in C1 and in C2, Anne said that the apple was red.

Elaboration on the kinds of contexts that support Inter-Contextual

Disquotational Indirect Reports

The examples of true indirect reports discussed in the previous section con-
stitute an objection to contextualism in part because the context that the

Refutation of Radical Contextualism

report is made in is relevantly different from the context of the reported utter-
ance. We want to draw your attention to four important kinds of differ-
ences between the reporting context and the reported context:

Basic differences between contexts. The point of our short description of

5stC was just to point out that the context we found ourselves in when
reporting on (1) and (2) was different from the original context of utter-
ance in a whole range of potentially important respects. Compare 5stC and
the context Anne found herself in when talking to her son about red apples.
Those contexts differ at least in the following respects: perceptual inputs,
accompanying activities, previous conversational context, purpose of con-
versation, nature of audience, and assumptions shared by conversational
participants.

Reports under ignorance. Sometimes the person reporting on an utterance

might be ignorant of the relevant contextual features of the original context
of utterance; that is, someone uttering (2.1) might not know what Anne had
in mind in the original context of utterance. The reporter might not know
whether Anne cared about the inside or the outside of the apple, whether
she looked at the apple through sunglasses or under water. That ignorance,
however, needn’t inﬂuence the truth value of the indirect report.

Reports under indifference. This is a variation on reports under ignorance,

but deserves separate mention. Someone can utter ‘A said that Rudolf is
red’ in a context where no kind of redness is salient. It’s what we’ll call an
indifferent context. Such contexts differ from some ignorant contexts, since in
some ‘ignorant’ contexts, the speakers care about what the relevant features
are; they just don’t know what they are. Indifference need not inﬂuence the
truth value of disquotational indirect reports.

Reports based on mistaken assumptions. Sometimes the reporter has false

beliefs about the original context of utterance. The speaker of (2.1) might
believe that in the original context of utterance redness on the inside
when seen through sunglasses was salient. Suppose he’s wrong. What was
salient was redness on the inside when seen without sunglasses. Such false
beliefs need not inﬂuence the truth value of the disquotational indirect
report.

It might be useful to pause here for a moment and compare the effects

on indirect reports of the controversial cases of context sensitivity (i.e.,
those discussed above and in more detail below) to the effects of real
context sensitive expressions (i.e., members of the Basic Set, e.g., ‘I,’ ‘that,’
‘now,’ etc.). Consider, for example, an utterance of ‘That’s a nice one.’ If
someone overheard this utterance, but did not know what was demon-
strated, one of four things would typically happen:

Objections to Radical Contextualism (I)

䊉

The potential reporter might say, I don’t know what he said, since I don’t
know what was demonstrated.

䊉

She might do some investigation, i.e., try to ﬁnd out what was
demonstrated and, if the investigation was successful, then indi-
rectly report.

䊉

She might provide a direct quote, i.e., say: ‘She said “That’s a nice
one.” ’

䊉

She might say something like ‘She said some demonstrated object
was nice.’

This contrasts radically with what we would do if we overheard, e.g., ‘A is
red’ uttered in a context where we didn’t know what ‘kind’ of redness was
in question (i.e., whether the speaker intended to describe A as red when
washed, scrubbed, painted, red on the inside or outside, etc.). As pointed
out above, in such cases we do not (typically) ﬁnd it problematic to report
the utterance with: She said that A is red. (Same point applies to the other
expressions discussed above.) Notice, for example, you never hear people say
things like: ‘She said that A is red for some contextually salient way of being
red.’

More illustrations of allegedly context sensitive expressions that fail the

Inter-Contextual Disquotational Indirect Report Test

We venture the following daring hypothesis: Every instance of an allegedly
context sensitive expression that is not a member of the Basic Set fails this
test. What we will now do is simply report on our own intuitions involving
disquotational indirect reports.

Consider the claims below, in part, as reports of our own intuitions, and

in part, as hypotheses susceptible to falsiﬁcation. These are test cases for
the various contextualist claims that have been made (and documented in
Chapter 2) about knowledge, moral and psychological attributions, about
the context sensitivity of ordinary nominals, attributive and comparative
adjectives, about weather and temporal reports, about possessive construc-
tions, about quantiﬁer expressions, about ‘ready’ and ‘enough,’ geometri-
cal expressions, and various modal constructions. We are keeping an open
mind on all these cases, and it is certainly possible that someone can devise
a scenario in which these predictions fail (and we can’t explain why they
fail), but here is how we see things for the time being. (In all these cases
assume that the indirect report takes place in a context that’s relevantly dif-
ferent from the original utterance.

)

Refutation of Radical Contextualism

4 i.e., they differ with respect to whatever features the contextualist thinks determine content.

(Know)
Any utterance of ‘A knows that he has a hand’ can be reported by
‘She said that A knows that he has a hand’ and any two such
utterances can be reported by ‘They both said that A knows that he
has a hand.’

(Believes)
Any utterance of ‘A believes that B is shady’ can be reported by ‘She
said that A believes that B is shady’ and any two such utterances can
be reported by ‘They both said that A believes that B is shady.’

(Nominals (i.e., Stanley’s (2002b) view), e.g., penguin)
Any utterance of ‘Penguins have soft beaks’ can be reported by ‘She
said that penguins have soft beaks’ and any two such utterances can
be reported by ‘They both said that penguins have soft beaks.’

(Adjectives in general, e.g., blue)
Any utterance of ‘Jackie has blue eyes’ can be reported by ‘She said
that Jackie has blue eyes’ and any two such utterances can be reported
by ‘They both said that Jackie has blue eyes.’

(Quantiﬁers, e.g., There is at least one)
Any utterance of ‘There is at least one duck in Norway’ can be
reported by ‘She said that there is at least one duck in Norway’ and
any two such utterances can be reported by ‘They both said that there
is at least one duck in Norway.’

(Enough)
Any utterance of ‘Steel isn’t strong enough’ can be reported by ‘She
said that steel isn’t strong enough’ and any two such utterances can
be reported by ‘They both said that steel isn’t strong enough.’

(Possessives)
Any utterance of ‘Rudolf ’s penguin is happy’ can be reported by ‘She
said that Rudolf ’s penguin is happy’ and any two such utterances can
be reported by ‘They both said that Rudolf ’s penguin is happy.’

(Comparative adjectives, e.g., tall)
Any utterance of ‘A is tall’ can be reported by ‘She said that A is tall’
and any two such utterances can be reported by ‘They both said that
A is tall.’

(Moral terms, e.g., bad)
Any utterance of ‘Killing penguins is bad’ can be reported by ‘She said
that killing penguins is bad’ and any two such utterances can be
reported by ‘They both said that killing penguins is bad.’

Objections to Radical Contextualism (I)

(Geometrical terms, e.g., hexagonal)
Any utterance of ‘France is hexagonal’ can be reported by ‘She said
that France is hexagonal’ and any two such utterances can be reported
by ‘They both said that France is hexagonal.’

(Modals, e.g., could have, would have)
Any utterance of ‘If Oswald hadn’t shot Kennedy, someone else would
have’ can be reported by ‘She said that if Oswald hadn’t shot Kennedy,
someone else would have’ and any two such utterances can be
reported by ‘They both said that if Oswald hadn’t shot Kennedy,
someone else would have.’

(Weather reports, e.g., It’s raining)
Any utterance of ‘It’s raining’ can be reported by ‘She said that it’s
raining’ and any two such utterances can be reported by ‘They both
said that it’s raining.’

(Temporal reports, e.g., It’s three p.m.)
Any utterance of ‘It’s three p.m.’ can be reported by ‘She said that it’s
three p.m.’ and any two such utterances can be reported by ‘They both
said that it’s three p.m.’

The main point of these illustrations is that if you agree with the

data, then you are endorsing our objection against any contextualist about
the expressions just discussed: Inter-Contextual Disquotational Indirect Reports,
though perfectly natural and ubiquitous, are inexplicable on the assumption that
RC (or some version of MC that extends to any of these above cases) is true. The
challenge is to explain away or challenge this data (or show that the test
is no good). (Keep in mind, though, that in order to do that a single
counterexample is not enough. We present these claims as generalizations
about all utterances of various sentences because that’s what we believe, but
this generalization is not essential to our objection. If a Radical (or Mod-
erate) Contextualist agrees that there are some true reports of these kinds,
then she has a problem. There should be no such readings according to RC
or MC.

) Here are several responses we imagine a contextualist might

tender.

Refutation of Radical Contextualism

5 Here’s a guess: whatever technical apparatus is introduced to account for the cases you can
hear will be powerful enough to account for the cases you allegedly can’t hear. That is to say,
in order for a theory to account for the disquotational reports you agree with us on, you’ll end
up introducing apparatus that’s powerful enough to account for the readings you disagree
with us about. If that is so, then it’ll turn out to be surprising if we couldn’t get all the read-
ings we can get. You’ll in effect be committed to the semantic possibility of even the ‘strange’
ones, and what’ll end up being strange is that you can’t hear them.

Reply 1: The indirect reports are false

The ﬁrst reply goes like this: ‘It’s just not true that these utterances can be
so reported. The indirect reports are all false. She didn’t say that John was
ready and they didn’t both say that John was ready. (Anne didn’t say that
both apples were red.)’

Our reply. The objection should be made more ﬁne grained. With respect
to (1.1), (1.2), (2.1), and (2.2), we’re quite conﬁdent that nonbiased intro-
spection supports our estimation of the facts. We’re equally sure that the
collective reports, those in (1.3) and (2.3), have overwhelming foundation
in our practice of indirect reporting. There might be contexts in which they
are less natural, but all we need for our point to stick is that there are con-
texts in which they seem perfectly natural. That is what we wholeheartedly
believe. It is an empirical claim and it seems to us to be almost obviously
true. As we see it, this is a cornerstone of our communicative practices,
namely, the idea that two or more people said the same thing.

Reply 2: There’s an abstract content that they all have in common

We can imagine two versions of this reply. According to the ﬁrst, all utter-
ances of ‘John is ready’ express a very abstract proposition, namely, the
proposition that John is ready for something or other. That’s why they can all be
reported in 5stC. That’s also why it makes sense to collectively report them
by ‘They all said that John is ready.’ So, the idea is to just ﬁnd something
very abstract that they all have in common, and that abstract property is
what’s attributed to John by all these utterances.

It should be obvious that this is not a view available to a Radical Con-

textualist. It is, in effect, the denial of RC. RC is the view that what’s said is
a richer content, to wit, a content speciﬁc to the context of utterance. To
endorse the view that the said-content is this abstract proposition is in
effect to relinquish RC. It’s to endorse Semantic Minimalism (except that
our view is not that all utterances of ‘John is ready’ express the proposition
that John is ready for something or other, but rather that they all express the
proposition that John is ready, and we don’t want to characterize this propo-
sition as abstract – see Part III).

A revised version might be this: Well, one of the propositions expressed

by any utterance of ‘John is ready’ is the abstract proposition. Then other
propositions are expressed too. These other propositions are speciﬁc to the
context of utterance.

We would be more than happy if the Radical Contextualist proffered this

reply. We say: Welcome to Semantic Minimalism and (a version of) Speech

Objections to Radical Contextualism (I)

Act Pluralism, i.e., to endorse this view is to accept (a version of) our view,
so we strongly encourage this move.

Reply 3: ‘What’s said’ doesn’t track content/propositions expressed

The Radical Contextualist might respond: ‘Look, all this evidence is based
on intuitions we have about indirect reports and those intuitions are noto-
riously unreliable. Even if we can do weird stuff in indirect reports, that’s
no evidence of the content expressed by the utterance reported. There’s a
sharp distinction between how we think about what utterances say, and
what the real content of those utterances is. Evidence based on what’s said
doesn’t support conclusions about which propositions were expressed in
the original contexts of utterance.’ It will become clear below (Chapter 13)
that we endorse a version of this view. It is, however, not a reply available
to the Radical Contextualist. Without appeals to intuitions about what’s
said by utterances, there’s no basis for RC and no basis for their alleged evi-
dence against Semantic Minimalism. Radical Contextualists are the ones
who base their arguments on intuitions about what speakers say and about
what’s said by utterances. They are the ones who think intuitions about
what speakers say should be captured. They are the ones who think Seman-
tic Minimalism fails precisely because it fails to account accurately for what
speakers say. (See Chapter 2 for textual evidence.) If they give up such
appeals, we have no idea how anyone could argue for RC and against
Semantic Minimalism. (It is sometimes suggested to us that Radical Con-
textualists do not need to rely on intuitions about what utterances say, and
instead rely on direct intuitions about the truth value (or truth conditions)
of utterances. Our brief response is threefold: First, as a matter of exegesis
this is not true. See Chapter 2. All the arguments that we have encountered
for RC are based on intuitions about what speakers say. Second, we don’t
know what it is to have intuitions about the truth value of utterances as
such. If we are asked to have intuitions not about what an utterance says,
asserts, claims, etc. but just about its truth value, we are at a loss. Third, if
someone claims to have such intuitions, we would like to know why they
are not supportive of Semantic Minimalism, i.e., if we were to encounter
someone who claims to have intuitions just about the truth conditions of
an utterance of ‘John is ready’ or ‘The apple is red’ (and not the truth con-
ditions of what is said by such an utterance), he would have to explain to
us why those truth conditions are not just that John is ready or that the
apple is red.

Refutation of Radical Contextualism

Reply 4: ‘Said that’ is context sensitive

The Radical Contextualist might say: ‘Whether or not such indirect reports
are true depends on the context of the report. In some contexts, a disquo-
tational report of an utterance u might be acceptable, but in other contexts
not. It all depends on the context of utterance for the report.’

This might be true (we argued for a view like this (1997) and discuss it

further in Chapter 13), but even if it is, it has no bearing on the present
issue. All we are claiming is that Anne said that the apple was red. If what
we said by the previous sentence is true, and it is, then that settles the
matter. Suppose there are contexts in which ‘Anne said that the apple was
red’ is false (we are not saying there are such contexts, but suppose). All we
need is that it is true that Anne said that the apple was red. That’s it.

We turn now to our second test for context sensitivity.

Test 2: Context Sensitive Expressions Block

Collective Descriptions

If a verb phrase v is context sensitive (i.e., if it changes its semantic value
from one context of use to another), then on the basis of merely knowing
that there are two contexts of utterance in which ‘A v-s’ and ‘B v-s’ are true
respectively, we cannot automatically infer that there is a context in which
‘v’ can be used to describe what A and B have both done. In short, from
there being contexts of utterance in which ‘A v-s’ and ‘B v-s’ are true it
doesn’t follow that there is a true utterance of ‘A and B both v.’ This is
because the semantic value of ‘v’ in the previous sentence is determined in
one context, and we have no guarantee that that semantic value, whatever
it is, ‘captures’ (whatever that means) the semantic values of ‘v’ in those con-
texts of utterance where they were used solo.

On the other hand, if for a range of true utterances of the form ‘A v-s’

and ‘B v-s’ we obviously can describe what they all have in common by using
‘v’ (i.e., by using ‘A and B v’), then that’s evidence in favor of the view that
‘v’ in these different utterances has the same semantic content, and hence,
is not context sensitive.

A parallel point extends to singular terms.

Objections to Radical Contextualism (I)

6 The argument can be summarized as follows: If ‘v’ is a context sensitive term, then its
semantic value can change from one utterance to another. So, ‘A v-s’ and ‘B v-s’ might attribute
different properties to A and B. But it doesn’t follow that ‘v’ can be used to describe what A
and B share. Maybe by chance someone might be able to use ‘v’ in some context to refer to a
property they both share, but that would be a coincidence. In other words: e is context sensi-
tive only if there’s no guarantee of collective usage. Suppose there were a guarantee of collec-
tive usage, then a use of ‘v’ in one context would ‘denote’ (have as its semantic value) what all
other utterances of ‘v’ denote and we would be guaranteed collective descriptions.

If a singular term N is context insensitive and there’s a range of true

utterances of the form ‘N is F’ and ‘N is G,’ then we, for example, in this
context, can truly utter ‘N is F and G.’ If, however, N is context sensitive,
we shouldn’t be able to do this. As an illustration consider the context sen-
sitive ‘yesterday’: Suppose we know there are two contexts in which ‘Yes-
terday John left’ and ‘Yesterday Bill left’ are true respectively (though we
don’t know the times of these contexts). It doesn’t follow that there is a
context in which ‘Yesterday John and Bill left’ is true. Again, all of the
expressions we list as members of the Basic Set pass this test of collectivity.
We won’t bore the reader with more illustrations, so we leave these thought
experiments as homework.

Objection 2 to RC: RC postulates context sensitivity that fails the

Collective Descriptions Test

We ﬁrst present you with two circumstances involving Mr. Smith and Mr.
Jones. We then ask you to think about how Smith and Jones are to be prop-
erly (i.e., truly) described.

Circumstance 1. Smith, who is an astronaut, steps out of his spacecraft

onto a new and unexplored planet. As usual, he has brought his extremely
accurate scale with him (he does that since he’s curious about gravitation).
The ﬁrst thing he does is step onto it, in full astronaut outﬁt, and it
registers ‘80 kg.’ His fellow astronauts look at him, and utter: ‘Smith weighs
80 kg.’

Circumstance 2. Jones is at home on earth, it’s morning, he’s naked, he

hasn’t had breakfast, but he did go to the bathroom, and now he steps onto
an extremely accurate scale. It shows ‘80 kg.’ His friends gleefully exclaim
‘Jones weighs 80 kg.’

Consider the following facts about how we, and, we assume, every other

competent English speaker, would describe these circumstances. (Note: The
objection we are running focuses on how we would describe Mr. Smith and
Mr. Jones in these two circumstances; not how we would describe what
people say in these contexts, and not what we would have and could have
said in these circumstances.) We’re interested in how we (i.e., C&L) can actu-
ally now, in this one context, describe the facts. Here’s the description that
we ﬁnd natural of Circumstance 1:

(C1) Smith weighs 80 kg.

Here’s a description of Circumstance 2 that we also ﬁnd natural:

Refutation of Radical Contextualism

100

(C2) Jones weighs 80 kg.

It is also true that:

(C3) Both Smith and Jones weigh 80 kg.

Our objection should by now be obvious: RC postulates context sensitivity
in cases where collective descriptions are perfectly natural, contrary to
what’s demanded by the Collective Descriptions Test. In particular, RC
cannot explain how the utterance (C3) above follows from the truth of the
two stories we told about Smith and Jones. RC proponents claim that the
truth conditions for utterances of ‘Smith weighs 80 kg’ and ‘Jones weighs
80 kg’ vary between contexts of utterance, that they semantically express dif-
ferent propositions contingent upon the peculiarities of the context of
utterance. But if all of the circumstances of evaluation that make these dis-
tinct utterances true (that make the propositions semantically expressed
true) are ones in which Smith and Jones weigh 80 kg, then the RC claim is
false. The truth of (C3) provides evidence that all utterances, e.g., of ‘Smith
weighs 80 kg,’ are true just in case Smith weighs 80 kg; and that all such
utterances semantically express the same proposition, namely, that Smith
weighs 80 kg. Ditto for ‘Jones weighs 80 kg.’

More illustrations of expressions that fail the Collective Descriptions Test

Again, our intuitions involving collective descriptions are inconsistent with
RC (and MC) across the board. The collectivity test as formulated above

Objections to Radical Contextualism (I)

101

7 A related test for context sensitivity concerns the fact that context sensitive expressions
have ﬁxed interpretations in Verb Phrase ellipsis (or VP-ellipsis). In the following sentences

Frank likes my mother, and Bob does too.
George lives near you, and so does Bill.
Frank bought this, and Martha did too.

there are no available interpretations where Frank and Bob like different people, or George
and Bill live near different people, or Frank and Martha bought different things. Obviously,
noncontext sensitive expressions do not exhibit this feature, as in

John bought a car, and so did Bill.

Nothing in the meaning of this sentence requires that it be the same car that John and Bill
bought. With this understanding of context sensitivity we can see that we get the same results
as with the collective readings for the troublesome cases. In the circumstances described above
we can infer

(C4) Smith weighs 80 kg and Jones does too.

applies to verb phrases and noun phrases, and, to our ears, none of the verb
or noun phrases alleged to be nonobviously context sensitive by RC (and
MC) pass the Collective Descriptions Test for context sensitivity (again, we
simply report on our intuitions here, and encourage readers to come up
with counterexamples to change our minds):

(Know)
If there is a true utterance of ‘A knows that he has a hand’ in context
of utterance C, and another true utterance of ‘B knows that he has a
hand’ in relevantly different context C

¢, the following collective

description is perfectly natural: ‘Both A and B know that they have
hands’ and, as the VP-ellipsis Test (see note 7) suggests, so is ‘A knows
that she has hands, and B does too.’

(Believes)
If there is a true utterance of ‘A believes that B is shady’ in context of
utterance C, and relevantly different true utterance of ‘C believes that
B is shady’ in relevantly different context C

¢, the following collective

description is perfectly natural: ‘Both A and C believe that B is shady,’
and, as the VP-ellipsis Test suggests, so is ‘A believes that C is shady,
and B does too.’

(Nominals (i.e., Stanley’s (2002b) view), e.g., penguin)
If there is a true utterance of ‘Penguins are happy’ in context of utter-
ance C, and another true utterance of ‘Penguins are lazy’ in a rele-
vantly different context C

¢, then the following collective description

is perfectly natural: ‘Penguins are both lazy and happy,’ and, as the
VP-ellipsis Test suggests, so is ‘Penguins are happy, and lazy too.’

It’s fairly obvious how to extend the test so it applies to adjectives and
adverbs:

Refutation of Radical Contextualism

102

8 Interestingly enough, Stanley (2003a) invokes the VP-ellipsis Test himself in arguing against
contextualists about vague predicates, e.g., Soames (1999), Kamp (1981), and Raffman (1994,
1996). What’s especially interesting is that it looks to us as if the VP-ellipsis Test does not go
the way Stanley should want it to go given his comments on domain speciﬁcation. Recall that
according to Stanley and Szabó (2000a) quantiﬁer expressions like ‘a store’ should index dif-
ferent domain restrictions in different contexts of utterance, i.e., they are supposed to be
context sensitive. Yet under VP-ellipsis their interpretation does not seem ﬁxed. So, for
example, in the sentence

John went to a store, and so did Bill

we feel no compulsion to restrict the domain of quantiﬁcation of ‘a store’ elliptic in the second
conjunct to the same one as the one that occurs (explicitly) in the ﬁrst conjunct. The ﬁrst
might be restricted to stores in New Jersey, or car stores, or whatever; and the second may be
restricted to anything whatsoever.

(Adjectives in general, e.g., blue)
If there is a true utterance of ‘Jackie has blue shoes’ in context of utter-
ance C, and another true utterance of ‘Jackie has blue sunglasses’ in
a relevantly different context C

¢, then the following collective descrip-

tion is perfectly natural: ‘Jackie has blue shoes and sunglasses.’

(Geometrical terms, e.g., hexagonal)
If there is a true utterance of ‘France is hexagonal’ in context of utter-
ance C, and another true utterance of ‘Berlin is hexagonal’ in a rele-
vantly different context C

¢, then the following collective descriptions

are perfectly natural: ‘France and Berlin are both hexagonal’ and, as
the VP-ellipsis Test suggests, so is ‘France is hexagonal and so is
Berlin.’

(Comparative adjectives, e.g., tall)
If there is a true utterance of ‘Mount Everest is tall’ in a context of
utterance C, and another true utterance of ‘The Empire State Build-
ing is tall’ in a relevantly different context C

¢, then the following col-

lective description is perfectly natural: ‘Mount Everest and the Empire
State Building are both tall,’ and, as the VP-ellipsis Test suggests, so
is ‘Mount Everest is tall, and the Empire State Building is too.’

(Enough)
If there is a true utterance of ‘Jackie has had enough’ in a context of
utterance C, and another true utterance of ‘Jill has had enough’ in a
relevantly different context C

¢, then the following collective descrip-

tion is perfectly natural: ‘Both Jill and Jackie have had enough’ and,
as the VP-ellipsis Test suggests, so is ‘Jackie has had enough, and Jill
too.’

(Possessives)
If there is a true utterance of ‘Jill’s daughter is happy’ in a context of
utterance C, and another true utterance of ‘Jill’s dog is happy’ in a rel-
evantly different context C

¢, then the following collective description

is perfectly natural: ‘Jill’s dog and daughter are happy.’

It is considerably more difﬁcult to extend these tests to quantiﬁer expres-

sions, in part because there’s a great deal of dispute about where exactly to
locate the allegedly context sensitive component of a quantiﬁer expression.
Some posit a context sensitive argument place attached to the nominal
(Stanley and Szabó 2000a), some claim it is in the quantiﬁer itself (Wester-
ståhl 1989). The former option has been tested for above (we have shown
that nominals don’t pass the test); the latter is difﬁcult to test for since
some collective readings are blocked for simple logical reasons. But here is
an intuitive instance of the test extended to a quantiﬁer expression.

Objections to Radical Contextualism (I)

103

Consider the view that ‘at least two’ is context sensitive. It follows from

this view that the following collective description should be blocked, even
though (at least to our ears) it is obviously not (we should report that we
have come across a few stragglers who do not share our intuitions):

(Quantiﬁers)
If there are true utterances of ‘Jill bought at least two penguins’ and
‘Jill bought at least two ducks’ in two relevantly different contexts,
then the following collective description is perfectly natural: ‘Jill
bought at least two penguins and ducks.’

The challenge is the same as in connection with the Inter-Contextual

Disquotational Indirect Report Test: If RC (or MC) is true, none of these
collective sentences should be true. If you have the intuition that the col-
lective sentences are true in at least some of these cases, then you need either
to ﬁnd a way to explain away those intuitions or to ﬁnd a way to accom-
modate them. Our hypothesis: These intuitions can’t be explained away and
they can’t be accommodated within the framework of RC or MC.

Test 3: Context Sensitive Expressions Pass an Inter-

Contextual Disquotational Test and Admit of Real

Context Shifting Arguments

We now present two closely related tests (so closely related that they in
effect are the same test described in different ways) that a theorist can
perform to determine whether an expression e is context sensitive. These
(just as the previous two) tests are ‘live’ in this sense: they require the the-
orist to actually use e while performing the test. They require the theorist
to confront intuitions about her own language in use, and not just about
other people’s use of language (or her use of e in other contexts).

To introduce the ﬁrst such test, note, once again, that it is a constitutive

mark of a context sensitive expression e that it can be used with different
extensions (semantic values) in different contexts of utterance. This is
surely a big part of what it means to say of any expression that it is context
sensitive. It follows from this constitutive fact alone that for any context
sensitive expression e our use of e in this context (i.e., in the context of this
chapter) with whatever extension it takes on in this context need not be the
same as whatever extension it takes on in another context. There can be no
denying that this is so.

Refutation of Radical Contextualism

104

9 Notice that we are not here presenting a general collectivity test for quantiﬁers. We remain
neutral about that possibility.

Based on this constitutive fact about context sensitivity, the following

test recommends itself for judging whether an expression e is context sen-
sitive or not: Simply look and see whether e behaves as it should by actu-
ally using e in a context of utterance (and thereby ﬁxing its semantic value
in that context) and simultaneously describe another use of e with a dis-
tinct semantic value in another context.

Since e is not context sensitive unless its semantic values can shift from

context to context, and since the semantic value e takes in, say, this context
of utterance (i.e., the context of this book) can be distinct from the seman-
tic value it takes in some other context, to test whether e is context sensi-
tive or not, simply use e; in order to use e, put it in a sentence S and then
use S. e is context sensitive only if there is a true utterance of an instance
of the following schema for Inter-Contextual Disquotation (ICD, for short;
where S contains e):

(ICD) There are (or can

be) false utterances of

even though S.

(Alternatively, run the test in reverse.

) Unless e passes this ICD Test, it is

not context sensitive.

Here’s a concrete example that should help to concretize the discussion,

if you are not yet getting it.

Suppose we (i.e., C&L) are trying to determine whether ‘she’ is context

sensitive. To do so, according to the ICD Test, we choose a sentence S con-
taining ‘she,’ e.g., (1):

(1) She is French.

We then proceed to assert an instance of ICD with respect to (1), namely,
(2a):

(2a) There is (or can be) a false utterance of ‘She is French’ even

though she is French.

Objections to Radical Contextualism (I)

105

10 In our formulation of ICD we appeal to possible utterances. We do that because it is some-
thing our opponents, i.e., Radical (and Moderate) Contextualists, do all the time. However, if
you’re worried about quantiﬁcation over possible utterances (or worried it will be difﬁcult to
specify the relevant domain of possible utterances), run the test on actual (past, present, or
future) utterances.
11 That is, it can take the form of an utterance of ‘There is at least one true utterance of

even though it is not the case that S.’

12 Of course, care must be practiced in choosing S; e.g., S mustn’t contain any context sen-
sitive expressions other than possibly e, or, if it does, then restrict the domain of ‘There are
utterances’ so that additional context sensitive expressions take the same semantic values in
the imagined contexts as in the context of use.

Since, as a matter of fact, we are now actually pointing at a French woman,
(2a) is true, that is, in the context of this chapter, i.e., in 5stC; and so, ‘she’
is context sensitive. If you need convincing, ask yourself whether intuitively
you think this utterance of (2a) (taking into account the relevant inten-
tions, demonstrations, etc.) is true. Since the answer is obviously ‘yes’ – con-
sider someone else’s utterance of (1) who is pointing at a non-French
woman – (1), and so ‘she,’ are established to be context sensitive.

Here are additional illustrations involving further expressions from the

Basic Set:

(2b) There is a false utterance of ‘That’s nice’ even though that’s nice

[said pointing at Al’s car].

Suppose Al’s car is nice. Then, obviously, any utterance where someone
points at anything other than Al’s car that isn’t nice sufﬁces to establish
there are true utterances of (2b).

(2c) There is a false utterance of ‘I’m hungry’ even though I am

hungry.

Suppose I’m hungry. Then, obviously, any utterance of ‘I’m hungry’ by a
speaker who is not me and who is not hungry sufﬁces to establish that the
test utterance of (2c) expresses a truth.

(2d) There is a false utterance of ‘Tom is leaving now’ even though

Tom is leaving now.

Suppose Tom is leaving now. Then, obviously, any utterance of ‘Tom is
leaving now’ made at times other than now, say, a few days into the future
when Tom isn’t leaving, sufﬁces to establish that the test utterance of (2d)
expresses a truth.

These stories decisively illustrate that expressions from the Basic Set pass

ICD with ﬂying colors. That there are intuitively true utterances of (2a)–(2d)
in perfectly ordinary circumstances sufﬁces to establish that not all utter-
ances of the following biconditionals are true.

‘She is French’ is true just in case she is French.
‘That’s nice’ is true just in case that’s nice.
‘I’m hungry’ is true just in case I’m hungry.
‘Tom is leaving now’ is true just in case Tom is leaving now.

Real Context Shifting Arguments

A reasonable question to ask is how do we elicit intuitions in others that
an expression e does, as a matter of fact, pass ICD? We believe the best

Refutation of Radical Contextualism

106

manner in which to proceed is by providing a Real Context Shifting Argument.
Indeed, an expression e passes the ICD Test just in case it is possible to con-
struct a Real Context Shifting Argument involving e.

The context in which a Context Shifting Argument (CSA) is told we

are calling the Storytelling Context; and the context about which a CSA is
told we are calling the Target Context. In trying to elicit intuitions about
context shifting either of two sorts of stories can be devised from within the
Storytelling Context: with one sort, the alleged context sensitive expression
e doesn’t get used (in the Storytelling Context), but is instead only men-
tioned in describing its uses in Target Contexts; with the other sort, the
alleged context sensitive expression e is both used (in the Storytelling
Context) and also mentioned in describing its uses in Target Contexts. We’ll
call the ﬁrst sort of Context Shifting Argument Impoverished (ICSA, for
short) and the latter sort of Context Shifting Argument Real (RCSA, for
short).

Both Radical and Moderate Contextualists invariably rely on ICSAs (and

not on RCSAs) in order to convince us that the relevant expressions are
context sensitive. For examples consult Chapter 2. Here, though, are a few
illustrations of ICSAs well known from the literature.

When a contextualist about ‘know’ tries to convince us that knowledge

attributions are context sensitive, he appeals to intuitions we have about
ICSAs involving ‘know’ to provide evidence of context sensitivity; so, e.g.,
consider two Target Contexts, one in which the topic of conversation is
philosophical skepticism and one in which it is various issues about the
habits of birds (nothing philosophical).

Imagine an utterance of (3) in

each Target Context.

(3) Lewis knows that penguins eat ﬁsh.

Intuition is supposed to support the conclusion that the utterance of
(3) in the ﬁrst Target Context is false (because Lewis doesn’t, for example,
know how to rule out the possibility that he is a brain in a vat), while the
utterance in the second Target Context is true (since he’s fairly knowledge-

Objections to Radical Contextualism (I)

107

13 Note that this is where we ﬁnally tell you how we think Context Shifting Arguments can
be successfully put to use.
14 As Keith DeRose pointed out to us (personal communication), he prefers to use examples
in which the high standard context is nonphilosophical, e.g., a context in which the partici-
pants care very much about the evidence for the knowledge claim. The exact points we are
making here can be made about such examples by just changing the descriptions of the con-
texts appropriately. As far as we can tell, nothing at all hinges on what kinds of high or low
standard examples are chosen. We let the ‘high’ standard cases be philosophical just because
that’s the way Lewis (1996) proceeded.

able about ﬂightless water birds and their eating habits). So described this
story is an ICSA since in the Storytelling Context we are not told whether
or not ‘knows that penguins eat ﬁsh’ as used in that context applies to
Lewis.

Or, consider a Storytelling Context in which two Target Contexts are

described, one in which the topic is the heights of NBA basketball players
and the other in which the topic is the heights of Saudi Arabians. We’re
asked, again from within the Storytelling Context, to consider two distinct
utterances of (4) in these two Target Contexts.

(4) Osama bin Laden is tall.

Intuition leans towards concluding that the utterance of (4) in the ﬁrst
Target Context is false (because Osama is not tall for an NBA player), while
the utterance of (4) in the second Target Context is true (because he is tall
for a Saudi Arabian). So described this story is also an ICSA, since (4) is
never used in its Storytelling Context; it is only mentioned in describing its
uses in the two Target Contexts.

Note that these ICSAs (and others documented in Chapters 1–3)

differ from the stories we have been telling above about context shifting
with regards to members of the Basic Set, e.g., the stories surrounding
our discussion of (2a)–(2d). Reexamine those stories and you’ll note for
yourself that each constitutes an RCSA for its mentioned sentence. (2a),
e.g., tells a story in which (1) is used in an assertion in the Storytelling
Context. We used the word ‘she’ in the Storytelling Context (i.e., in this
chapter) to pick out some woman and say of her that she is French and
simultaneously we described a Target Context where the word ‘she’ got used
to pick out someone else, rendering that utterance of (1) in that Target
Context false.

In sum, for a story to be a legitimate RCSA for an expression e, it can’t

be just about utterances of a sentence S containing e; it must also be about
what S semantically expresses in the Storytelling Context. If an expression
e is genuinely context sensitive, we should be able to construct an RCSA for
e, i.e., we should be able, in a Storytelling Context, to use e in a sentence S
that semantically expresses a true proposition and simultaneously describes
a Target Context in which S is used falsely (or vice versa); thus, establish-
ing bona ﬁde context shifting. It is only through such stories that we should
be convinced that an expression e passes the ICD Test, and only such expres-
sions, as we have emphasized, are context sensitive.

All the expressions in the Basic Set pass the ICD Test and can be used to

construct legitimate and convincing RCSAs. The various contextualist can-
didates do not.

Refutation of Radical Contextualism

108

Objection 3 to RC: RC postulates context sensitivity that does not support

ICD and RCSA

As with (3) and (4), the mentioned sentences in (5a)–(5b) are context sen-
sitive only if there are true utterances of (5a)–(5b).

(5a) There are true utterances of ‘George knows that he has hands’

even though George doesn’t know he has hands.

(5b) There are true utterances of ‘Fire engines are red’ even though

ﬁre engines aren’t red.

We deny, however, there are any such utterances. We believe that, even at
this early stage of our argument, anyone who doesn’t already have theoret-
ical prejudices will ﬁnd it very hard to resist denying there are true utter-
ances of (5a)–(5b). If any utterance of ‘George knows he has hands’ is true,
George had better know he has hands; and if any utterance of ‘Fire engines
are red’ is true, ﬁre engines had better be red. These intuitions tell us that
‘know,’ ‘red,’ etc. fail the ICD Test. Compounded with the intuitive obvi-
ousness of the ICD Test for context sensitivity, this constitutes strong
prima facie evidence that these expressions are not context sensitive.

Of course, it would be boring were the entire debate reduced to a colli-

sion of intuitions: we say all utterances of (5a)–(5b) are intuitively false; our
opponents insist that they can hear some as true. How do we press forward?
Well, since it’s supposed to be news that these expressions are context sen-
sitive, anyone who thinks there can be true utterances of (5a)–(5b) needs to
bolster her case: she could try to do so by bringing us to recognize some of
these utterances as true, perhaps by getting us to reﬂect further upon the
sorts of data presented in CSAs. Thinking about the cases presented by
Cohen, DeRose, Lewis, and others might enable us to recognize there are
true utterances of ‘George knows that he has hands’ even though George
doesn’t know he has hands. Thinking about Travis’s and Searle’s examples
involving ‘weighs 80 kg’ might enable us to recognize there are true utter-
ances of ‘Smith weighs 80 kg’ even though Smith doesn’t weigh 80 kg. Etc.

Objections to Radical Contextualism (I)

109

15 To repeat why: Suppose ‘George knows he has hands’ is context sensitive, i.e., that the
proposition expressed by (and the truth conditions of) ‘George knows that he has hands’ varies
across contexts of utterance. If so, this sentence in this context (i.e., the context of this chapter)
expresses a certain proposition, and has certain truth conditions. This proposition and these
truth conditions needn’t be the same as those of its other utterances. In other words, it’s a
trivial implication of the assumption that ‘George knows that he has hands’ is context sensi-
tive that it has (at least potential) utterances that are not true just in case George knows he
has hands. At least one of these is true even though George doesn’t know he has hands, i.e.,
some utterance of (5a) (in the context of this chapter) is true.

It’s extremely telling, in this regard, that the stories presented in

defense of contextualism never take the form of an RCSA; instead, we
are given an ICSA. In what follows we will try to elicit just how hard it is,
if possible at all, to devise RCSAs for ‘know,’ ‘red,’ ‘weighs 80 kg,’ etc., that
is, a story in which these words are both used (in the appropriate way) and
mentioned. It’s crucial to keep in mind that our stories are contextualized,
i.e., understood as uttered, if you like. That’s important because if con-
textualism is true, then these RCSAs will contain a context sensitive
expression.

We call the ﬁrst alleged RCSA Known Rupert:

Known Rupert. Right now, I’m doing philosophy and thinking about
Rupert. Rupert, however, is not now doing philosophy. Instead, he’s
home making tea. Rupert doesn’t know he is 30 years old. For Rupert
to know he is 30 years old, he has to rule out the possibility that he
is a brain in a vat. Rupert, however, is unaware of (or not thinking
about) this possibility.

And so he’s ignoring a possibility that must

be ruled out in order for anyone to know anything at all. Still, when
Rupert utters in the comfort of his home, ‘I know I am 30 years old’
what he says is true, because he’s ignoring this possibility, even
though this possibility has got to be considered in order for Rupert
to know anything at all.

To see the point of the Known Rupert scenario remember that according to
contextualism ‘know’ is context sensitive, and so, its semantic value is ﬁxed
in a context of utterance. When we use ‘know’ in the Storytelling Context
(i.e., this context) to describe a Target Context, it takes on the semantic value
it has in the Storytelling Context, and not the semantic value it would have
had had it been used in the Target Context. After all, we are not in the Target
Context; we’re in the Storytelling Context using ‘know’ to describe the
Target Context. More generally, when we use ‘A knows that p’ in this context
to describe a possible world, it is the standards of this context that deter-
mine whether an object in that possible world is correctly described by that
utterance, i.e., whether A knows that p. That, by the way, is why Known
Rupert contains the modal claim that Rupert is ‘ignoring a possibility that
must be ruled out in order for anyone to know anything at all.’ Remember:
All of this follows directly from contextualism itself. If ‘know’ is context sensitive,

Refutation of Radical Contextualism

110

16 We’ve been assured by our contextualist friends that no contextualist would require
Rupert to occurrently rule out the possibility of being a brain in a vat, or to actively check it
off in any manner. All that would be required is that Rupert be disposed to handle the possi-
bility properly, perhaps by being capable of eliminating it on the basis of his evidence. We
assume nothing in our thought experiment turns on this distinction.

then Known Rupert should be true; we should have the intuition that it is
true. Our intuition is that Known Rupert is blatantly false.

Here’s the same point applied to ‘red.’

Red Rupert. In order to be red, an apple has to have red skin. That’s a
necessary condition for being a red apple. It is irrelevant, for instance,
whether an apple is red on its inside. Here’s an apple, call it Rupert;
Rupert is red. On the inside, Rupert is white. Nonetheless, there are
utterances of ‘Rupert is red’ that are false, not because Rupert’s color
changes, but because the speaker cares about what’s inside Rupert
rather than whether it is red or not. This affects the truth value of the
utterance even though the color of the inside of the apple is com-
pletely irrelevant to whether Rupert is red.

We assume you’ll agree with us that these Rupert stories do not provide
clear and convincing intuitive support for contextualism. However,
compare them with the following Now scenario (again, reading this passage
as contextualized).

Now. Right now, Stephen is not wearing a hat. Yesterday he was
wearing a hat. And when he then uttered ‘I’m wearing a hat now’ what
he said then was true, even though he’s clearly not wearing a hat now.

Objections to Radical Contextualism (I)

111

17 The point can be strengthened: Suppose someone reads Known Rupert and says: ‘Yeah, I
can still hear that as true.’ So far we have nothing but an incredulous stare to reply with. But
we could catch our breath and go further. We can ask this person: ‘Why do you think Rupert’s
utterance of “I know that I am 30 years old” is not true in this context?’ The reply, we suppose,
is going to be something like: ‘Because Rupert doesn’t know he’s not a brain in a vat.’ But now
ask: ‘What is the meaning of “know” in that reply?’ Is it what it means in this context or in
the context of utterance? It can’t be either, so our opponent is in a bind. What this shows is
that in order for a CSA to be effective, it must be motivated by descriptions that are context
insensitive. (We’ll see this point again when we discuss color words and their alleged context
sensitivity below: color terms cannot be used to describe the Target Context in a CSA that aims
to establish that such words are context sensitive, because if they are, those words would take
on the meaning they have in the Storytelling Context.)
18 Skinny Rupert. Rupert has been dieting for the last eight weeks. Rupert now weighs 80 kg!
In order to weigh 80 kg a person must weigh 80 kg on an accurate scale, naked, before break-
fast, in the morning. What he weighs with his clothes on at lunch is irrelevant. It has no bearing
on whether or not Rupert weighs 80 kg. Nonetheless, there are utterances of ‘Rupert weighs
80 kg’ that are false, not because Rupert weighs more naked before breakfast in the morning,
but because the speaker cares about what a scale would show when he steps on fully clothed
after lunch. Suppose, for example, Rupert is about to get on an elevator with a capacity of no
more than an extra 80 kg. If someone were to utter ‘Rupert weighs 80 kg’ her utterance would
be false, even though he weighs 80 kg. The utterance would be false, not because Rupert’s
weight has changed, but because the speaker is concerned with something other than what
Rupert weighs, for example, with what a scale registers were he to step on it fully clothed.

The Rupert scenarios are unconvincing to us, and compare quite unfavor-
ably to the Now scenario. Unlike the Rupert scenarios, we take Now to be
an RCSA that provides evidence for the view that ‘now’ is context sensitive.
Yet there is nothing cagey about the Rupert stories; they parallel exactly the
Now one. If you don’t like them, or think that they are prejudicially slanted,
try devising one of your own.

A point worth reemphasizing is that anyone who wants to defend the

context sensitivity of an expression e while conceding that RCSA cannot be
devised for it has placed himself in an unfortunate position. Anyone who
is in this position is asking us to take it as an article of faith that the expres-
sion in question is context sensitive. Though faith has its place, we don’t
believe that place is in the philosophy of language.

Penultimate Point: It would be Surprising if there were

Surprising Context Sensitivity

This penultimate point is a bit vague, but it’s important. If expression e is
context sensitive, then it’s obviously context sensitive. Speakers should not
have to theorize about it in order to realize that it is context sensitive.

We remind you of some basic facts about communication: Conversa-

tions happen fast. Someone speaks; sounds hit the audience’s eardrums;
they must be processed; often a reply is expected immediately. There’s little
time for reﬂection and exploration. It is because our linguistic devices are
so effective for communication that conversation is able to be as ﬂuid as it
is. They are easy to use and it is not surprising that they are easy to use.

To use the dubious metaphor of language as a tool for a moment: if

words are tools, then they had better be pretty easy to use because they don’t
come to us with instruction manuals and even if they did, there would be
no time for us to consult these instruction manuals when we’re steeped in
the middle of a fast and furious conversation.

We’re highlighting these obvious features of communication in order to

register a very simple point: If an expression e has its semantic value ﬁxed
in a context of utterance, that had better be obvious to all of us. Context
sensitivity can’t be some obscure phenomenon that you need to read schol-
arly books and articles about in order to recognize and master. Context
sensitivity is a surface phenomenon. Every speaker knows it when he’s
confronted with it; and he knows that every other competent speaker of his
language knows it as well, and all speakers know how to exploit context sen-
sitivity in the heat of a conversation. None of this should come as a sur-
prise. If you ask a speaker of some particular expression whether or not it’s
context sensitive, she should be able to tell you right away that it is, and
how it is, context sensitive.

Refutation of Radical Contextualism

112

Every expression in the Basic Set is obviously context sensitive. No one

needs a theory to ﬁgure this out. You can gift wrap it in fancy language and
appeal to all of the philosophical tests, as we did above, but you don’t have
to do that to convince yourself of the context sensitivity of any single
member of the Basic Set. We all know that the referent of an utterance of
‘yesterday’ depends on the day it’s uttered on. We all know that the refer-
ent of ‘I’ is always its user; and so on for each member of the Basic Set. But
compare that to the idea that ‘blue’ is context sensitive; that not every single
thing it applies to is blue! Any nonphilosopher apprised of this philo-
sophical ‘discovery’ would, and should, scoff at it. That reaction itself is
philosophically signiﬁcant and presents more than just a prima facie case
against the alleged discovery.

Final Point: Caution

We end this chapter with a cautionary note to be elaborated upon in Part
III of the book: The tests in this chapter obviously have to focus on what’s
communicated; we are after all communicating with the reader, and so the
tests are tests that go via communicated content. Our view, as presented in
Part III, is that there is a sharp distinction between communicated content
and semantic content. Nonetheless, we use communicated content (the
content we succeed in communicating to our readers) to ‘get at’ semantic
content (semantic context sensitivity in particular). There is of course no
other way to proceed. The purpose of the tests is to generate contexts in
which semantic content is salient. Think of the tests like this: They are ways
to get the audience to notice semantic features of sentences uttered. They
create contexts in which our attention is drawn to features of the semantic
content expressed by the utterances in question. These issues are addressed
further in Chapter 13.

A P P E N D I X 1. C O N T E X T UA L

SA L I E N C E A B S O R P T I O N

Here is a reply that we guess some Radical Contextualists might devise in
response to the ICD/RCSA Tests and objection in this chapter.

Since we’ve

never seen it in print, we’re not sure exactly how to put it, but we imagine

Objections to Radical Contextualism (I)

113

19 Participants at a Rutgers Graduate Seminar (fall, 2003) unanimously claimed that no sen-
sible Radical Contextualist would appeal to Contextual Salience Absorption. Hence, the
appendix status of this reply.

it going something like this: Schematically, the situation we theorists
ﬁnd ourselves in is this: We’re in context 5stC. Call this the Storytelling
Context. In this context, we are imagining two Target Contexts, C1 and C2,
in which someone utters ‘John is ready.’ We ask ourselves what these two
utterances say. In order to determine an answer to this question, we vividly
imagine these Target Contexts C1 and C2 from within our Storytelling
Context 5stC. We imaginatively place ourselves ﬁrst in Target Context
C1 and then in C2. To imaginatively place ourselves in C1 and C2
automatically triggers what we shall call Contextual Salience Absorption. While
in 5stC, we are to imagine ourselves in C1; doing so, however, requires
that whatever is salient in C1 becomes salient to us in 5stC. But then when
we try to imagine ourselves in the other Target Context C2, what’s salient
there becomes salient to us in 5stC. So, what’s contextually salient in 5stC
varies contingent on what we happen to be thinking about in 5stC. So far
so good. But why is Contextual Salience Absorption relevant to the
ICD/RCSA Test?

It is because it provides the Radical Contextualist with an RC-

compatible explanation for why native speakers intuitively ﬁnd the relevant
RCSAs problematic (e.g., why they ﬁnd the Known Rupert story counter-
intuitive). In these cases, the theorist is asked to vividly imagine a Target
Context from a Storytelling Context. If imagining a context, say, C1, from
the perspective of another context, say, 5stC, renders what is salient in C1
salient in 5stC, then that undermines the idea that separate contextual
standards are in play. The test for context sensitivity assumes that the con-
textually salient standards are clearly distinct. To say that there’s Contex-
tual Salience Absorption is to say that the standards are blurred. That
would make the test unreliable. To put it pointedly: As soon as you think
about a context C, the standards of C affect the standards of the context
you’re in (the context in which you do your thinking). Just thinking about
C changes the context you are in so that there is no longer a clear distinc-
tion in contextual standards between the Storytelling Context and the
Target Context. If this is so, our test is faulty: it presupposes that there
is a rather sharp distinction between the Storytelling and Target Contexts
standards.

We have left this reply rather vague in part because we’ve never seen it

worked out in detail and there are many ways to do that. But, we think, no
matter how it’s elaborated, this response on behalf of RC fails miserably for
(at least) two reasons:

1 The reply gets the dialectic deeply wrong and leaves the Radical

Contextualist without an argument against Semantic Minimalism.

2 An appeal to Contextual Salience Absorption triggers what we

believe ultimately leads to an internal inconsistency in RC.

Refutation of Radical Contextualism

114

1. Appeal to Contextual Salience Absorption leaves RC

dialectically impotent

Let’s review the dialectical situation: A Semantic Minimalist about an
expression e denies there is contextual variation in the semantic content of
utterances of sentences containing e (other than whatever is contributed by
expressions other than e). The contribution of e to the proposition seman-
tically expressed by utterances of sentences containing e is invariant across
all contexts of utterance. One way to capture this idea (not the only way,
but it captures a common core of many versions of Semantic Minimalism)
is this:

If S is a sentence containing e (and no other indexical components),
then every utterance of an instance of schemas (I) and (II) should be
true:

(I) ‘S’ is true just in case S.

(II) ‘S’ expresses the proposition that S.

Context Shifting Arguments are supposed to convince Semantic Minimal-
ists that their position is ﬂawed. The evidence takes the form of a coun-
terexample: intuitive evidence that there is at least one utterance u of S that
semantically expresses a different proposition or has different truth condi-
tions (maybe even a different truth value with respect to the same circum-
stance of evaluation) than another utterance of S in the Storytelling
Context. But to endorse Contextual Salience Absorption is in effect to grant
that no such intuitive counterexample is forthcoming.

Whenever in a Storytelling Context we think about or describe a Target

Context, our intuitions about the e-standards of the two contexts become
uniﬁed or get blurred. To get a clear counterexample of this kind the
Radical Contextualist presupposes clearly distinct standards. In sum:
Appeal to Contextual Salience Absorption is not a defense of contextual-
ism; rather, it’s a concession of defeat. For the argument in defense of RC
was based entirely on intuitions about context shifting; Contextual Salience
Absorption, however, is invoked to explain why such intuitions are not
forthcoming.

2. Contextual Salience Absorption renders

contextualism internally inconsistent

According to RC, every expression is context sensitive. So, if RC is true, then
distinct uses of ‘ready’ should induce intuitions about context sensitivity.

Objections to Radical Contextualism (I)

115

Here’s a corollary of point (A): If we assume for the sake of argument that
a sentence S is context sensitive, and if the context sensitive components of S
absorb, then there’s overwhelming evidence in favor of the truth of (D1) and
(D2):

(D1) Every utterance of ‘A is ready’ is true iff A is ready.
(D2) Every utterance of ‘A is ready’ expresses the proposition that A

is ready.

Remember, if ‘A is ready’ is absorbent, it follows that every utterance u of
‘A is ready’ can be characterized truly by an utterance of (D1.1) and (D2.1).

(D1.1) u is true iff A is ready.
(D2.1) u expresses the proposition that A is ready.

This follows because if there is Contextual Salience Absorption, then every
utterance of (D1.1) and (D2.1) is true. This is because if ‘A is ready’ is
absorbent, then the truth conditions for ‘A is ready’ are the same in the Sto-
rytelling Context as they are in the Target Context, where ‘A is ready’ is sup-
posed to be imagined to be true. The truth of (D1.1) and (D2.1) provides
overwhelming evidence in favor of the generalizations (D1) and (D2). But
they are the denial of RC. In other words, a Radical Contextualist who
claims that sentence S is absorbent has an incoherent view.

A P P E N D I X 2 . D I AG N O S I S : M O N S T E R S

A N D U S E – M E N T I O N FA L L AC I E S

If we are right, then ‘know’ and ‘red’ fail obvious tests for context sensitiv-
ity. We think the same applies to every other example mentioned in Chapter
2. So what’s going on? How do philosophers convince themselves that obvi-
ously noncontext sensitive expressions are context sensitive?

These positions are typically justiﬁed by an appeal to Impoverished

Context Sensitive Arguments. That raises the question: If you have an
Impoverished Context Shifting Argument for an expression e, but e does
not pass other tests for context sensitivity (including the tests discussed
here), then what’s going on?

First of all, you have strong evidence that whatever intuitions were trig-

gered by the ICSAs concern nonsemantic content. They are intuitions that
reveal what the utterance might succeed in communicating, but not its
semantic content. But there’s more than that going on. We have two ten-
tative diagnoses of why contextualists are so easily seduced by their ICSAs.

Refutation of Radical Contextualism

116

First diagnosis: Treating ‘in context C’ as a monster

Contextualists talk quite freely about knowing that p in context C and not
knowing that p in context C

¢, or being rich in C but not being rich in C¢,

etc., as if the semantic role of ‘in context C’ were to map S (the sentence to
which it attaches) from a context of utterance onto a distinct context of
utterance. Unwittingly, we presume, in this regard they are treating ‘in
context C’ as what David Kaplan has called a monster. A monster is an oper-
ator ‘which when preﬁxed to a sentence yields a truth if and only if in some
contexts the contained sentence (not the content expressed by it) expresses
a content that is true in the circumstance of that context’ (Kaplan 1989a,
p. 510). If ‘in context C’ were a monster, it could transport us from our
current context of utterance to imagined ones. It would allow us to use an
expression in this context, within the scope of the monster, to say what that
expression would say if it were uttered in another context. This, so it seems
to us, is what often transpires when contextualists try to devise (fake) RCSA.
And, of course, if English had monsters and ‘in context C’ were one of them,
then most of our objections would be undermined. For example, you would
be able to use ‘know’ or ‘red’ in a Storytelling Context C when describing a
Target Context, but they would not take on the semantic values they have
in C. If there were monsters and contextualists could use them, the effect
would be this: the theorist could keep a sharp distinction between the stan-
dards of the Storytelling Context and the Target Context, and could
describe what goes on in the Target Context by using, say, ‘know,’ either
outside a monster or inside a monster. With this device in hand, she could
say one thing within the monster and another outside it. Playing on this
discrepancy she gets herself to believe she has something like an RCSA
rather than the obviously unsatisfying ICSA.

Contextualists certainly write as if they can use ‘know’ or ‘red’ in this

way. But you could succeed with these uses only if there were monsters, and,
as Kaplan has pointed out, ‘in context C’ is not a monster in English. He
writes:

Let us try it:

(8) In some contexts it is true that I am not tired now.

For (8) to be true in the present context it sufﬁces that some agent of
some context not be tired at the time of that context. (8), so inter-
preted, has nothing to do with me or the present moment. But this
violates principle 2!

Principle 2 can also be expressed in a more

Objections to Radical Contextualism (I)

117

20 i.e., the thesis that indexicals pick out their referents directly from the context of utter-
ance, without mediation. This means that the value of an indexical is ﬁxed by the context of
its utterance, and cannot be changed by the logical operators in whose scope it may occur.

theory-laden way by saying that indexicals always take primary scope.
If this is true – and it is – then no operator can control the character of the
indexical within its scope, because they will simply leap out of its scope to the
front of the operator. I am not saying we could not construct a language with
such operators, just that English is not one. And such operators could not be
added to it. (Kaplan 1989a, p. 510; emphasis our own)

Indirectly, we have been championing Kaplan’s view. Our ﬁrst diagnosis of
what’s gone wrong with contextualism is that in convincing themselves that
the expressions in question are context sensitive, the locution ‘in context C’
winds up being treated as a monster. A case in point is Graff ’s discussion
of color words:

Suppose I want you to hand me a certain book. If the book in ques-
tion is colored a very light grayish-blue, and it’s sitting amongst a
bunch of other books all of which are colored a very light grayish-red,
I may say, ‘Hand me the blue one.’ If, on the other hand, the book I
want is sitting with a bunch of richly-colored cobalt blue books, I may
say, ‘Hand me the gray one.’ I take it that it would be true to say in
the ﬁrst case that the book I wanted was blue, and in the second case
that the book I wanted was gray. (Graff 2000, p. 56)

Focus carefully on the last sentence of her story: ‘I take it that it would be
true to say in the ﬁrst case that the book I wanted was blue, and in the
second case that the book I wanted was gray.’ Call its tokening u and the
context of u the Storytelling Context. The color words ‘blue’ and ‘gray’ are
used there in an indirect quote that occurs in the Storytelling Context.

Context sensitive expressions in indirect quotes receive the semantic

value of the context of the indirect quote itself. If we say, in the context of
this chapter, that Jason said our view is crazy, ‘our’ refers to Cappelen and
Lepore, not to Jason and company. Indirect quotation is no monster. In
other words, Graff ’s use of ‘blue’ and ‘gray’ get their semantic values from
the Storytelling Context, and not from Target Contexts. But then u makes
no sense. Graff ’s point is that the semantic values of ‘blue’ and ‘gray’ shift
between what she calls the ﬁrst and the second case (we call these the ﬁrst
and second Target Contexts). That wouldn’t work were both given their

Refutation of Radical Contextualism

118

21 Lewis, though he doesn’t use the terminology, is acknowledging the same point when he
writes: ‘To be sure, we could speak a language in which “As for you, I am hungry” is true iff “I
am hungry” is true when the role of speaker is shifted from me to you – in other words, iff
you are hungry. We could but we don’t. For English, the speaker is not a shiftable feature of
context’ (Lewis 1980, pp. 27–8; cf. also Evans 1985, pp. 357–8).

interpretations in the Storytelling Context. She wants ‘in the ﬁrst case’ to
create a context such that what occurs after it is interpreted as uttered in
the ﬁrst Target Context. She wants what occurs after ‘in the second case’ to
be interpreted as uttered in the second Target Context. On this interpreta-
tion, she is treating ‘in the ﬁrst (second, etc.) case’ as monsters.

But again,

there are no monsters in English. When someone says to you, ‘In case one,
I’m not happy,’ there is no context in which his use of ‘I’ picks out you –
which it should be able to do were ‘in case one’ a monster.

For a second illustration, consider a passage from Soames (1999), where

he’s endorsing the view that ‘looks green’ is context sensitive. Here’s how
he describes the context sensitivity:

In this model the rule governing contextual adjustment of the pred-
icate looks green should be roughly as follows:

If in a context C something x is explicitly judged to look green,
then the extension of looks green in C includes everything percep-
tually indistinguishable in color from x, as well as everything that
looks greener than x. If in C something y is judged not to look
green, then the anti-extension of looks green in C includes every-
thing perceptually indistinguishable in color from y, as well as
everything that looks less green than y.

(Soames 1999, pp. 211–12)

This entire statement is uttered in the context of Soames’s book. Call it the
SB-Context (for Soames’s Book Context). Focus on the ﬁrst conditional and
the ﬁrst use of ‘to look green’ in that conditional. Call this utterance of ‘to
look green’ u (so u is preceded by an utterance of ‘If in a context C some-
thing x is explicitly judged . . .’). Now ask yourself: What’s the semantic
value of u? The expression ‘to look green’ is, according to Soames, context
sensitive, so its semantic value is determined by its context of utterance. The
context of utterance in this case is the SB-Context. So, u should get its
semantic value from SB. But that’s clearly not what Soames intends. For
then he would end up talking only about contexts in which someone
uttered (or made a judgment) about things that fall under ‘looks green’
according to the standards of the SB-Context. What Soames wants is for u

Objections to Radical Contextualism (I)

119

22 Alternatively, she treats ‘what was said’ as a monster. The same point about monsters
extends to DeRose, when he writes: ‘In Bank Case B . . . when, in the face of my wife’s doubt,
I admit that I don’t know that the bank will be open on Saturday, I don’t contradict an earlier
claim to know that I might have made before the doubt was raised and before the issue was
so important because, in an important sense, I don’t mean the same thing by “know” as I
meant in the earlier claim’ (DeRose 1992, p. 921).

to take its semantic value from the imagined context C. He wants u to have
the semantic value it would have had if it had been uttered in this imag-
ined context. That’s why he puts ‘If in a context C something x is explicitly
judged . . .’ in front of u. In other words, he is treating ‘in a context C’ as a
monster.

But, again, there are no monsters in English; and surely, it

should be obvious that ‘in a context C’ is not one.

For a third illustration, consider this story from Cohen:

Mary and John are at the LA airport contemplating taking a certain
ﬂight to New York. They want to know whether the ﬂight has a layover
in Chicago. They overhear someone ask if anyone knows whether the
ﬂight makes any stops. A passenger Smith replies ‘I do. I just looked
at my ﬂight itinerary and there is a stop in Chicago.’ It turns out that
Mary and John have a very important business contact that they have
to make at the Chicago airport. Mary says: ‘How reliable is that itin-
erary, anyway? It could contain a misprint. They could have changed
the schedule since it was printed, etc.’ Mary and John agree that Smith
doesn’t really know that the plane will stop in Chicago on the basis
of the itinerary . . . (Cohen 1999, pp. 58–9)

Cohen presents his thought experiment in the Storytelling Context. Con-
sider the utterance u of ‘They overhear someone ask if anyone knows
whether the ﬂight makes any stops.’ Since u is uttered in the Storytelling
Context, if contextualism is true, ‘know’ in u takes on a semantic value in
that context. Suppose the proposition expressed by u in the Storytelling
Context is p. Now consider the direct quote of Smith that directly follows
u (where Smith is quoted as saying ‘I do’). Call the utterance of this direct
quote u

¢. Cohen clearly thinks u¢ indicates that Smith agrees with the

semantic content of u. But it’s a mystery how it could. u is uttered in the
Storytelling Context, the context of a philosophy paper, and, again assum-

Refutation of Radical Contextualism

120

23 Note also the following passage from Soames: ‘What, then, is going on? The answer, in
our analysis, is that there is no contradiction here at all. According to the analysis, when x

initially characterized as looking green, this is done with respect to a certain set of standards,
S. Later, when it is characterized as not looking green, this is done with respect to a new set
of standards, S*. But there is no contradiction in the observation that something may look
green with respect to one set of standards and not look green with respect to a different set
of standards’ (Soames 1999, p. 313). Note the occurrence of ‘looking green’ in ‘when x is char-
acterized as looking green.’ Here ‘characterized as’ is being used as a monster.
24 ‘In context C, I am speaking’ picks out me no matter who is speaking in context C. Why
should it be any different for ‘looks green’?

Schlenker (2003) argues that Kaplan is wrong: English does contain monsters. We choose

not to engage in a full discussion of that issue here because even if the examples Schlenker
presents are genuine examples of monsters (something we doubt), they are not the kinds of
monsters that would help contextualists.

Objections to Radical Contextualism (I)

121

ing contextualism, the fact that p is expressed by u is due to peculiar fea-
tures of that context. Smith’s utterance u

¢, however, is made in the Target

Context, a context with low epistemic standards. The epistemic standards
for ‘know’ are ﬁxed by the kind of considerations Cohen and other philoso-
phers are concerned with. Since Smith is not in that sort of context, how
could he end up agreeing with p? It’s all very peculiar. What’s going on?
One possibility is that Cohen wants us to read u as if it was uttered in the
Target Context. To do so would be to treat the act of storytelling as tacitly
introducing a monster.

Our own view about ‘in a context C’ is that it has no impact on the

semantic value of the controversial cases. If an object is round in Room 300,
it’s round. If someone knows in context C what penguins eat, then he
knows what penguins eat.

Second diagnosis: How to avoid monsters: direct quotation

Aren’t we just being pedantic? Couldn’t a contextualist easily avoid these
troubles just by being more careful? In particular, couldn’t they avoid mon-
sters simply by using direct quotation instead? That’s probably what they
think and it’s probably why they allow themselves to be so sloppy. Here’s
David Kaplan on the connection between monsters and direct quotation:

There is a way to control an indexical, to keep it from taking primary
scope, and even to refer it to another context. Use quotation marks.
If we mention the indexical rather than use it, we can, of course,
operate directly on it. Carnap once pointed out to me how important
the difference between direct and indirect quotation is in

Otto said ‘I am a fool’.
Otto said that I am a fool.

(Kaplan 1989a, p. 511)

Does direct quotation provide an easy way out for the monstrous con-

textualist? No. If in order to avoid monsters a contextualist resorts to direct
quotation, then she would be left with nothing more than intuitions about
an ICSA to bolster her case. But as we have been arguing, those intuitions
alone are insufﬁcient to establish contextualism unless a corresponding

25 The continuation is equally puzzling. At the end of Cohen’s story he says, ‘Mary and John
agree that Smith doesn’t really know that the plane will stop in Chicago on the basis of the
itinerary.’ The ‘know’ in that utterance should take on its semantic value in the Storytelling
Context, and it’s peculiar how Mary and John end up agreeing with an utterance in the
Storytelling Context.

intuitive RCSA can be devised. But an RCSA requires using the disputed
expression in an assertion (or a denial). It is tempting to do that in a mon-
strous way (in which case the contextualist would have her RCSA), but there
are no monsters! The way to avoid monsters is to resort to direct quota-
tion, but then the contextualist is back to an ICSA.

Cautionary note to be elaborated upon in Part III of the book. The tests in this
chapter obviously have to focus on what’s communicated; we are after all
communicating with the reader, and so the tests are tests that go via com-
municated content. That’s not a problem if we ensure that the context is
one where semantic content is being focused on. Think of the tests like this:
They are ways to get the audience to notice semantic features of sentences
uttered. They create contexts in which our attention is drawn to features of
the semantic content expressed by the utterances in question.

Refutation of Radical Contextualism

122

C H A P T E R 8

Objections to Radical

Contextualism (II): Makes

Communication Impossible

123

The objection we raise in this chapter is certainly not original with us; its
variations go back at least to Frege. The simple idea is this: If RC were true,
it would be miraculous if people ever succeeded in communicating across
diverse contexts of utterance. But there are no miracles; people do succeed
in communicating across diverse contexts of utterance with boring regu-
larity. So, RC isn’t true.

Only slightly more elaborated, it goes like this: If RC were true, then

what’s said by an utterance by a speaker A in context of utterance C
depends, at least in part, on very speciﬁc features of C. Here’s a helpful list
from Bezuidenhout (all Radical (and Moderate) Contextualists hold similar
views) summarizing various contextual features that can determine what is
said:

(i) knowledge that has already been activated from the prior dis-

course context (if any)

(ii) knowledge that is available based on who one’s conversational

partner is and on what community memberships one shares with
that person

(iii) knowledge that is available through observation of the mutual

perceptual environment

(iv) any stereotypical knowledge or scripts or frames that are asso-

ciatively triggered by accessing the semantic potential of any of
the expressions currently being used

(v) knowledge of the purposes and abilities of one’s conversational

partner (e.g., whether the person is being deceitful or sincere,
whether the person tends to verbosity or is a person of few words,
etc.)

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

(vi) knowledge one has of the general principles governing con-

versational exchanges (perhaps including Grice’s conversa-
tional maxims, culturally speciﬁc norms of politeness, etc.).
(Bezuidenhout 2002, p. 117)

Focusing on (i)–(iii) from above, notice that:

(a) If RC were true, then for us to understand what you said by an utter-

ance of ‘Philosophy is fun’ we would have to know what knowledge has
been triggered by previous conversations you have been engaged in, we have
to know whom you are talking to, what you know about them, what knowl-
edge you can assume is shared between you and your audience, the nature
of your mutually shared perceptual environment, and so on and so on.
That’s a lot of requirements just to ﬁgure out what you said. When the full
RC story is told, it will turn out to be a miracle every time anyone manages
to ﬁgure out what someone had said. But there are no miracles. People do
not need to access all of this knowledge in order to ﬁgure out what has been
said. So, RC is false.

(b) The situation, as hard as it may be to believe though, is even worse.

Not only would we have to gather all this information in order to ﬁgure
out what you said with your utterance. According to RC, the content
communicated, the proposition semantically expressed, the explicature
(in Relevance Theory terminology), is generated through a complex and
interesting psychological process. The speaker and audience ﬁx on a
common proposition by virtue of the fact that they have the knowledge
listed in (i)–(vi). So, suppose we know that a speaker A and his audience
share a range of beliefs, B

. . . B

. Of course, we might not believe B

. . . B

even if we believe they (i.e., A and his audience) have them. We might think
they’re wrong. But then, if RC is right, it would again be miraculous if we
managed to ﬁx in our minds the same proposition that they centered on.

In sum: If RC were true, it would be a miracle if speakers in different con-
texts were ever able to agree, disagree, or more generally, share contents.

Reply: Bezuidenhout on miracles. As far as we know, Bezuidenhout is the only
Radical Contextualist who addresses this (or at least a related) worry. She
discusses Recanati’s view according to which

words are not primitively associated with conditions of application,
but with a certain semantic potential. Sense or content must be con-
structed in context out of such a semantic potential. Semantic poten-
tial is deﬁned in terms of a set of source-situations. A source-situation
is a situation that a speaker has learned to associate with a term

Refutation of Radical Contextualism

124

because during learning the speaker has seen the term legitimately
applied to that situation. (Bezuidenhout 2002, p. 116)

She’s sympathetic to this view, but worries that

it seems to relativize semantic knowledge to a speaker. Source-
situations are relativized to the learning histories of particular
speakers. This entails that the semantic potential of a term will differ
from speaker to speaker. One might think that communication
between speakers is thus jeopardized, as speakers will not be using
terms with the same meanings. (Bezuidenhout 2002, p. 118)

Here’s her response to the worry:

But of course the whole idea of this new picture is that terms don’t
have ﬁxed meanings, but only semantic potentials, and that sense/
content must be constructed in context. If one thinks of this sense
creation as a joint action of the conversational partners, one might be
able to avoid both the relativism and the apparent consequence that
communication is jeopardized. Even though the semantic potential
of ‘playing baseball’ differs for you and for me, we may in a particu-
lar conversational setting be able to converge on the same contextual
understanding for that term as it is used in that setting. This
needn’t be too much of a miracle if we assume that human beings, in
virtue of their common species heritage, are apt to perceive and reason
about the world in ways similar to one another. (Bezuidenhout 2002,
p. 118)

We applaud her for acknowledging the dangers lurking here, but she fails
to see the full reach of the problem this and related views face; her reply to
this more restricted worry serves to highlight the remaining problem.

The way she construes it, the problem is how people who ﬁnd themselves

in a shared context can communicate. She thinks (hopes) that this problem
dissolves if we ‘assume that human beings, in virtue of their common
species heritage, are apt to perceive and reason about the world in ways
similar to one another.’ But where does that leave those of us not in that
context; those unfamiliar with the reasoning the participants have under-
gone; those unfamiliar with what they perceive?

Reply 1. Similarity of content is sufﬁcient for successful communication

Suppose the reply is: ‘Of course, we’ll never be able to say exactly what
someone else said; we’re never able to match the content exactly. But who

Objections to Radical Contextualism (II)

125

cares? That’s not needed for communication to proceed smoothly. All we
need is sufﬁciently similar content. If what A says in context C is sufﬁciently
similar to what B says in context C

¢, then that’s enough.’

Our reply

1 Enough for what? Not enough to account for the intuition that we

have said the very same thing. We can say what you said. You can say what
we say; not just something similar to what we said. It is a fact about ordi-
nary speakers that they deeply believe this. It is reﬂected in practically every
aspect of our linguistic practice. Of course, we might discover that our con-
ception of communication is thoroughly mistaken, some sort of bad folk-
theory of language. The Radical Contextualist has discovered that we have
been confused all of these years.

2 Similarity of content is undermined at least in this sense: We can’t say

what it is for two utterances to be sufﬁciently similar in content; that the
propositions they express are close enough; that their truth conditions are
virtually alike. We doubt any such notion will stand up to scrutiny (cf.,
Fodor and Lepore 1992, 1999). But even if it did, speakers surely don’t
behave as if they are getting by with nothing more than similarity. We often,
indeed, want to know exactly what someone said; what he literally said;
what he really said. Is RC committed to the view that such aspirations must
be dashed?

Reply 2. Biting the bullet

A Radical Contextualist might, of course, just bite the bullet and say some-
thing like: ‘Look: There’s no way to secure successful communication across
contexts. When people ﬁnd themselves with different background assump-
tions, different audiences, different perceptual conditions, it is difﬁcult,
very difﬁcult indeed, to latch on to the exact same propositions. Is that such
a big deal?’

Our response is not particularly original. First of all, it’s not a bullet we

would like to bite. Second, and more importantly, if we were to take this
view seriously, we don’t even know which bullet we are being asked to bite,
and our opponents don’t even know which bullet we are asking them to
bite, and vice versa. We know that much about RC is written in places other
than New York and New Jersey, lots of it is being written and discussed all
over in Europe, some of it in the Midwest, some in California; most of it by
people whose background assumptions, lifestyles, audiences, and percep-
tual inputs are radically different from our own and from each other. So, if
we have understood their bullet biting correctly (which, of course, it would
be on their own account a miracle if we did), it would also be miraculous

Refutation of Radical Contextualism

126

if we’ve understood any of all these RC writings we’ve been reading lately
(and it would be a miracle if they understood anything of what we’re saying
in response).

Of course, what we really think is this: There are no miracles and we do

understand all these papers and books we’ve been reading about RC (well,
most of them). So, we conclude that RC is wrong.

Perhaps another illustration will help you to see our point if you haven’t

already: Suppose right now, Cappelen is about to say something; he’s about
to do it sitting with Lepore in the café on 5th Street between Avenues A and
B in Manhattan (i.e., you are about to witness another live test, not a quote,
but a real utterance, made in 5stC):

Napoleon was an interesting character.

Call this the Napoleon Speech Act. We think the following is obviously true:
Other people in other contexts have said, could have said, and will say
exactly what Cappelen said with the Napoleon Speech Act. This would be
miraculous if RC were true. Those features which supposedly are constitu-
tive of the content of (the what is said by) the Napoleon-Utterance are not
shared by other contexts of utterance. The only way for it to be true that
others have said what was said by the Napoleon-Utterance is if the speciﬁcs
of the context of utterance are irrelevant to content determination. (We
have much more to say about contextualism and communication in our
(2004).)

Objections to Radical Contextualism (II)

127

128

C H A P T E R 9

Objections to Radical Contextualism

(III): Internal Inconsistency

Some biographical background: whenever we read a presentation of, and
arguments for, RC we wind up screaming at each other: ‘They can’t even
say what they believe! They break their own rules in the very formulations
of their views.’ Initially, we thought the offending formulations were just
the result of insigniﬁcant sloppiness or a legitimate simplifying device. But
we no longer think that. In this chapter we show it’s intrinsic to RC that its
presentations be internally inconsistent. To interpret the sentences that
express RC you have to assume RC is not true.

First Inconsistency Charge: RC Implies there are True

Utterances of ‘RC is False’

RC is the view that all sentences are context sensitive. If so, then the propo-
sitions expressed by utterances of an arbitrary sentence S change depend-
ing on its contexts of utterance. The truth conditions for an utterance of
an arbitrary sentence S vary between contexts of utterance. Let’s apply this
view to sentences used to express and defend RC, i.e., consider a sentence
such as:

(1) RC is true.

If RC is true, then (1) expresses different propositions depending on its con-
texts of utterance.

Let’s consider a context in which an utterance of (1) doesn’t express

a proposition Radical Contextualists can accept, i.e., consider contexts
relative to which an utterance of (1) expresses a proposition that’s false

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

according to RC. (If you think there’s no such context, see the imagined
replies below; for now we’ll assume there’s such a context.)

It is not entirely easy to engage in the thought experiment we’re propos-

ing. The central complication is this: For this thought experiment to work,
you have to assume that the context we (i.e., C&L) are writing in is one
in which ‘true’ expresses the property that Radical Contextualists think
applies to RC. That is, we have to assume that ‘true’ in this context expresses
a property that Radical Contextualists think applies to RC.

(We have to

admit we have no idea what it even means to make that assumption, but it
seems to follow from RC that we can make it, and since this is an internal
objection we’re entitled to it.) From this context we are imagining a distinct
context C in which an utterance u of ‘RC is false’ is true.

Keep in mind what this last sentence means according to Radical Con-

textualism: it means that u has the property that ‘true’ picks out as used in
this context. Of course, for that to be so we have to assume ‘false’ as used
in C doesn’t pick out what ‘false’ expresses in this context. We’ll make use
of the following simplifying device (ultimately not an acceptable simplify-
ing device, but put that aside for now): Let’s name the semantic value of
‘true’ in this context true and the semantic value of ‘true’ in C true* (and
correspondingly for ‘false’ we introduce false and false*).

We expect the Radical Contextualist to agree that there might be an

utterance u of ‘RC is false’ that’s true, but not worry about it. After all, u
doesn’t deny RC is true. It denies RC has some other property, namely, being
true*. That is, all they have granted is that RC is false* and that’s not the
same as being false.

Here is where our objection kicks in. Why doesn’t the Radical Contex-

tualist worry about being false*? What’s so special about being true? Why is
that more important than being true*? Why just be concerned about avoid-
ing falsehood and not falsehood*? Why privilege the property that ‘true’ picks
out in this context? It is as if the Radical Contextualist has arbitrarily chosen
to care intensely about what ‘true’ picks out in this context, and not care
about what it picks out in other contexts. We’re seeking a justiﬁcation for
that.

Think of our objection as a challenge. We don’t really know what Radical

Contextualists think about ‘true’ and other words such as ‘content,’
‘express,’ ‘context,’ etc., that are central to the articulation of RC. Consider
an analogy with ‘red.’

According to Radical Contextualists, there are many, many ways to be

red. Let’s suppose something analogous is true about ‘true’: there are many
ways to be true. Of course, the Radical Contextualist doesn’t think one of

Objections to Radical Contextualism (III)

129

1 We also have to assume that ‘RC’ expresses (cross-contextually) the view they would like to
defend, but we’ll leave that complication aside for the moment.

these ways of being red captures redness any more than the others (that’s
the whole point). Why should it be any different for ‘true’?

The Radical Contextualist might make any one of three replies:

1 She might argue that ‘true’ and other central semantic terms are

stable, that they don’t exhibit the kind of context sensitivity that all other
expressions do. They have the same semantic value in all contexts of utter-
ance. Of course, for this strategy to be workable, the Radical Contextualist
has to present us with convincing arguments to the effect that these terms
are different from all other terms. We know of no such argument, but we’re
all ears.

2 She might argue that ‘true’ picks out different properties in different

contexts of utterance, but that whichever property is picked out, it applies
to RC. Whichever property is picked out by ‘true,’ it is true to utter ‘RC is
true.’ To show that this is so would be a rather extraordinary feat. We just
don’t have any idea how the Radical Contextualist would go about it. In
particular, since the problem recurs at the meta-level: Take utterances of
‘ “RC is true” expresses a truth when uttered in any context of utterance.’
The Radical Contextualist would have to show that this sentence also
expresses a truth in all contexts of utterance. As far as we can tell, the only
way to do this is to opt for (1). We might be wrong, so again, we’ll just point
out that no Radical Contextualist has ever shown that this strategy can be
implemented. We’re all ears.

3 Finally, she might argue that what ‘true’ picks out in this context is

more important than what ‘true’ picks out in other contexts. There’s some-
thing special about true and false, and this allows them to ignore (or not
worry about) the true utterances of ‘RC is false.’ On this view, a Radical Con-
textualist has no reason to worry about RC being false*, or any of the other
properties picked out by ‘false.’ Of course, if this is correct, it’s extraordi-
narily important to place yourself in the right kind of context in order to
do semantics. If you’re unfortunate enough to ﬁnd yourself in a context in
which ‘true’ picks out a property you don’t care about, one that’s not sig-
niﬁcant, then any kind of discourse (in that context) about what in those
contexts are called ‘truth values’ is pointless. Again, we ultimately do not
know how to evaluate this suggestion before we see it worked out in detail.
And again, we’re all ears.

Preliminaries for Second and Third Inconsistency

Objections

Some preliminaries for the next two objections.

Refutation of Radical Contextualism

130

Negative vs. positive characterizations of RC

Most of the RC literature consists of a series of claims about what’s not the
case about communication in general. Here are some sample illustrations
of the kind of formulations we have in mind:

What is expressed by the utterance of a sentence in a context goes
beyond what is encoded in the sentence itself. Truth-conditional
content depends on an indeﬁnite number of unstated background
assumptions, not all of which can be made explicit. A change in back-
ground assumptions can change truth-conditions, even bracketing
disambiguation and reference assignment. That is, even after disam-
biguating any ambiguous words in a sentence and assigning seman-
tic values to any indexical expressions in the sentence,
truth-conditions may vary with variations in the background.
(Bezuidenhout 2002, p. 105)

There is no sentence that we can produce that can settle all questions
about how some original sentence is to be understood, since language
doesn’t function that way. It is not self-interpreting. Any sentence
that is produced itself has its content only against a background of
assumptions. (Bezuidenhout 2002, p. 113)

According to contextualism, the sort of content which utterances
have (in virtue of the speech acts they serve to perform) can never be
fully encoded into a sentence; hence it will never be the case that the
sentence itself expresses that content in virtue solely of the conven-
tions of the language. Sentences, by themselves, do not have deter-
minate contents. What gives them the determinate contents they have
(in context) is the fact that they are used in performing meaningful
actions. In brief, contextualism says that the gap between sentence
meaning and speaker’s meaning can never be closed. (Recanati 2003,
p. 194)

What words mean plays a role in ﬁxing when they would be true; but
not an exhaustive one. Meaning leaves room for variation in truth
conditions from one speaking to another. (Travis 1996, p. 451)

These kinds of remarks say nothing about how what is asserted is deter-
mined. They provide purely negative characterizations of RC. We take it

Objections to Radical Contextualism (III)

131

that you are not ﬁnished formulating your view if all you have done is
present a range of negations, i.e., a set of sentences of the form ‘It’s not the
case that A, B, C . . .’ It’s bad for theology and it’s no better for semantics.

We’re introducing the notion of a negative characterization of RC just

so we can easier locate our attack. Our objections take this form: As soon
as Radical Contextualists go from presenting purely negative claims to
making positive claims, these two aspects of RC conﬂict. The positive claims
directly contradict the negative ones. We have in mind two such conﬂicts:

(1) RC descriptions of conditions under which utterances of a sen-

tence S are true provide the basis for generalized truth conditions,
i.e., for a completely general description of the conditions under
which an arbitrary utterance of a sentence S is true. Exactly what
they deny is possible.

(2) RC descriptions of what speakers say by uttering sentences in

various contexts are incompatible with the Radical Contextualist
claim that there’s no common core, no one proposition expressed
in all these contexts, i.e., incompatible with the claim that there’s
no semantically minimal proposition expressed.

Second Inconsistency Charge: RC

Generalizations are Inconsistent with

Negative RC-Claims

Semantic Minimalists have a response to the question What does it take for
an utterance of ‘Smith weighs 80 kg’ to be true? Such utterances are true just in
case Smith weighs 80 kg. Such utterances all express the proposition that
Smith weighs 80 kg. What’s the Radical Contextualist’s reply to this ques-
tion? She denies our view, so she obviously thinks utterances of ‘Smith
weighs 80 kg’ can be true even though Smith doesn’t weigh 80 kg. But, if
it’s not Smith’s weighing 80 kg that renders the utterances true, what does?
Travis has responded to a related concern:

What could make the given words ‘The leaves are green’ true, other
than the presumed ‘fact that the leaves are green,’ is the fact that the
leaves counted as green on the occasion of that speaking. Since what
sometimes counts as green may sometimes not be, there may still be
something to make other words ‘The leaves are green’ false, namely
that on the occasion of their speaking, those leaves (at that time) did
not count as green. (Travis 1997a, pp. 101–2; see also Travis 1996,
p. 457)

This passage suggests the following response to our question (of course,
we’re applying what he says here to ‘Smith weighs 80 kg,’ so just change

Refutation of Radical Contextualism

132

from ‘leaves’ to ‘Smith’ and from ‘green’ to ‘weighs 80 kg’): an utterance of
‘Smith weighs 80 kg’ can be true when Smith doesn’t in fact weigh 80 kg
because he might count as weighing 80 kg.

Apparently, as Travis sees things, Smith can count as weighing 80 kg even

if he doesn’t weigh 80 kg.

So, in response to the question ‘What, other than

Smith weighing 80 kg, can sufﬁce for the truth of an utterance of “Smith
weighs 80 kg”?’ Travis has a positive proposal: It sufﬁces that Smith counts
as weighing 80 kg (in a situation) and Smith can count as weighing 80 kg
in some situation even though he doesn’t weigh 80 kg.

Notice that as soon as Travis provides this answer (more generally, any

answer to our question) he has, in effect, speciﬁed (generalized) truth con-
ditions for ‘Smith weighs 80 kg.’ His positive position now seems to be that
this sentence is true in a context C just in case Smith counts in C as weigh-
ing 80 kg. Travis has provided no argument for thinking that this doesn’t
follow from the meaning of ‘Smith weighs 80 kg’ alone. So, the correct
T-sentence for this sentence, according to Travis, as determined by its
meaning, is something like (A):

(A) ‘Smith weighs 80 kg’ is true in a context C iff Smith counts in C

as weighing 80 kg.

Notice that some version of (A) is available for whatever answer Travis, or
any other Radical Contextualist, gives to the question ‘What can make an
utterance of “Smith weighs 80 kg” true except Smith weighing 80 kg?’

Travis’s position is internally inconsistent. For, as soon as he tells us

what would sufﬁce for the truth of ‘Smith weighs 80 kg’ in a context C other
than Smith weighing 80 kg, and he has to tell us that in order to provide a posi-
tive proposal to answer our claim that nothing else would sufﬁce, he has provided
a (generalized) T-sentence.

Think of it like this: There’s a predicate ‘counts as weighing 80 kg’ which

is such that it can be used, in the context Travis ﬁnds himself while writing
this passage, to pick out the correct truth conditions for ‘Smith weighs 80
kg’ in any context of utterance. Now, we don’t think this is the correct way
to describe these truth conditions, but they are an attempt to do so. At this
point, we’re not debating whether ‘Smith weighs 80 kg’ has generalized
truth conditions or not; we are simply debating the correct form of these
truth conditions. Should it be something along the lines of

Objections to Radical Contextualism (III)

133

2 A natural extension would be that he thinks that an utterance of ‘Smith weighs 80 kg’ can
be false in a context C, when Smith does weigh 80 kg, as long as he doesn’t count in C as weigh-
ing 80 kg.

For any utterance u of ‘Smith weighs 80 kg’ in a context C, u is true
just in case Smith weighs 80 kg.

For any utterance u of ‘Smith weighs 80 kg’ in a context C, u is true
just in case Smith counts as weighing 80 kg in C.

We imagine three possible replies:

(1) According to the ﬁrst, the Radical Contextualists should just give

up the claim that sentences lack truth conditions. The Radical
Contextualist says something like: ‘OK, if that’s enough for it to
be the case that sentences have truth conditions, then have it your
way. Call them truth conditions.’

(2) According to the second reply, these ‘counts as’ truth conditions

do not depend on the meaning of the words, and that’s the central
discrepancy between the two possibilities.

(3) According to the third reply, Radical Contextualists should not say

the sorts of things Travis here says.

Our response to the ﬁrst reply. This is not a reply available to a Radical
Contextualist since it is, in effect, to give up RC. See the representative
quotes in the section ‘Preliminaries for the Second and Third Inconsistency
Objections’ above.

Our response to the second reply. We have an objection to RC only if what ﬁlls
in the right hand side (RHS) of (A) is determined by the meaning of the
words (and their mode of combination). We don’t see how a Radical Con-
textualist could deny that the RHS of (A) is so determined. We are supposed
to be able to ﬁgure out what the truth conditions of an utterance are from
what sentence is uttered and the context in which it is uttered. We don’t go
on nothing presumably, and different people ﬁgure out pretty much the
same thing. (How else can Travis rely on our reaction to his examples?) But
if that’s true, then there must be some sort of tacit rule guiding our inter-
pretations. Something like: Look for the relevant information (where this pre-
sumably has something to do with intentions and social practices) that actually
determines the truth conditions. If RC were right about the practice, then this
is a routine feature of interpretation of sentences, and everyone recognizes

Refutation of Radical Contextualism

134

3 A fourth reply is indirectly considered in the subsection of Chapter 10 entitled ‘PoC 3.’ We
there consider the view that it cannot be ‘real semantic truth conditions’ if it appeals to
speaker’s intentions.

that everyone else follows this general rule when we use sentences. Thus,
that we follow these rules in looking for the relevant information looks like
a convention associated with our use of sentences, and if we always (or
almost always) use it, then there’s no reason to deny that it attaches to the
use of the words as a matter of convention, and hence, to their conventional
meaning.

Our response to the third reply. In direct response to our argument, Bezuiden-
hout says the following:

this objection arises simply because of the way Travis happened to
frame the idea of occasion-sensitivity. When I introduced Travis’s
ideas . . . this is what I said: ‘If there were no occasion-sensitivity . . . we
might say that the English words “is blue” correctly describe an item just in
case it is blue. But there is occasion-sensitivity, which means that there are dif-
ferent possible understandings of “is blue”, depending on the circumstances.
Thus at best we could have a rule for English which says that “is blue” cor-
rectly describes an item in a certain context just in case it is then correct to
apply “is blue” to it.’ I think it is clear that when phrased in this way,
Travis’s notion of occasion-sensitivity cannot be exploited to give rise
to a rule of the sort [A], and hence the idea of occasion-sensitivity does
not give rise to an inconsistency. (Bezuidenhout 2002, pp. 128–9;
emphasis our own)

In effect, Bezuidenhout recommends that Radical Contextualists should
refuse to answer the question What, other than Smith weighing 80 kg, could make
an utterance of ‘Smith weighs 80 kg’ true? This is just a variation on the ques-
tion What does it take for an object to satisfy ‘is blue’? It’s no response to that
question to say, as Bezuidenhout does, that ‘is blue’ correctly describes an item
in a certain context just in case it is then correct to apply ‘is blue’ to it. She claims
that ‘is blue’ can be satisﬁed by things other than things which are blue.
When we ask her what, other than being blue, can sufﬁce for satisfying ‘is
blue,’ her reply is simply ‘Whatever it takes to satisfy “blue.” ’ This is no
answer to our question.

Why isn’t it acceptable for her to refuse to generalize about what it takes

to satisfy ‘is blue’? Why wouldn’t it be acceptable for a Radical Contextu-
alist to remain mute? We’re tempted to say: Because, if she opts for this
rejoinder, then she doesn’t have a theory. You don’t have a theory if all you
say, in general, about the satisfaction conditions for ‘blue’ is that something
satisﬁes ‘blue’ just in case it is correct to apply ‘blue’ to it. We’re very
tempted to give this reply, but we will resist this temptation since we’re not
in the mood for a protracted discussion or debate about the nature of
theories, whether one is needed, etc.

Objections to Radical Contextualism (III)

135

Here’s a more direct and useful response to the suggestion that Radical

Contextualists can refuse to answer our question: It makes no difference to
us which questions Radical Contextualists opt to answer. It makes no dif-
ference to us what they choose to say in response to our questions. What
does matter to us is only what they are committed to, i.e., what follows from
their positive claims and from their methodology. So, for us, the interesting
question is not what they actually say (or which questions they choose to
answer), but rather whether it would be true (given other claims they make
and given their methodology) to say what Travis says. Does Bezuidenhout
really want to deny Travis’s claim? Let’s break it down.

The ﬁrst claim made in the Travis quote above simply says that ‘What

sometimes counts as green may sometimes not be,’ i.e., a thing can satisfy
‘is green’ even though it is not green. No Radical Contextualist can deny
this, because to do so is, in effect, to endorse Semantic Minimalism. It is
constitutive of Semantic Minimalism that only something green can satisfy
‘green,’ and so anyone who endorses this claim endorses Semantic
Minimalism.

The second central claim in the Travis quote is this: ‘If an utterance of

“The leaves are green” isn’t true, it is because on the occasion of their speak-
ing, those leaves (at that time) did not count as green.’ Does Bezuidenhout
really want to deny this claim, i.e., does she really want to claim that an
utterance of ‘The leaves are green’ can be true even though in that context
of utterance those leaves didn’t count as green? If so, she’s committed to
presenting us with a context of utterance in which an object both isn’t green
and doesn’t count as green, but still satisﬁes ‘is green.’ We don’t think she,
or any other Radical Contextualist, can (or even wants to) come up with
any such a scenario.

Third Inconsistency Charge: RC Descriptions

of what Speakers say are Inconsistent with

Negative RC-Claims

Our second inconsistency charge against RC arises from the way a Radical
Contextualist describes truth conditions, and, more generally, the circum-
stances that allegedly make the propositions expressed by various utter-
ances true. We now look at passages in which Radical Contextualists
describe speciﬁc circumstances of evaluation that are alleged to make the
propositions expressed by various utterances true.

Here is a passage in which Bezuidenhout talks about utterances of the

word ‘blue.’ Pay close attention to the way she uses ‘blue’ (in particular, the
uses we have put in boldface):

Refutation of Radical Contextualism

136

Consider again the example of Pia and her ink. When Pia asks the sta-
tioner for blue ink, and he hands her an opaque box with a bottle
inside, saying ‘It’s blue,’ how is he to be understood? First, the
meaning of ‘is blue’ in English imposes some constraints. This pred-
icate is part of a system for describing the colors of objects, which
includes categories such as blue, green, red, black, etc. This system is
governed by certain principles, such as that an object’s color is a
matter of its looking a certain way in normal conditions. But what
constitutes normal conditions will vary depending on the particular
application that is being made of the color system. On one possible
way of applying the color system to ink, the color of ink is seen by
writing with it. On another way of applying the system, ink’s color
can be seen in the bottle. Let us assume that Pia wants to write with
this ink. Then being given the information that the ink in the box is
blue in the sense that it writes blue would be of value to her. On the
other hand, being given the information that the ink in the box is blue
in the sense that it looks blue in the bottle is of less value. Pia attrib-
utes a certain level of worldliness and reasonableness to the stationer.
He presumably knows that most customers come into his store to buy
ink for the purpose of writing with it, since this is the stereotypical
use for ink. Hence he is to be understood as saying that the ink is blue
in the sense that it will write blue (on normal paper). (Bezuidenhout
2002, p. 120)

Call the context in which Bezuidenhout writes, the B-context. In this objec-
tion we focus on four sentences uttered in the B-context. (It’s (4) that’s par-
ticularly important right now, but to put it in the proper context we list all
of (1)–(4).)

(1) Then being given the information that the ink in the box is blue

in the sense that it writes blue would be of value to her.

(2) On the other hand, being given the information that the ink in

the box is blue in the sense that it looks blue in the bottle is of
less value.

(3) Hence, he is to be understood as saying that the ink is blue in the

sense that it will write blue (on normal paper).

(4) When Pia asks the stationer for blue ink, and he hands her an

opaque box with a bottle inside, saying ‘It’s blue,’ how is he to be
understood?

First, put aside the confusing ‘is blue in the sense . . .’ This construction is
ungrammatical, and we’re not sure how to interpret it; we’re not even sure

Objections to Radical Contextualism (III)

137

it can be rendered intelligible, but we will resist the temptation to base any
of our argument on these worries. The best we can do is to suggest a para-
phrase of it as a sort of metalinguistic claim, for instance, as ‘Can be described
by the word “blue” when that word is used to mean . . .’

The third objection against Bezuidenhout is elicited by focusing on

(4). Notice that its ﬁrst occurrence of ‘blue’ is used, and not quoted; i.e.,
it is not mentioned. Bezuidenhout uses ‘blue’ when she describes what
Pia asked for. She does that for a very good reason. She does it because
it is true that what Pia asked for is blue ink. But if that is true, and it is,
then the implications are devastating for RC. What it implies is that
Bezuidenhout can use ‘blue’ in the B-context to pick out the same property
as Pia did when she used ‘blue’ in her context. Finally, and this is a crucial
step in our argument, there’s nothing special about Pia’s utterance of ‘blue’
that renders this indirect report acceptable. It is not as if Bezuidenhout
sat down and thought hard about the details of Pia’s context of utterance
and concluded that it was OK to utter (4). No. (4) would be true no matter
what was going on in a context where Pia came into a store and uttered
‘Can I have some blue ink please?’; more generally, any context in which
she asked ‘Can I have some blue ink please?’ is one in which Pia asked for
blue ink.

What follows is that Bezuidenhout in the B-context can use ‘blue’ to pick

out the property (or to mean the same as, whichever is your favorite termi-
nology) that an utterance of ‘blue’ is used to pick out in an arbitrary context
of utterance. More generally, Bezuidenhout can use ‘Pia asked for some blue
ink’ to describe what Pia asked for by any utterance of ‘Can I have some
blue ink please?’

Remember, RC is committed to the negative claim that there’s no one

property picked out by various utterances of ‘blue’ and no one proposition
expressed by various utterances of ‘Can I have some blue ink please?’ That’s
not compatible with Bezuidenhout being able to use an utterance of ‘blue’
in the B-context to pick out that which ‘blue’ picks out in an arbitrary
context. Nor is it compatible with Bezuidenhout being able to use ‘She
asked for blue ink’ to describe the request made by an arbitrary utterance
of ‘Can I have some blue ink please?’

Of course, Bezuidenhout could deny that all these different people who

utter ‘Can I have some blue ink please?’ ask for blue ink. But there’s a reason
why she doesn’t deny it. Pia did ask for blue ink. To deny this would be, if
not absurd, at least a move that would place RC deeply out of joint with
our intuitions about what speakers say. That would be a peculiar conse-
quence since a central goal of RC is to present a theory that’s faithful to
our intuitions about what is said by utterances.

Refutation of Radical Contextualism

138

Corollary of the third inconsistency charge

Here’s a corollary of the above point. Consider sentences (1)–(3) from the
passage quoted above.

(1) Then being given the information that the ink in the box is blue

in the sense that it writes blue would be of value to her.

(2) On the other hand, being given the information that the ink in

the box is blue in the sense that it looks blue in the bottle is of
less value.

(3) Hence, he is to be understood as saying that the ink is blue in the

sense that it will write blue (on normal paper).

Focus on the second occurrence of ‘blue’ in each of (1)–(3). What do they
mean? It certainly looks as if Bezuidenhout is using ‘blue’ (in her B-context)
to refer to something that the two phenomena she describes have in
common, i.e., it certainly seems that she is using ‘blue’ to refer to something
that blue writing ink and blue looking ink have in common. But what exactly
do they have in common? Well, the one looks blue, and the other writes blue.
What can ‘blue’ refer to that describes both these things? Another way of
asking this question is this: If something writes and looks blue, what do
the writing and the looking have in common?

We’re not metaphysicians, so we’re not going to provide an elaborate

answer: Let’s just call whatever they have in common blue

blue

(i.e., we’re going

to use ‘blue

blue

’ to refer to whatever it is that all these uses have in common).

We can leave it to the metaphysicians to determine what exactly blue

blue

since the answer to that metaphysical question is irrelevant to our concerns.
(If the answer is, and we suspect it is, no more than blue

blue

= blue, that’s

ﬁne with us.)

Our critical point is linguistic: To make sense of Bezuidenhout’s passage,

you have to assume that her ‘blue’ refers to something that all these differ-
ent ways of being blue have in common. So ‘blue’ as used in (1)–(3) refers to
blue

blue

. But now go back to the use of ‘blue’ in (4). Bezuidenhout uses ‘blue’

when she describes what Pia asked for. As we pointed out above, she does
that for a very good reason. She does it because it is true that what Pia asked
for is blue ink. So, she uses ‘blue’ in the B-context to characterize the
content of Pia’s utterance. So, the semantic value ‘blue’ takes when it is used
in the B-context can capture what Pia said by using ‘blue’ in Pia’s context.
If what she is asking for is the ink of that color which the writing and
the looking have in common, i.e. blue

blue

, then in the imagined context of

utterance, Pia uses ‘blue’ to refer to blue

blue

(or uses it with blue

blue

as its

semantic value).

Objections to Radical Contextualism (III)

139

We conclude: The ways Bezuidenhout uses ‘blue’ to describe truth con-

ditions for utterances of ‘blue’ other than her own, and to describe what is
said by these other utterances involving the word ‘blue,’ are incompatible
with RC. They assume that her own utterances of ‘blue’ refer to blue

blue

and

that this is what Pia also referred to.

Even if you followed what we had to say above and even if you happened

to agree with it, you might still think we’re just exploiting some accidental
features of a particular discussion of a particular example by a particular
Radical Contextualist. How are we to establish that our critical points
generalize?

Well, all we can say is this: They all do. We can only respond to what they

say and write. If there’s some other way to do it, then show it to us. More
importantly, for someone not to present the cases as Bezuidenhout does
would be bizarre. It would involve denying that Pia asked for blue ink,
denying that ink that writes blue and looks blue have something in
common, i.e., that the one looks the same color as the other one writes.
There might be positions that incorporate such views, but we haven’t seen
them made explicit.

Refutation of Radical Contextualism

140

P A R T I I I

Semantic Minimalism and

Speech Act Pluralism

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

C H A P T E R 1 0

Semantic Minimalism

143

So far we’ve tried to convince our readers that certain views about seman-
tic content and context sensitivity are wrong. In this third and last part of
the book we’ll present the view we think is correct. The correct view is the
combination of two views: Semantic Minimalism and Speech Act Pluralism.
In this and the next two chapters we present and defend Semantic
Minimalism and in the ﬁnal chapter, Chapter 13, we defend Speech Act
Pluralism. Since the two views go together, we will assume Speech Act
Pluralism in this and the next two chapters.

Semantic Minimalism: Basic Idea

The idea motivating Semantic Minimalism is simple and obvious: The
semantic content of a sentence S is the content that all utterances of S share.
It is the content that all utterances of S express no matter how different
their contexts of utterance are. It is also the content that can be grasped
and reported by someone who is ignorant about the relevant characteris-
tics of the context in which an utterance of S took place.

The minimal proposition cannot be characterized completely indepen-

dently of the context of utterance. Semantic Minimalism recognizes a
small subset of expressions that interact with contexts of utterance in
privileged ways; we call these the genuinely context sensitive expressions. When
such an expression occurs in a sentence S, all competent speakers know that
they need to know something about the context of utterance in order to
grasp the proposition semantically expressed by that utterance of S, and
to recognize the truth conditions of its utterance. These context sensitive
expressions exhaust the extent of contextual inﬂuence on semantic
content.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

Semantic Minimalism: Elaboration

Semantic Minimalism can be presented in a more elaborated form as the
conjunction of the following seven theses:

(1) One of the propositions expressed by a genuine utterance of an

English sentence is the proposition semantically expressed.

(2) That there is a proposition semantically expressed is presupposed

by any coherent account of linguistic communication, i.e.,
accounts which fail to recognize a semantically expressed propo-
sition (more or less as characterized in this book) are incoherent.

(3) All semantic context sensitivity (i.e., context sensitivity that

affects the proposition semantically expressed) is grammatically
triggered, i.e., it is triggered by a grammatically (i.e., syntactically
or morphemically) articulated sentential component.

(4) There are only a handful of context sensitive expressions in natural

language and they all pass the tests presented in Chapter 7.

Here is a simple effort to provide a list of the semantically context sensitive
expressions in English: The personal pronouns ‘I,’ ‘you,’ ‘he,’ ‘she,’ ‘it’ in
their various cases and number (e.g., singular, plural, nominative,
accusative, genitive forms), the demonstrative pronouns ‘that’ and ‘this’ in
their various cases and number, the adverbs ‘here,’ ‘there,’ ‘now,’ ‘today,’
‘yesterday,’ ‘tomorrow,’ ‘ago’ (as in ‘He left two days ago’), ‘hence(forth)’ (as
in ‘There will be no talking henceforth’), and the adjectives ‘actual’ and
‘present.’ Words and aspects of words that indicate tense also have their
reference so determined (cf., Kaplan 1989a, p. 489).

(5) In order to ﬁx or determine the proposition semantically

expressed by an utterance of a sentence S, follow steps (a)–(e):
(a) Specify the meaning (or semantic value) of every expression

in S (doing so in accordance with your favorite semantic
theory, i.e., we want Semantic Minimalism to be neutral
between the different accounts of how best to assign seman-
tic values to linguistic expressions; e.g., objects, sets, proper-
ties, functions, conceptual roles, stereotypes, or whatever).

(b) Specify all the relevant compositional meaning rules for

English (doing so also in accordance with your favorite
semantic theory; again, we insist upon Semantic Minimalism

Semantic Minimalism and Speech Act Pluralism

144

1 We don’t mean to rule out the possibility that there are other context sensitive expressions.
There might also be unclear (potential borderline) cases.

being neutral between different accounts of how best to
respect compositionality).

(c) Disambiguate every ambiguous/polysemous expression in S.
(d) Precisify every vague expression in S.
(e) Fix the semantic value of every context sensitive expression in

(6) The proposition semantically expressed by an utterance u of a

(declarative) sentence S does not exhaust the speech act content
of u. An utterance of a sentence S typically says, asserts, claims,
etc. a wide range of propositions in addition to the proposition
semantically expressed.

(7) An epistemic corollary of thesis 6, one that cannot be emphasized

enough, is that intuitions about, and other evidence for, speech
act content are not direct evidence for semantic content: an intu-
ition to the effect that an utterance u said that p is not even prima
facie evidence that p is the proposition semantically expressed by
u. This is so no matter how reﬁned, reﬂected, or ‘equilibriumized’
the intuition in question might be.

Semantic Minimalism: Illustration

As a simple illustration of our position on Semantic Minimalism, consider
an utterance of sentence (1):

(1) She’s happy.

(1) includes (at least) two indexical elements: ‘she’ and a tense indicator
indicating simple present tense. To locate the proposition semantically
expressed by an utterance of (1), ﬁrst make sure there’s no ambiguity
present (there isn’t). Precisify every vague expression (for the sake of argu-
ment, assume there are none). Fix the propositional components that cor-
respond to the two indexical components. That gives you the contextually
salient female

b and the time of utterance t. The proposition semantically

expressed can, for heuristic purposes, be represented as an ordered triple
along the lines of

·b, t, happyÒ, (where ‘happy’ corresponds to the semantic

value of ‘happy,’ whatever that might be). This proposition is not the only
proposition that the speaker said or asserted in uttering (1). The speaker
might, for example, be saying that b is no longer angry, or that her medi-
cation is working, or that she’s ready to meet her sister. If the speaker’s
utterance of (1) is ironic, it can be used effectively to say that b is in a bad

Semantic Minimalism

145

2 As mentioned above, we take no stand on the exact nature of the reference ﬁxing rule for
‘she.’

mood. There cannot be a systematic theory that predicts in advance what
someone can use a particular utterance of (1) to say. We suspect that it just
isn’t the sort of subject matter that lends itself to any kind of serious, or
rigorous, theorizing.

Semantic Minimalism: Points of

Clariﬁcation

Obviously, much more needs to be said about theses (1)–(7), but seven
points of clariﬁcation are of immediate signiﬁcance.

PoC 1. As already indicated, throughout this book we have tried to remain
neutral between competing semantic frameworks. In particular, we have
presented all our arguments in a manner that remains neutral about
whether a semantic theory must appeal to something like a proposition
semantically expressed in addition to assigning interpretive truth condi-
tions, or whether a semantic theory that speciﬁes interpretive truth condi-
tions alone sufﬁces.

PoC 2. In line with this neutrality, there are two ways to render more precise
what we mean by saying of an expression e that it is context sensitive.

(a) To say that e is context sensitive is to say that its contribution to

the propositions expressed by utterances of sentences containing
e varies from context to context. For example, if a sentence S con-
tains the word ‘I,’ the contribution of that word to the proposi-
tions semantically expressed by utterances u of S is the utterer (or
the speaker) of u himself.

(b) To say that e is context sensitive is to say that its contribution to

the truth conditions of utterances u of a sentence S containing e
(in some way or other) references various aspects of the context of
u. For example, the truth conditions for an utterance u of a sen-
tence S containing ‘I’ will, on the right hand side of a biconditional
specifying the interpretive truth conditions for u, make reference
to the speaker (e.g., if u is an utterance of ‘I am hungry’ by a speaker
A, then u is true just in case A is hungry at the time of u).

PoC 3. A huge chunk of the current debate about these issues is distorted
by a peculiar obsession with how to use the words ‘semantic’ and ‘prag-
matic.’ We emphatically don’t care about these terminological issues. We
do, however, think the debate about these issues has reached such an ele-
vated stage of confusion that it is advisable not to frame it in terms of the
‘semantic–pragmatic’ distinction. There are just too many such distinctions

Semantic Minimalism and Speech Act Pluralism

146

and hence any debate framed in those terms is too easily led astray. For this
reason, we distinguish between semantic content and speech act content.
We stay clear of the term ‘pragmatic.’ If someone would like to use the
word ‘semantic’ for something other than what we use it for, we’re OK with
that.

PoC 4. A closely related point has to do with a distinction sometimes drawn
between wide and narrow context. This distinction plays an important role
in the work of, for example, Recanati, Bach, and Carston. Recanati, in par-
ticular, seems to think the distinction plays an important role in an objec-
tion he raises to what he calls Semantic Minimalism. As far as we can tell,
his objection is purely terminological, and that’s why we address it here.

The argument can be found in Recanati (2004). He ﬁrst introduces the

distinction with a quote from Bach:

There are two quite different sorts of context, and each plays quite a
different role. Wide context concerns any contextual information rel-
evant to determining the speaker’s intention and to the successful
and felicitous performance of the speech act . . . Narrow context con-
cerns information speciﬁcally relevant to determining the semantic
values of [indexicals] . . . Narrow context is semantic, wide context
pragmatic. (Recanati 2004, p. 66)

It’s not entirely clear from this passage (or the rest of the text) exactly how
to deﬁne the wide–narrow distinction, but for the purposes of Recanati’s
argument the only important point is that wide context takes speaker’s inten-
tions into account, whereas narrow context does not.

He then goes on to argue that some of the expressions Semantic

Minimalists (such as us) treat as semantically context sensitive depend on
wide context. He says about demonstratives:

It is generally assumed that there is such a rule, namely, the rule that
the demonstrative refers to the object which happens to be demon-
strated or which happens to be the most salient, in the context at
hand. But the notions of ‘demonstration’ and ‘salience’ are pragmatic
notions in disguise. They cannot be cashed out in terms merely of the
narrow context. Ultimately, a demonstrative refers to what the speaker
who uses it refers to by using it. (Recanati 2004, p. 67)

About ‘here’ and ‘now’ he says:

We encounter the same sort of problem even with expressions like
‘here’ and ‘now’ which are traditionally considered as pure indexicals

Semantic Minimalism

147

(rather than demonstratives). Their semantic value is the time or place
of the context respectively. But what counts as the time and place of
the context? How inclusive must the time or place in question be? It
depends on what the speaker means, hence, again, on the wide
context. (Recanati 2004, p. 68)

These are insights that we might agree with, i.e., we might agree that the
semantic value of semantically context sensitive expressions might be ﬁxed
by the intentions of the speaker. We don’t (need to) assume that the seman-
tic value is ﬁxed by purely ‘objective’ features of the context of utterance.
Our view need not take a stand on the question of the exact nature of these
reference ﬁxing mechanisms, but as far as we can tell, Recanati is probably
right: wide context is involved. (At least we would be willing to concede that
for the sake of argument.)

Recanati, however, thinks this reliance on wide context is incompatible

with Semantic Minimalism. He says:

It follows that semantic interpretation by itself cannot determine
what is said by a sentence containing such an expression: for the
semantic value of the expression – its own contribution to what is said
– is a matter of speaker’s meaning, and can only be determined by
pragmatic interpretation. (Recanati 2004, pp. 67–8)

Recanati seems to hold the view that whenever the semantic content relies
on wide context (i.e., on speaker’s intentions) to ﬁx semantic values, that
semantic content isn’t pure semantic content, but is in part pragmatic.
Hence, it shouldn’t be called semantic content. It’s somehow cheating to call
it semantic content. He says:

To be sure, one can make that into a semantic rule. One can say that
the character of a demonstrative is the rule that it refers to what the
speaker intends to refer to. As a result, one will add to the narrow
context a sequence of ‘speaker’s intended referents,’ in such a way that
the nth demonstrative in the sentence will refer to the nth member of
the sequence. Formally that is ﬁne, but philosophically it is clear that
one is cheating. We pretend that we can manage with a limited, narrow
notion of context of the sort we need for handling indexicals, while
in fact we can only determine the speaker’s intended referent (hence
the semantic referent, which depends upon the speaker’s intended
referent) by resorting to pragmatic interpretation and relying on the
wide context. (Recanati 2004, pp. 67–8; emphasis our own)

Semantic Minimalism and Speech Act Pluralism

148

We ﬁnd this remark about ‘cheating’ bafﬂing.

Why is it cheating? Of

course, if someone claimed that the semantic content didn’t depend in any
way on speaker’s intentions, it would be cheating, but we don’t know of
anyone who makes that claim. It is telling in this respect that Recanati
doesn’t quote a single person who does. We certainly don’t. Maybe he thinks
it’s cheating because he thinks the word ‘semantic’ should be used to
describe only those features of communicated content that do not depend
on speaker’s intentions. If that’s how Recanati wants to use the term
‘semantics,’ that’s OK with us. It’s just not how we use it.

PoC 5. Another related point concerns a distinction made early in
Kaplan’s classic paper ‘Demonstratives.’ Kaplan distinguishes between
pure indexicals and true demonstratives. Utterances of sentences contain-
ing the latter must be accompanied by a demonstration in order to be
complete:

The linguistic rules which govern the use of the true demonstratives
‘that,’ ‘he,’ etc. are not sufﬁcient to determine their referent in all con-
texts of use. Something else – an associated demonstration – must be
provided. The linguistic rules assume that such a demonstration
accompanies each (demonstrative) use of a demonstrative. (Kaplan
1989a, p. 490)

Indexicals, on the other hand, need not be accompanied by a demonstra-
tion: ‘The linguistic rules which govern their use fully determine the refer-
ent for each context’ (Kaplan 1989a, p. 491). We turn your attention to this
distinction merely to emphasize that it is one we will intentionally ignore.
There have been extensive (and interesting) debates about the nature, and
the correct representation, of the rules that ﬁx the semantic values of
various context sensitive terms, about the role of demonstrations, about the
importance of speaker’s intentions, and about the nature of contextual

Semantic Minimalism

149

3 The ‘cheating’ remark is endorsed by Carston in the following passage: ‘We can, of course,
stipulate that it (or this or that) encodes a rule to the effect that it refers to what the speaker
intends to refer to, and we can add to the set of contextual parameters a sequence of “speaker’s
intended referents”, arranged in such a way that each demonstrative maps onto a referent as
required. But, as Recanati (2002) says, while that may be ﬁne from a formal point of view,
“philosophically it is clear that one is cheating”. To proceed in this formal way is to avoid
dealing with an undeniable cognitive reality, which is that the assignment of referents to the
vast range of linguistic referring expressions relies on a wide notion of context and requires
the intervention of pragmatic principles or strategies that are geared to the recovery of the
speaker’s intended meaning’ (Carston 2004, p. 7).
4 Where being complete is different from being nonvacuous (Kaplan 1989, p. 491).

salience. These are interesting and important issues but have no bearing
whatsoever on the arguments presented here.

PoC 6. We should point out that even though much about our view is
Gricean in perspective, our way of classifying contents (i.e., to contrast the
proposition semantically expressed by an utterance u of a sentence S with
the speech act content of u) distinguishes us (at least terminologically) from
Grice. For Grice, there’s an important distinction between what a speaker
says with an utterance and what she conversationally implicates with that
utterance. He thinks of the former more or less along the lines of what we
have been calling the proposition semantically expressed. It is important
for us (a) not to identify the proposition semantically expressed with the
proposition asserted (or said), and (b) for that reason not to reserve the label
‘what is said’ for the proposition semantically expressed. (See Chapters 12
and 13 for further discussion.)

PoC 7. It should be clear that Semantic Minimalism, as we understand it,
is not in the business of conceptual analysis. Semantic Minimalists are
happy to use the words and sentences they theorize about to characterize
the semantic content of those words and sentences. They need not be in the
business of analyzing the meanings of words. For example, it’s perfectly
acceptable, according to Semantic Minimalism, to say that ‘red’ is a word
that applies to red things, and that ‘Ducks have soft beaks’ expresses the
proposition that ducks have soft beaks, and is true just in case ducks have soft
beaks. The goal is not to analyze the basic expressions of the language being
studied. It is to reveal the structure of that language. In the words of
Bertrand Russell:

It is no more necessary to be able to say what a word means than it is
for a cricketer to know the mathematical theory of impact and of pro-
jectiles. Indeed, in the case of many object-words, it must be strictly
impossible to say what they mean, except by a tautology, for it is with
them that language begins. (Russell 1940, p. 26)

Semantic Minimalism: Arguments

The central components of the arguments in favor of Semantic Minimal-
ism have already been presented in earlier chapters and what remains will
be presented in the next two. Here we simply outline the two main argu-
mentative strategies that support Semantic Minimalism (when combined
with Speech Act Pluralism). The ﬁrst is an Argument by Elimination. The
second is an Argument from Explanatory Force.

Semantic Minimalism and Speech Act Pluralism

150

Argument by elimination

There are two alternatives to Semantic Minimalism: Moderate Contextual-
ism (endorsed by lots of contemporary philosophers, be they semantic
opportunists or just confused semanticists) and Radical Contextualism
(often thought of as a rather extreme view and less popular than Moderate
Contextualism). In Part I of this book we showed that MC is an unstable
position. A consistent Moderate Contextualist should become a Radical
Contextualist. In Part II we showed that Radical Contextualism is internally
inconsistent and fails to account for some of the basic features of our lin-
guistic practices. It is radically empirically inadequate in so far as it fails to
account for a wide range of intuitions that by the standards of Radical Con-
textualism should be accounted for. If we’re right, Semantic Minimalism is
the only game in town at the end of the day.

As a reminder for those who, for reasons we can’t comprehend, are not

convinced that MC is an unstable position: MC on its own has all the
defects of RC (except for the internal inconsistency) and the most popular
(and most plausible) version of MC has a wide range of additional prob-
lems presented in Chapter 6 (i.e., overgenerates necessary truths, predicts
anaphoric relations where there are none, overgenerates a priori truths). So,
even if you’re unconvinced that MC is unstable, Semantic Minimalism is
still preferable to MC.

Argument from explanatory force: The seven virtues

of Semantic Minimalism

Of course, the primary reason we think Semantic Minimalism is a good
theory is that it can explain all the data that a semantic theory should
explain. It accounts for all the evidence a semantic theory should account
for. In particular, it can account for the ways in which the content that an
utterance communicates is shaped by the context of that utterance.

Here, in summary, are what we take to be the seven primary virtues of

Semantic Minimalism as defended in this book:

First virtue. The most characteristic feature of Semantic Minimalism is

that it recognizes only a very limited set of semantically context sensitive
expressions. In Chapter 7 we argued that the members of this set are the
only expressions that pass our various tests for context sensitivity: the Inter-
Contextual Disquotational Indirect Report Test, the Collective Descrip-
tions Test (and the VP-ellipsis Test), and the ICD/RCSA Test. No semantic
theory should classify as context sensitive any expression that does not pass
these tests. Semantic Minimalism doesn’t. All other theories do.

Semantic Minimalism

151

Second virtue (a corollary of the ﬁrst virtue). Semantic Minimalism, and no

other view, can account for how Inter-Contextual Disquotational Indirect
Reports can be true where the reporter and the reportee ﬁnd themselves in
radically different contexts. In such cases, the reported content is the
semantic content. As an illustration, when A knows that B has uttered
‘Peter’s book was red,’ but knows nothing else about the context in which
that utterance took place, A can still say truly that B said that Peter’s book
was red (as we can do, in this very context). The content attributed to B’s
saying is the minimal content.

Theories that increase the magnitude of

contextual inﬂuence on content cannot explain the ease with which we
make such Inter-Contextual Disquotational Indirect Reports.

Third virtue. Semantic Minimalism, and no other view, can account for

how the same content can be expressed, claimed, asserted, questioned,
investigated, etc. in radically different contexts. It is the semantic content
that enables audiences who ﬁnd themselves in radically different contexts
to understand each other, to agree or disagree, to question and debate with
each other. It can serve this function simply because it is the sort of content
that is largely immune to contextual variations.

We should point out something entirely obvious in connection with the

second and third virtues: The underlying idea, namely, that it is essential
for a philosophy of language to explain how content can be shared across
contexts, does not originate with us. It is one of the fundamental strands
in twentieth century philosophy of language. A version of the underlying
idea is found, for example, in Frege’s objections to psychologism. Here’s a
pertinent passage from his paper ‘The Thought’:

If every thought requires an owner and belongs to the contents of his
consciousness, then the thought has this owner alone; and there is no
science common to many on which many could work, but perhaps I
have my science, a totality of thoughts whose owner I am, and another
person has his. Each of us is concerned with contents of his own con-
sciousness. No contradiction between the two sciences would then be
possible, and it would really be idle to dispute about truth; as idle,
indeed almost as ludicrous, as for two people to dispute whether a
hundred mark note were genuine, where each means the one he
himself had in his pocket and understood the word ‘genuine’ in his
own particular sense. (Frege 1977, p. 17)

Semantic Minimalism and Speech Act Pluralism

152

5 See comments in Chapter 13 about the speech act content of indirect reports for some
further qualiﬁcations.

In this passage Frege takes it to be self-evident that we can share thoughts,
that we can have a common science, and that we can agree and disagree.
His opinions are, in some vague form, a version of the (self-evidently true)
presumption that communication (as we understand it) is possible despite
radical differences. At the risk of offending Frege scholars everywhere, we
ask you to consider this (no doubt deeply anachronistic and, at some deep
level of exegesis, inaccurate) analogy. Frege in this passage is, at least in part,
trying to show that if thoughts were psychological states, then it would be
hard to see how individuals could communicate. The analogy is this: If com-
municated contents are restricted to (or, essentially tied to) speciﬁc con-
texts of utterance, then it is hard to envision how speakers who ﬁnd
themselves in different contexts can communicate, i.e., under such circum-
stances communication between contexts is thrown into doubt. To ﬂesh
out the analogy further, consider the following rewrite of Frege’s remarks
in ‘The Thought’:

If every thought requires a context and belongs to that context (i.e.,
is essentially tied to that context, could only be expressed in that
context), then the thought belongs to that context alone; and there is
no science common to many on which many could work, but perhaps
one context has one science, a totality of thoughts that belong to it;
and another context has its. Each context is concerned with its own
contents. No contradiction between the two sciences would then be
possible, and it would really be idle to dispute about truth; as idle,
indeed almost as ludicrous, as for two people to dispute whether a
hundred mark note were genuine, where each means the one he
himself had in his pocket and understood the word ‘genuine’ in his
own particular sense.

Semantic content, as characterized by Semantic Minimalism, is, we claim,
the only way to avoid the kind of contextual isolationism or semantic solip-
sism our (modiﬁed) Frege here mocks.

The ﬁnal four virtues we will defend in later chapters, but we mention

them brieﬂy here.

Fourth virtue. Semantic Minimalism, and no other theory, is based on a

correct understanding of the relationship between speech act content and
semantic content: any view according to which it is a requirement on a
semantic theory that it accounts for (or explains) our basic (most reﬁned)
intuitions about what speakers can use sentences to say, assert, state, etc.
fails. Semantic Minimalism rejects all such requirements on semantics. The
resulting separation of semantic content from speech act content is neces-
sary both in order to do semantics properly and in order to do speech act

Semantic Minimalism

153

theory properly. This point was elaborated on in Chapter 4 and will be
pursued further in Chapter 13.

Fifth virtue. Semantic Minimalism, and no other view, can account for

how what speakers say can be determined, in part, by the context of those
who report on what those speakers say even though the semantic content
is not so determined. No theory that ties semantic content and speech act
content closely together can do this (see Chapter 13). Using terminology
introduced in Chapter 13, the point can be put like this: Only Semantic
Minimalism (when combined with Speech Act Pluralism) can avoid Origi-
nal Utterance Centrism.

Sixth virtue. In Chapter 11 we show that Semantic Minimalism gets the

connection between semantics and metaphysics right. The anti-minimalists
end up requiring that semanticists do metaphysics. They are in effect, on
this interpretation, requiring that semantics provides answers to questions
such as ‘What is redness?’, ‘What is dancing?’, ‘What is tallness?’, ‘What is
it to be a fast giraffe?’, and so on. To require of semantics that it provide
answers to such questions is not only to misunderstand the division of
labor between the philosophy of language and metaphysics, but it is also
to make both disciplines impossibly difﬁcult.

Seventh virtue. Semantic Minimalism, and no other view, has a notion of

semantic content that is psychologically realistic. We argue for this in
Chapter 12.

Looking ahead

The previous section contains many promissory notes. In the next three
chapters we will try to deliver on those. In Chapter 11 we further clarify
the nature of the propositions that are semantically expressed according
to our view. We do that by responding to an objection. In so doing we
also present how our version of Semantic Minimalism construes the
relation between semantics and metaphysics. In Chapter 12 we discuss the
psychological–cognitive role of semantic content. Again, we do that by
responding to an objection, one that Carston, Recanati, and King and
Stanley raise against Semantic Minimalism. Then in the ﬁnal chapter,
Chapter 13, we present our reasons for Speech Act Pluralism.

Semantic Minimalism and Speech Act Pluralism

154

C H A P T E R 1 1

Semantics and Metaphysics

155

You’ve just read ten short chapters about semantic context sensitivity. We’ve
presented what seems to us a pretty obvious view (i.e., Semantic Mini-
malism) and refuted both of its apparent alternatives (i.e., Radical and
Moderate Contextualism). In this chapter we turn to a discussion of the
relationship between semantics (minimalism, in particular) and meta-
physics. Why, you might ask, would we want to do that? That’s a good ques-
tion, and to be honest, we don’t really want to, but objections that can be
(and have been) raised against us, properly understood, we will argue, are
metaphysical. The goal of this chapter is to respond to those objections and
clarify how minimalists should think of the division of labor between
philosophy of language and metaphysics.

Semantic Minimalism, as should be clear by now, is committed to each

of the following, some of which are ostensibly surprising:

(S1) An utterance of ‘A is red’ expresses the proposition that A is red

and it is true just in case A is red.

(S2) An utterance of ‘A dances’ expresses the proposition that A dances

and it is true just in case A dances.

(S3) An utterance of ‘A has had enough’ expresses the proposition

that A has had enough and it is true just in case A has had enough.

(S4) An utterance of ‘A is ready’ expresses the proposition that A is

ready and it is true just in case A is ready.

(S5) An utterance of ‘It’s raining’ expresses the proposition that it is

raining and it is true just in case it is raining.

(S6) An utterance of ‘A is tall’ expresses the proposition that A is tall

and it is true just in case A is tall.

One objection we’re imagining to our position is this:

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

I acknowledge that your arguments against Radical and Moderate
Contextualism are seemingly sound, but when a positive view ends up
endorsing claims like these, in particular, claims like S3–S6, isn’t that
a reductio of the view? How, for example, can there be any such thing
as being ready simpliciter? How can there be any such state of affairs?
What is it? Even worse: How can there be such a thing as being tall
simpliciter? What is that?

Unless you provide a positive account of

these, I don’t even understand Semantic Minimalism. To put it even
more pointedly: If you don’t offer more of an elaboration on your pos-
itive theory, you don’t yet even have a theory on offer – just a set of
objections to other views. If you’re right that the views you criticize
exhaust the realm of possibilities, then given your ‘positive’ view
it follows that there must be an error in your reasoning against its
alternatives.

At this point it is both extraordinarily important, and somewhat difﬁcult
(we’ve found), to keep the dialectic straight. In particular, it is important to
keep two issues distinct:

The metaphysics of minimal propositions. What are minimal propositions?

The cognitive–psychological and communicative role of minimal propositions.
Given that we have an understanding of what minimal propositions
are, we can ask: What role do they play in communication?

This chapter is about the ﬁrst issue. We’re imagining someone saying: The
minimal propositions aren’t real propositions. You might think that’s a
borderline silly objection when it comes to ‘A is red,’ you might think it’s
slightly less silly when it comes to ‘A is ready,’ but most philosophers think
it is a reasonable objection when it comes to ‘A is tall.’ So, the goal of this
chapter is to gradually move you over on our side by starting with ‘A is red’
and moving to ‘A is tall’ via ‘A is ready.’

The second set of issues, not discussed in this chapter, concerns the role

of minimal propositions in communication. Suppose we have convinced
you that there are propositions such as that A is red and that A is ready
and that A is tall. It doesn’t follow we ever express them; it doesn’t follow
they are the propositions semantically expressed. It doesn’t follow they
have a role to play in communication. That set of issues is discussed in
the next chapter. It makes sense, we think, to tell you about what we

Semantic Minimalism and Speech Act Pluralism

156

1 In particular, we expect those who have strong incompleteness intuitions (see the second
half of Chapter 2) to raise this objection. This entire chapter, in effect, can be thought of as a
response to those moved by Incompleteness Arguments of the kind described in the second
half of Chapter 2.

think minimal propositions are before we say more about their role in
communication.

Overview of our Strategy

Our strategy for responding to the metaphysical objection is simple (at least
in the abstract). It consists of three parts:

Part One. We discuss what we call the Metaphysical Objection to S1 and S2,
and show that:

䊉

If there’s a metaphysical puzzle about S1 or S2, then it is not about
S1 and S2 speciﬁcally, but about any claim to the effect that two
objects can share the same property or engage in the same activity. There’s
no special problem or puzzle about the properties of redness or
dancing per se.

䊉

But there is no such general metaphysical worry; and hence, there
is no worry about ‘red’ and ‘dance.’

Part Two. We discuss a version of the Metaphysical Objection directed at
claims S3–S5, and show that:

䊉

If there’s a metaphysical problem or puzzle with respect to S3–S5,
then there’s one with respect to S1 and S2 as well.

䊉

Since there’s no metaphysical problem or puzzle with respect to S1
and S2, there is none with respect to S3–S5 either.

Part Three. At the end of the chapter we discuss the Metaphysical Objection
directed at claim S6, and show that:

䊉

If there’s no metaphysical problem or puzzle with respect to S1–S5,
then there’s none with respect to S6 either.

Metaphysical Objection Version 1:

Redness and Dancing

Here’s the kind of worry or objection we are imagining being raised against
Semantic Minimalism:

The Semantic Minimalist says that an utterance of ‘A is red’ seman-
tically expresses the proposition that A is red. He also says that this
proposition is not the same proposition as the proposition that A is
red on the inside, or as that A is red on the outside, or as that A is red when

Semantics and Metaphysics

157

looked at through red sunglasses, or as that A is red in normal light. It’s just
the proposition that A is red. But which proposition is that? What, for
example, is it to be just red? What is it that all these things (those that
are red when seen through red sunglasses, red on the inside, red in
the dark, red on the outside, red when washed, etc.) have in common?
What is the state of affairs that the proposition that A is red picks out?

First of all, note that this worry has nothing speciﬁcally to do with

redness. It generalizes. What, for example, do all dancers have in common?
Some ﬂy in the air; some underwater; some with music; some without; some
stand on their feet; some crawl; some touch other people; some don’t. What
is the state of affairs that the proposition that A dances picks out?

Answer (the short and accurate one). The proposition that A is red is true just
in case A is red. That’s it. That’s the whole semantic story. End of discus-
sion. The proposition that A dances is true just in case A dances. That’s the
most informative true answer anyone can provide. If you try to push the
issue any further, you’ll regret it because it’ll just turn into a rather large
metaphysical mess (not a mess of our making, just the regular mess meta-
physicians inevitably like to throw themselves into).

Metaphysical worry (again): Seriously, what is it to be red?

You might be unsatisﬁed with this reply, and push the issue further as
follows:

You’re not going to get off the hook that easily: Seriously, what is it
for A to be red? It sounds like you’re claiming that the proposition
that A is red is true just in case A is red in some way, or under some con-
ditions, or if it looks red under some conditions. Is that your view? If so, then
even the White House turns out to be red. Uma Thurman has red eyes.
(She also has blue eyes and green eyes.) But that’s absurd. It’s a reduc-
tio of Semantic Minimalism.

Answer (more elaborate, but ultimately, not more informative). OK, here we go.
First some general comments about the relationship between semantics
and metaphysics, and then a bit of metaphysics.

A preliminary point: We’re most certainly not claiming that something

is red just in case it is red in some way (or in some respect or under some condi-
tions). Our view is that to be red is just to be red. We’re adamant about this
short answer for two reasons. First, we refuse to mix metaphysics and
semantics (for more on that see below). Second, to stick the quantiﬁer in
there would make it seem as if we think either that there’s an argument

Semantic Minimalism and Speech Act Pluralism

158

place linguistically represented, say, in LF, that refers to a way of being red
and that that place can be bound by a quantiﬁer, or that the quantiﬁer
magically occurs (as with Perry’s unarticulated constituents). We hold
neither of these views.

Our response to this ﬁrst part of the Metaphysical Objection comes in

three stages. First, we explain why we think these objections are meta-
physical, and not semantic. Second, we show that whatever worries there
are with respect to redness and dancing generalize to every property and
activity. Third, we explain why we don’t think these metaphysical worries
should be taken very seriously anyway.

First stage of reply to Metaphysical Objection Version 1: The division

of labor between semantics and metaphysics

To respond to the Metaphysical Objection we ﬁrst have to outline how we
see the division of labor in philosophy, in particular, how we see the
relationship between metaphysics and philosophy of language. The Meta-
physical Objection is exactly that: it’s metaphysical. It is a ‘worry’ that
remains no matter how we opt to do semantics. If it is a worry, it is not one
that arises because of Semantic Minimalism, and it is not one that can be
solved by doing semantics in any particular manner.

Think about what metaphysicians do. For at least the last two mil-

lennia, metaphysicians have been asking What-Do-They-Have-in-Common
Questions (CQ, for short). Suppose you’re curious about what it is to be G.
Then you ask (this is at least one of the questions you ask):

(CQ) What do all G things have in common?

Suppose, for example, you’re interested in what consciousness is. Then you
ask a question like ‘What do all conscious things have in common?’ If you’re
interested in what causation is, you ask ‘What do all events A and B have
in common in which A causes B?’ If you’re interested in what personal
identity over time is, then you ask ‘What do all processes through which a
person undergoes change have in common?’ If you need more illustrations,
take an introductory course in metaphysics.

Notice that none of these questions are about language. They are not about

the expressions ‘conscious,’ ‘cause,’ or ‘identity.’ They are not about how
people use those expressions. They are nonlinguistic questions. Not only is
there no reason to think these worries can be solved by doing semantics, there
is no reason to think they have anything at all to do with semantics. In par-
ticular, these problems (or questions, or whatever you want to call them)
arise no matter what views you hold about linguistic context sensitivity.

Here are two illustrations of these points.

Semantics and Metaphysics

159

First illustration: Red. Some red things are red on the inside, some are red on
the outside, some are red when scrubbed, some are red in the dark, some
are red when peeled, some are red when seen through red sunglasses, etc.
You might ask the age-old metaphysical query: What do they all have in
common? What is this property of being red that they all share?

We ourselves have not given much thought to this question, so what we

have to say won’t be of much help. Suppose, for the sake of argument,
that the metaphysician of color concludes that one thing they all have in
common is that they look red on some surface or other under some conditions (we
are not endorsing this reply; we’re simply using it to illustrate a general
point). Inasmuch as this account still uses ‘red,’ presumably there remains
metaphysical work to be done unless, of course, as we suspect, no ‘reduc-
tive’ response is forthcoming, and so, all you end up with is that red things
are all red. But let’s suppose, for the sake of argument, that this answer con-
stitutes at least part of the reply to the CQ question.

Imagine someone registering this objection: Doesn’t that make it very

easy, indeed, too easy, to be red? Doesn’t it render both the White House
and Uma Thurman’s eyes red? Both things, after all, are red on some surface
under some conditions? Reply: Maybe, yes. When you think hard about what
it is to be red, maybe that’s all it takes. If so, then it turns out that it’s not
that hard to be red. But regardless of whether this is so, or whether it is
problematic, it is most certainly not a problem that arises because of views
one might hold about the context sensitivity of ‘red.’ On the face of it that
should be fairly obvious. Notice: the claim that many things can be red

(a) is not a claim about language; in particular, it is not a claim about

the word ‘red’; and

(b) does not involve thinking about or referring to contexts of utter-

ance for any linguistic item; the claim takes place in a context of
utterance (in particular, in this chapter), but it itself is not about
a context of utterance.

is to think about the properties of certain things in the world, and
whether those things are all red.

Suppose, for the sake of argument, that you think ‘red’ is a context sen-

sitive expression, i.e., imagine you’re a contextualist about ‘red.’ Then ask
yourself, on that assumption, What do things that are red on the inside, red on the
outside, red when seen through red sunglasses, etc. have in common? The assump-
tion that ‘red’ is context sensitive does not make asking the question any
easier or more difﬁcult, nor does it make the task of answering it any less
pressing or any less interesting. It certainly doesn’t follow from the view
that ‘red’ is context sensitive that these questions do not make sense. Hence,

Semantic Minimalism and Speech Act Pluralism

160

the puzzle (if there is one) doesn’t arise or disappear depending on what
view one might hold about contextualism with respect to ‘red.’

Second illustration: Dancing. Now some remarks on dancing and ‘dance.’
Some people dance by stepping, some crawl around the ﬂoor (like Martha
Graham), some have music, some don’t have music, some jump in the air,
some wave their arms, some hold on to other people, some are alone, some
slide on ice, some ﬂy in the air, etc. What do all these activities have in
common in virtue of which they are all dancing? Again, this is not our area
of expertise, but suppose the dance metaphysician informs us that: To
dance is to move in some way W. Again, we’ll remain ofﬁcially neutral on
the question of whether ‘W’ will turn out to be a reductive analysis or not.

Imagine the following objection: But doesn’t that make it way too easy

to dance? If that’s all it takes, aren’t we all dancing all the time? Suppose
the answer is: Maybe, yes. When you think real hard about what it is to
dance, perhaps that’s all it does take. If so, then it’s not that hard to dance.
But, again, regardless of whether this is so, or whether it is something you
ﬁnd problematic, it’s not a semantic problem. It is not a problem that arises
because of any view one might have about the semantics for ‘dance.’

Most importantly for this book: It is not a problem that has anything to

do with the semantics of context sensitivity. Again, suppose for the sake of
argument that ‘dance’ is a context sensitive expression, i.e., imagine that
you’re a contextualist about ‘dance.’ Then ask yourself: What do people who
dance by creeping around on the ﬂoor, or ﬂying in the air, or shaking their hands, etc.
all have in common? What makes all of these movements instances of
dancing? As in the case of ‘red,’ the assumption that ‘dance’ is context
sensitive neither prevents this question from arising nor provides an answer
to it. As far as we can tell, it is once again irrelevant to the metaphysical
question.

In sum, the ﬁrst stage of our response is this: The issues raised here are

the staple of metaphysics. We poor semanticists don’t have much to con-
tribute on these issues. Whatever problems might arise here are not solved,
dissolved, or illuminated by any view one might hold about semantic
context sensitivity. If contextualism about ‘red’ were true, it would still be
possible to ask the relevant CQs, i.e., ‘What do things red on the inside, red under
water, etc. have in common?’ The answer to this question doesn’t follow from

Semantics and Metaphysics

161

2 Suppose, for reasons we can’t fathom, a contextualist were to respond: If ‘red’ is context sen-
sitive, then the question ‘What do things that are red on the inside, red on the outside, red in the dark, etc.
have in common?’ makes no sense. If so, this contextualist claims that what we call below Meta-
physical Nihilism follows from contextualism. For reasons that will become clear below, that
should be considered a reductio of contextualism.
3 But we can’t resist the temptation to register our suspicion that ‘dance’ will occur in ‘W.’

contextualism, but whatever the answer might be, that’s the answer to the
puzzle that started this chapter, i.e., what is it to be red simpliciter?

Second stage of reply to Metaphysical Objection Version 1: Generalization

Suppose our reader is unconvinced so far. Suppose you have a residual sense
that there’s something deeply ﬁshy about claiming that there is such a thing
as being red simpliciter. And that there’s something ﬁshy about expressing
propositions that attribute redness (simpliciter) to objects. We now want to
show this skeptic that if it’s puzzling what it is to be red simpliciter, then it’s
equally puzzling what it is to be red on the inside, or red under water, or red
when washed on the left side. More generally: if it’s puzzling what it is to
be red simpliciter, then what you’re really puzzled about is the idea that a
range of disparate things can share a single property; that they can all be F,
whatever F is.

The problem with being red simpliciter is that it is hard, for our opponent

at least, to understand what all red things have in common. Remember,
she’s not satisﬁed with the answer that they are all red. But if that’s puz-
zling, it’s equally puzzling what it is for a range of things to be red on the
inside, or red under water, or red when washed on the left side. If you don’t
see this, consider the following examples of things each of which is red on
the inside:

䊉

a human that’s red on the inside

䊉

an apple that’s red on the inside

䊉

a gas that’s red on the inside

䊉

a liquid that’s red on the inside

䊉

a bottle that’s red on the inside

䊉

a car that’s red on the inside

䊉

a planet that’s red on the inside.

Here’s the analogous metaphysical puzzle: ‘What is this thing you are
calling being red on the inside? What do all these things have in common? You
say that they’re all red on the inside, but what exactly is that? Just as there
are indeﬁnitely many ways to be red, there are indeﬁnitely many ways to be
red on the inside.’

Now this skeptic, remember, is one who was not satisﬁed with the

response that all red things have one thing in common, namely, that they
are red. Why, then, should she be satisﬁed by our response: What all things
that are red on the inside have in common is that they are red on the inside? There’s
an exact analogy here between the alleged problems that arise for things
that are red simpliciter, and things that are red on the inside. If she’s unhappy

Semantic Minimalism and Speech Act Pluralism

162

with (or puzzled by) the response in connection with red, there’s no reason
why she should be happy with the analogous response in connection with
red on the inside. (If she is happy, we want to hear why and be convinced
that the reasons for her satisfaction don’t apply to the original case.)

What this reveals, we think, is that our opponents, i.e., those who claim

there is no such thing as being red simpliciter, should, in order to be con-
sistent, say that there’s a puzzle about the very idea of two or more things
sharing properties or engaging in the same activities. The puzzle they raise
has nothing speciﬁcally to do with redness or dancing. To raise objections
against S1 and S2 is to be committed, in effect, to a form of Metaphysical
Nihilism.

Third stage of reply to Metaphysical Objection Version 1:

Don’t worry about Metaphysical Nihilism

If the choice is between Semantic Minimalism, on the one hand, and Meta-
physical Nihilism, on the other, should we still worry? Is Metaphysical
Nihilism preferable to S1 and S2? We think not. We don’t have much to say
about Metaphysical Nihilism; it is such an extreme view that it hardly
merits a refutation. We’ll settle with several brief critical comments:

1. We know of no good arguments for Metaphysical Nihilism. It’s a

bizarre view and for us to take it seriously, we would have to see some pretty
good arguments in its defense. Absent such arguments it’s a view we don’t
even know how properly to evaluate.

2. Metaphysical Nihilism, if taken seriously, undermines all of natural

science. So, clearly it’s a view everyone should avoid. We don’t see it as an
attractive opponent. Any neutral observer who had the choice between
Semantic Minimalism and Metaphysical Nihilism should choose the
former over the latter.

3. Metaphysical Nihilism, if taken seriously, doesn’t just undermine our

semantics of context sensitivity; it makes all of the philosophy of language
impossible. After all, philosophers of language talk about language, and
about utterances of sentences. We assume that the same expression can be
uttered in different contexts. If Metaphysical Nihilism were true, there
would be no such thing as the same sentence uttered in different contexts.
Again, this is just to highlight a particularly nasty implication of this view;

Semantics and Metaphysics

163

4 Where by ‘Metaphysical Nihilism’ we mean nothing more than, for example, whatever
worries there about red and dancing generalize to everything; i.e., where the entire idea that
two things can have the same properties or engage in the same activities is doubtful.

it’s just another reason for not going down this road, i.e., not to endorse
the objections to S1 and S2.

Final comments about the reply to Metaphysical Objection Version 1

We’re trying to push all these worries onto the metaphysicians. Whatever
story in the end they settle on about what makes something red, that’s what
it takes to satisfy the semantic truth conditions of ‘A is red.’ Whatever in
the end they tell us about dancing, that’s what it takes to satisfy the seman-
tic truth conditions for ‘x dances.’ Semanticists can (and should) just tell
you that the proposition semantically expressed by ‘A is red’ is the propo-
sition that A is red; and that the proposition semantically expressed by ‘A
dances’ is the proposition that A dances. Metaphysicians, however, should
they so choose, can tell you as much as they like about what it is for A to
be red and to dance.

We can imagine three attempts to force us to engage the metaphysical

issues more seriously.

First attempt to force metaphysics on us. Our opponent complains: ‘You’re
claiming that when someone utters a simple sentence like “A is red,” it has
truth conditions that require metaphysical assistance in order to grasp?
You’re saying that speakers don’t know the semantic truth conditions of
their own utterances? Instead, they have to study thousands of years of
metaphysics in order to get a clue about which propositions they express
when they speak, and even then they’ll most likely end up dazed and con-
fused. Isn’t that a peculiar view of what it is to express semantic content?’

Our reply. You don’t need metaphysical assistance in order to know which
proposition you semantically express. It’s the proposition that A is red. You
don’t need metaphysical assistance in order to know the semantic truth
conditions: An utterance of ‘A is red’ is true just in case A is red. That’s the
entire story. You now know all that you need to know. Notice, even if you’re
some kind of veriﬁcationist who thinks he only knows what it is for A to
have red eyes if he knows how to ﬁgure out that A has red eyes, we have a
response to you: First, don’t be a veriﬁcationist; it’s a false view. Second, if
you can’t resist being one, we’ll tell you how to verify the proposition that
A has red eyes: go ﬁnd out whether she has red eyes.

Second attempt to force metaphysics on us. Our opponent complains: ‘This
move you make by appealing to disquotational truth conditions is a cheap
trick, and the way to call your bluff is to ask: Is the proposition semanti-
cally expressed by “Uma Thurman has red eyes” true? Are the interpretive

Semantic Minimalism and Speech Act Pluralism

164

truth conditions of an utterance of “Uma Thurman has red eyes” satisﬁed?
To answer these questions you must resort to what you are calling meta-
physics. So, you can’t avoid doing metaphysics.’

Our reply. This objection is even more deeply confused than the ﬁrst. No
one, we hope, expects a semanticist to tell you the color of Uma Thurman’s
eyes. We’re not in the business of telling you what the world is like. Who
ever thought otherwise? We take it to be trivial that a proposition with a
truth value is expressed by a felicitous utterance of ‘100,000 years ago an
insect moved over this spot.’ We have no idea whether it’s true or not. We
have no idea how to ﬁnd out whether it is true or not.

Third attempt to force metaphysics on us. Our opponent complains: ‘OK, let’s
try this: How are we able to communicate by semantically expressing a
proposition we don’t know the truth value of? You say speech act content
is, typically, different from semantic content, but there has to be a process
that gets us from the proposition semantically expressed to the proposi-
tion(s) communicated by a speech act. How does that work when we can’t
actually settle the truth value of the semantic content? A corollary of this
conclusion: How can speakers go around making assertions by uttering sen-
tences that semantically express propositions they don’t even believe (since
they certainly can’t believe the semantic content unless they have beliefs
about the actual truth values)?’

Our reply. These are interesting questions and we address them in the next
chapter on semantics and psychology.

Summary of reply to ﬁrst part of Metaphysical Objection

If you’re a bit confused by our strategy here, we don’t entirely blame you.
Here’s what we tried to do: We ﬁrst imagined someone objecting to Seman-
tic Minimalism by saying there’s no such proposition as that A is red. We
tried to provide reasons to believe there is such a proposition and that what-
ever is puzzling about such a proposition is not puzzling because of views
you might hold about semantic context sensitivity. The puzzles, if there are
any, are familiar metaphysical ones and the fact that they arise should not
be held against Semantic Minimalism.

You might still be confused by all of this: Why, you might wonder, would

we imagine someone objecting that there’s no such proposition as that A is
red? That’s a silly objection, you might think, not worth all of the pages we
have spent on it.

We hope you have this reaction. There’s a sense in which we agree. Only

when seen in the context of what follows can we justify these spent pages.

Semantics and Metaphysics

165

Our hope is that this discussion will prepare you for the idea that there are
propositions such as that A is ready and that A is tall.

Metaphysical Objection Version 2: What about

Readiness (and Enoughness)?

Reminder: We believe S3–S5:

(S3) An utterance of ‘A has had enough’ expresses the proposition

that A has had enough and it is true just in case A has had enough.

(S4) An utterance of ‘A is ready’ expresses the proposition that A is

ready and it is true just in case A is ready.

(S5) An utterance of ‘It’s raining’ expresses the proposition that it is

raining and is true just in case it is raining.

The uninitiated might respond to our commitments to S3–S5 with an
incredulous stare, expressed, maybe, with something like this: ‘Huh? I know
what it is to be ready for an exam, but to be ready simpliciter? What’s that? I
know what it is to have had enough oysters, but what is it to have had enough
simpliciter? I don’t get it.’ How can there even be such a proposition?

What

claim does it make about the world?

The observant reader will not be surprised that our response has the fol-

lowing structure:

䊉

The alleged metaphysical worry about ‘ready’ and ‘enough’ is no dif-
ferent from the alleged metaphysical worry about ‘red’ and ‘dances.’
S3–S5 introduce no new additional worries.

䊉

Furthermore, as a corollary to the previous point, if there are
puzzles in connection with S3–S5, then they have nothing whatso-
ever to do with Semantic Minimalism, and in particular, they have
nothing to do with what we have to say about the semantics for
context sensitivity. They would arise on any account of context
sensitivity.

First stage of reply to Metaphysical Objection Version 2: The metaphysics

of readiness and enoughness

Think about what people who are ready have in common. To make this
vivid, imagine A’s being ready to commit a bank robbery, B’s being ready to
eat dinner, and C’s being ready to take an exam.

Semantic Minimalism and Speech Act Pluralism

166

5 For an extensive selection of objections along these lines, see the references under ‘Incom-
pleteness Arguments’ in Chapter 2.

Thinking about A, B, and C, you’ve got two options:

(a) You might think, as we do: Well, they have a common relation

they stand in to their respective projects: There’s something in
common between A’s relation to the bank robbery, B’s relation to
the dinner, and C’s relation to the exam. What they have in
common is that they are all ready.

(b) Alternatively, you might think that there’s nothing these people

have in common. The fact that we would describe them as all
being ready for their various projects doesn’t mean that they have
anything whatsoever in common. There’s no state of readiness
that they share with respect to their respective tasks.

We ﬁnd (a) overwhelmingly plausible. It’s not just a pun that we feel com-
fortable describing them all as being ready. They really are all ready. That’s
different from their all being done with the tasks, or excited about them, or
prepared for them, or good at them, etc.

We don’t have a proof that (b) is wrong. But if it were correct, it would

be some kind of interesting metaphysical discovery that is in need of an
argument. In no way will we give this away as a primitive datum. If there is
an issue or puzzle here, it is surely not a consequence of Semantic Mini-
malism. It will arise for anyone who thinks that these people are all ready,
i.e., for anyone who thinks that these people all have something in common.
Suppose you’re a contextualist about ‘ready.’ You certainly could agree with
us that they all are ready for their respective tasks. There’s nothing in con-
textualism as such that blocks this question from arising. If so, the con-
textualist too would have to say something about what exactly it is to stand
in that relation (the relation of readiness) to a task or project. This is not
an issue that magically evaporates at the moment you become a contex-
tualist about ‘ready.’ As far as we can tell, the metaphysical and the seman-
tic issues are orthogonal to one another.

Semantics and Metaphysics

167

6 We hope it’s obvious how this argument goes with respect to ‘enough’:

Again, we’re no experts, but you do something like this: Consider a bunch of people who

all have had enough. For example, one person who has had enough wine, one has had enough
turkey, and one has had enough cocaine. All these people have something in common: They
have all had enough. It is true that they all have had enough of different things, but what they
have in common is that they all have had enough. The metaphysical worry, then, is to deter-
mine in virtue of what is it the case that they all have had enough. That, in a nutshell, is the
problem of the metaphysics of enoughness.

This problem, again, is most certainly not one about which we claim any special expertise,

but here’s a possible start of a solution: Every person who has had enough has had enough of
something that’s contextually salient to him. What all people who have had enough have in
common is that for each there’s something contextually salient that he has had enough of.

Second stage of reply to Metaphysical Objection Version 2:

What can people be ready for?

We have tried to convince you that the problem, worry, or puzzle that arises
in connection with readiness or enoughness, if there is one, is metaphysi-
cal, and not semantic; and that it is independent of any view one might hold
about semantic context sensitivity.

Suppose, despite our best efforts, you remain unconvinced. Suppose

you still think there’s something special about readiness and enoughness.
We don’t know what it might be, but suppose there’s something that
continues to prevent you from being convinced.

In this second stage of our discussion, we will attempt to convince you,

just as we did in connection with redness, that the alleged persistent
problem occurs in connection not just with readiness (or enoughness; see
note 6), but just as clearly (if not clearer) in connection with is ready for an
exam (and has had enough pasta). That is, we will attempt to convince you that
there is nothing special about readiness and enoughness simpliciter. This will
bolster our claim that this worry is entirely general, and not a speciﬁc one
that occurs because of some peculiar view we hold about the semantics of
‘enough’ and ‘ready.’

Suppose our opponent denies that a bunch of people can all be ready

(by being ready for different (kinds of) things (or projects)). To be honest,
we can’t even imagine what her reasons might be for holding this view, but
let that pass. Focus on this question: If a range of people can’t all be ready,
can they at least all be ready for an exam? Is that something a range of people
can have in common? We suppose the answer must be ‘yes.’ Clearly, a bunch
of people can all be ready for an exam. Even a contextualist doesn’t want
to deny that (or if she does, she’s out of the game already). But, look, what

Semantic Minimalism and Speech Act Pluralism

168

The following concern might now be raised: Doesn’t that make it very easy to have had

enough? If that’s all it takes, haven’t we all had enough all of the time? Suppose the answer is
‘yes’ (though we have no idea whether this is correct or not; presumably, it all depends on
doing more serious metaphysics, but suppose it’s correct). When you think real hard about
enoughness, maybe that’s all it takes. If so, then it’s not that hard to have had enough. But if
this is something you ﬁnd problematic, please remind yourself that this is not a problem of
semantics. It is not a problem that arises because of any view one might take about the seman-
tics for ‘enough.’ If this problem troubles you, take it up with your local metaphysician, and
not with a semanticist.

This issue arises for anyone who thinks these folks have all had enough, i.e., that these people

all have something in common. If you’re a contextualist about ‘enough’ and you agree with
us that they all have had enough, you too have this problem. So, it’s not one that has anything
to do with the semantics for ‘enough,’ in particular, it is independent of whether or not ‘A has
had enough’ is context sensitive or not.

was thought problematic and puzzling with respect to readiness simpliciter,
namely, that it seems impossible to get a grip on what these people all have
in common, i.e., readiness, can now quite easily be regenerated with respect
to the property of being ready for an exam. There are indeﬁnitely many ways
of being ready for an exam. If you were puzzled by the claim that a bunch
of people could all be ready, you should, as far as we can tell, be equally
puzzled by the claim that they can all be ready for an exam.

Here are some examples to get your imagination running. Someone can

be ready:

䊉

to give an exam

䊉

to write an exam

䊉

to grade an exam

䊉

to take an exam

䊉

to fail an exam

䊉

to proctor an exam

䊉

for a driving exam

䊉

for a karate exam

䊉

for an oral exam

䊉

for a sailing exam.

All these people have something in common. They are all ready for an exam.

The point of this is embarrassingly simple: Even if you were to deny that

several people can all be ready simpliciter, the worry (if we understand it
right) recurs as soon as you tell us what each person is ready for. So, again,
this bolsters our claim that the alleged worry has nothing at all to do with
the semantics for ‘ready.’

The conclusion is the same as before: Even if you were to deny that

several people can all have had enough simpliciter, the alleged metaphysical
worry recurs as soon as you tell us what it is that they can have in common.
This bolsters our claim that the alleged worry has nothing to do with the
semantics for ‘ready’ or ‘enough.’

Semantics and Metaphysics

169

7 With respect to enoughness, imagine our opponent is a contextualist who denies there is
a general property of having had enough. We can’t imagine what her reasons might be for her
denial, but, again, let that pass. We ask her: If a range of people can’t all have had enough, can
they at least all have had enough pasta? Is that something a range of people can have in
common? We suppose the answer must be ‘yes.’ Clearly, a bunch of people can all have had
enough pasta. But what is supposed to be problematic and puzzling about enough simpliciter
can now be regenerated with respect to having had enough pasta. There are indeﬁnitely many
ways of having had enough pasta. Here are some rough examples to get your imagination
going: a supermarket has enough pasta; a chef making a complicated meal in a restaurant has
enough pasta; a child carrying groceries home has enough pasta; a freight boat has enough
pasta; a car that runs on pasta (new invention, by the way) has enough pasta; and so on.

Third stage of reply to Metaphysical Objection Version 2: Don’t worry

Our opponent could say, as in connection with ‘red’ and ‘dance’: ‘Yes. You’re
right. My worry is completely general. I’m worried about how two things
can be ready, or be ready for an exam, or be ready for an oral exam on
Wednesday, or be ready to proctor a written exam on Friday. It’s all puz-
zling to me.’

Once again Metaphysical Nihilism rears its worrisome head. For reasons

we provided earlier, we don’t have much to say about this view other than
that it is deeply unattractive and that we know of no arguments for it and
that it undermines all of natural science and all of philosophy of language.

Metaphysical Objection Version 3:

Tall Giraffes

In some sense, we hope you think all the previous pages of this chapter were
borderline superﬂuous. Your main worry isn’t that there are no such propo-
sitions as those described in S1–S5. Your main worry, maybe, is that it’s
hard to see what role it plays in communication. Something like that. Fine.
We address that in the next chapter. However, virtually everyone with whom
we have discussed Semantic Minimalism draws a line in the sand when the
topic is comparative adjectives.

Philosophers and linguists alike have deeply entrenched intuitions

about comparative adjectives being context sensitive. Therefore, it is not
surprising that we regularly encounter this sort of response to our claim
that an utterance of ‘A is tall’ expresses the proposition that A is tall and that
it is true just in case A is tall: ‘Propositions like that A is red, that A has had
enough, and that A is ready are odd enough, but the proposition that A is tall
is a deal breaker. If Semantic Minimalism implies that there is such a propo-
sition or, even worse, that we sometimes express it, then that is an unim-
peachable reductio of your view.’

Our reply, in sum, is that this is just but one more version of the same

old metaphysical worry we raised above.

First stage of reply to Metaphysical Objection Version 3:

Metaphysics of tallness

As with redness, readiness, etc., this worry is metaphysical, and not seman-
tic. To show that this is so, we’ll engage in a little bit of tallness-metaphysics.
Consider, for example, the Empire State Building, Mount Everest, and
Osama bin Laden. Ask yourself, what, if anything, do they all have in

Semantic Minimalism and Speech Act Pluralism

170

common? Naturally, one answer is that they are all tall. If that’s so, and it
is, then it triggers the following metaphysical question: What is it in virtue
of which it is the case that they are all tall? Or, what do they all have in
common? Tallness? But what’s that? What does it take for something to
instantiate tallness? Because, as in all matters metaphysical, we are rank
amateurs, we don’t have much to say, but here are four preliminary options
(there are obviously others):

1. For something to instantiate tallness there must be some comparison

class or other with respect to which it’s tall. If that’s all it takes to instan-
tiate tallness, it’s very easy to do so. We take this to be an exceedingly
unpromising account of tallness.

2. It might be that to instantiate tallness it’s insufﬁcient to be tall with

respect to some comparison class. For each object there might be one such
class that’s privileged, say, for natural kinds, the natural kind they belong
to, for artifacts the artifact they instantiate. Since objects belong to many
kinds, work would have to be done to show one of these is privileged.

3. A third option is that the circumstances the object is in at a time t

singles out a comparison class that’s the one the object has to be tall with
respect to in order to be tall at t. Again, work would have to be done to
ﬁgure out how this comparison class is picked out.

4. The property of being tall corresponds to being taller than the average

height for all objects that have height. Since we have no idea how many
objects have heights we have no idea exactly what has this property.

If you agree that there’s a property of tallness – how could you not? –

but have a better account of what it is to instantiate it, that’s ﬁne with us.
The only serious objection we can think of is someone who actually denies
there’s any such thing as tallness. Such cynicism would be to endorse Meta-
physical Nihilism about tallness: i.e., the view that there’s nothing A and B
have in common if A is tall for a G and B is tall for an F. That view is, as
far as we can tell, a rather bizarre view to hold because no one, as far as we
know, denies that there is any such a thing as being tall with respect to some
comparison class. No one can deny there’s such a thing as being tall with respect
to a privileged comparison class or a contextually salient comparison class. If so,
everyone agrees with us that at least for these three accounts of what tall-
ness is, each picks out something that exists.

Semantics and Metaphysics

171

8 A more elaborate discussion of these options would, in some ways, mirror contemporary
debates about knowledge attributions. Both Stanley and Hawthorne propose theories accord-
ing to which knowledge is some kind of interest relative property, but where this does not
necessarily make ‘know’ a context sensitive expression (see Stanley 2003b and Hawthorne
2003). We imagine analogous arguments being made in connection with comparative adjec-
tives. For some suggestions along these lines, see Graff (2000), and a reply by Stanley (2003a).

Imagine someone who thinks ‘tall’ is context sensitive (i.e., imagine being

anyone except us). Presumably, such a philosopher still thinks that many
things can all be tall. A contextualist about ‘tall’ won’t deny that Osama bin
Laden and Mount Everest are both tall. She won’t deny that all tall things
have something in common, i.e. they are all tall. What that shows is that
the alleged metaphysical problem, the problem that’s supposed to show
that it’s absurd to claim that there is a proposition such as that A is tall, isn’t
solved or dissolved by accepting the view that ‘tall’ is semantically context
sensitive.

Second stage of reply to Metaphysical Objection Version 3: Tall for a giraffe

As with redness, readiness, and enoughness, we have tried to convince you
that the alleged puzzle concerning tallness is metaphysical, not semantic.
Still, you may remain unconvinced. Suppose, for reasons we can’t imagine,
you think there’s a special problem for tallness and that this special problem
is indeed semantic, and not purely metaphysical.

In this second stage of our reply, we will again try to establish that if a

problem exists here, it is not speciﬁc to tallness. If there’s an issue here, it
occurs just as much in connection with the property of being tall for an F.
Our rejoinder is dialectically signiﬁcant because our opponents tend to
hold that this alleged problem occurs on account of our commitment to
Semantic Minimalism for ‘tall.’ The ﬁx, according to our contextualist
opponents, is supposed to reside with relativizing comparative adjectives to
comparison classes, i.e., with a commitment to contextualism for ‘tall’ and
other comparative adjectives.

Adjectives like ‘tall’ are to be treated as relational with an unpronounced

place for a comparison class that gets indexed in a context of use. So, in
effect, ‘A is tall’ has in its linguistic representation as it were at LF ‘A is tall
for an F,’ where ‘F’ is an indexical that receives its semantic value in context.
For a sentence like ‘Osama bin Laden is tall,’ in one context of utterance
the indexed comparison class (or property, or whatever) might be NBA
players; and in another the indexed comparison class (or property, or what-
ever) might be Saudi Arabians.

We think there are no sound arguments for this position, and we think

introducing indexed comparison classes or properties or norms into the
semantics for comparative adjectives adds more problems than it allegedly
solves, but let’s leave that aside for the moment. What we want to show
right now is just this:

If the kind of metaphysical problems we have imagined arise for
tallness/‘tall,’ then exactly the same problem arises for tall for an
F/ ‘tall for an F.’

Semantic Minimalism and Speech Act Pluralism

172

Recall, the alleged problem for tallness is that it’s mysterious what it is to be
tall simpliciter: ‘There can be no such thing as tallness simpliciter. To claim
Osama bin Laden, Mount Everest, and the Empire State Building all have
something in common – namely, tallness – is a mistake, and any semantics
that presupposes there could be such a thing must be mistaken. Since
Semantic Minimalism, as you have characterized it, is committed to this
possibility, it should be rejected.’

If this objection issues from anyone who is content with there being such

a thing as being tall for an F, then it is terribly misplaced. Take tall for a giraffe
as an example, i.e., we’re imagining an opponent who thinks that a bunch
of things can all be tall for giraffes. Before proceeding, consider the following
basic giraffe facts: Giraffes have hairy ears. The ﬂeshy part of the ear stops
before the hairs on the ears stop. Not every giraffe can stretch his neck all
the way up; some are old and arthritic. (With help they might be able to
stretch their necks further than without help.) Giraffes can stand on their
back legs and lift their front legs into the air, and thereby push themselves
further up into the air. That makes them longer. They have hoofs, and these
hoofs wear down with usage.

Holding these simple giraffe facts in mind, consider two giraffes, say, A

and B. What would it be for A and B to be tall for giraffes? The problem is
this: There are many ways to be tall for giraffes. For starters, there are in-
deﬁnitely many ways to measure the tallness of giraffes. Consider these few
illustrations. A giraffe’s height can be measured

䊉

from the bottom of a hoof to the ﬂeshy tip of his ear with a self-
stretched neck;

䊉

from the bottom of a hoof to the tip of his nose with a self-stretched
neck;

䊉

from the bottom of a hoof to the hairy tip of an ear with a self-
stretched neck;

䊉

from the bottom of a hoof to the tip of his nose when standing on
his back legs with his front legs lifted into the air;

䊉

all of the above, with an artiﬁcially stretched neck, i.e., by a machine
or something else that can stretch the neck out further than the
giraffe can by herself. (Remember, some giraffes are arthritic, and
have very stiff necks.)

Then, of course, there’s the question of which group of giraffes we are to
compare any given giraffe to. Here are but a few options:

䊉

all living giraffes

䊉

a stereotypical giraffe

䊉

French giraffes

Semantics and Metaphysics

173

䊉

all giraffes that have ever lived, are alive, and will ever live

䊉

all possible giraffes

䊉

all giraffes in the vicinity of a certain giraffe.

Then, of course, there’s the question of the (optimum) conditions under
which to measure a particular giraffe (holding the method of measurement
and the comparison class ﬁxed). Here are but a few of indeﬁnitely many
options:

䊉

right after a bath (giraffes shrink a bit after having taken a bath)

䊉

right after a long walk (their hoofs wear down)

䊉

when dead (again, death shrinks us all)

䊉

when hungry (they tend to stretch their necks further)

䊉

when pregnant (their necks are rendered less ﬂexible).

Now ask yourself: What is it to be tall for a giraffe? What is giraffe-tallness? It
all depends on which giraffes you compare any given giraffe to, how you
measure it, the conditions of the giraffe when being measured, and so on.
The ‘and so on’ is vital. There are, we conjecture, no limits on the different
variations on giraffe-tallness.

Fast giraffes. The objection: There’s no such thing as fastness simpliciter. At
this stage, we don’t need to repeat what we hope is now a familiar point
and counter-point. Our critics claim a comparison class is required. But
were we to introduce one, say, the class of all giraffes, would that really help
us ﬁgure out what it is to be fast for a giraffe? Is it any easier to understand
fast for a giraffe than it is to understand fast simpliciter?

Consider the proposition that A is fast for a giraffe. Ask yourself, as we did

for the proposition that A is fast, what it is for A to be fast for a giraffe, or,
to be a fast giraffe? Is that any less puzzling than being just fast? It seems
to us the answer is obviously ‘no.’ Is it to be a

䊉

fast runner?

䊉

fast walker?

䊉

fast eater?

䊉

fast in the morning?

䊉

fast when tired?

䊉

fast when sick?

䊉

fast when hungry?

䊉

fast when scared?

䊉

fast when sleeping?

䊉

fast after snorting cocaine?

䊉

fast when pursuing a female giraffe?

Semantic Minimalism and Speech Act Pluralism

174

Telling us what the comparison class is doesn’t tell us what kind of fast-
ness we are supposed to be comparing, and it doesn’t tell us what the con-
ditions are under which we are to check for that fastness.

Suppose the comparison class for tall giraffes is all living giraffes with

normal health while eating. But eating what? When they like the food or
don’t like it? Etc. No matter how you limit the comparison class, it will be
insufﬁcient. More importantly: The objection here is supposed to be that
Semantic Minimalism is absurd for a certain reason. We have shown that
this reason applies quite generally. That it does is important because it
makes it obvious that there is no reductio here. There’s, maybe, a substan-
tive metaphysical disagreement, but not one that should carry any impli-
cations for how we do semantics.

The alleged metaphysical worry was about what all tall things have in

common. Our opponent claims there is nothing they have in common,
since it all depends on the relevant comparison class. What we have shown
is that this worry is not solved by introducing a comparison class. The
‘What do they have in common?’ question can be raised in exactly the same
way when asking what it is that all things that are tall for a giraffe have in
common.

We conclude that the alleged problem has nothing speciﬁcally to do with

Semantic Minimalism about ‘tall.’

Third stage of reply to Metaphysical Objection Version 3: Don’t worry

As with the previous two parts of the Metaphysical Objection, we might, of
course, be facing an opponent who’s a Metaphysical Nihilist. Again, for
reasons provided earlier, we don’t take Metaphysical Nihilism all that
seriously. What we have to say about it in this case is no different than
what we had to say about it in connection with redness, readiness, and
enoughness.

We conclude that the issues that motivate the objections discussed in

this chapter cannot be resolved by doing semantics, in particular they
cannot be resolved by adding context sensitivity to sentences containing
words like ‘red,’ ‘ready,’ and ‘tall.’ The real worry, we suggest, is either a
metaphysical one (one that different semantic theories are neutral with
respect to) or a worry about how semantic content functions in communi-
cation (an issue we address in the next chapter).

Semantics and Metaphysics

175

176

C H A P T E R 1 2

Semantics and Psychology

According to our view (Semantic Minimalism combined with Speech Act
Pluralism), speakers use sentences to make claims, assertions, suggestions,
requests, claims, statements, or raise hypotheses, inquiries, etc., the contents
of which can be (and typically are) radically different from the semantic con-
tents of (the propositions semantically expressed by) these utterances. The
speech act content (i.e., what was said, asserted, claimed, asked, etc.) depends
on a potentially indeﬁnite range of facts about the speaker, his audience,
their shared context, the reporter (i.e., the person recounting what was said),
the reporter’s audience, and their shared context. These facts have no
bearing on the semantic content of the utterance.

In this and the next chapter we outline the relationship between seman-

tic content and speech act content. In this chapter we discuss two versions
of what we will call the Psychological Objection to Semantic Minimalism. Vari-
ations on this objection have been raised by, among others, the so-called
Relevance Theorists. We focus on two versions of their argument, both found
in the writings of Carston (1988, 2002, 2004), but closely related to argu-
ments found in Recanati (2002, 2004), Sperber and Wilson (1986), and in
King and Stanley (forthcoming). We ﬁrst respond to these arguments
and then at the end of the chapter we reverse the charge: It is contextualists
who have a psychologically unrealistic notion of content, and not Semantic
Minimalists.

In Chapter 13 we elaborate on, and defend, Speech Act Pluralism (a view

we so far have just brieﬂy sketched, but constantly referred to).

The Psychological Objection Version 1

Here’s a potential worry about Semantic Minimalism: What communica-
tors actually care about in a discourse exchange is the speech act content

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

and only the speech act content. What they care about is what the speaker
said, asserted, claimed, stated, suggested, asked, etc. If the semantic content
is, so to speak, always hidden, if it never surfaces, then what purpose does
it serve? Isn’t it just an idle wheel? This objection might seem particularly
worrisome given what we ourselves said about the propositions semanti-
cally expressed in Chapter 11: they are peculiar, to say the least.

What we are calling Carston’s Psychological Objection is a version of

exactly this sort of worry. The conclusion of this argument is expressed in
the following passage:

So there simply does not seem to be any wholly semantic notion of
‘what is said’ . . . Of course, various minimalist notions of ‘what is
said’ can be deﬁned; they are ‘minimalist’ in that they keep pragmatic
contributions to a minimum, for instance, allowing just reference
assignment and disambiguation, or just saturation, or just whatever
it takes to achieve truth-evaluability. But none of the results of these
subtractions from the full range of pragmatic processes involved in
explicature derivation has been shown to have any cognitive reality.
Given decoded linguistic type meaning and a pragmatic processor
which takes this as its input in deriving what is communicated (expli-
catures and implicatures), it is difﬁcult to see a role for a further
notion of ‘what is said,’ whether subpropositional or minimally
propositional, which articulates a meaning that lies somewhere
between linguistic meaning and explicature. (Carston 2004, pp. 28–9)

Recanati (2001) makes the same point:

That minimal notion of what is said is an abstraction with no psy-
chological reality, because of the holistic nature of speaker’s meaning.
From a psychological point of view, we cannot separate those aspects
of speaker’s meaning which ﬁll gaps in the representation associated
with the sentence as a result of purely semantic interpretation, and
those aspects of speaker’s meaning which are optional and enrich or
otherwise modify the representation in question. They are indisso-
ciable, mutually dependent aspects of a single process of pragmatic
interpretation. (Recanati 2001, p. 88)

We focus on the version of this argument presented in Carston (1988). It’s
helpful, we think, to present two versions of her argument. At ﬁrst we
present the objection exactly as she does. She directs her objection at what
she calls the ‘Gricean view.’ Since what’s called the ‘Gricean view’ in the lit-
erature is often also called ‘Minimalism,’ it’s important we clarify exactly
how Semantic Minimalism differs considerably from this view and why the

Semantics and Psychology

177

objections Carston raises against the Gricean view would be off target were
they directed at us. Elucidating why they would be serves two purposes:
ﬁrst, it affords us an opportunity to explain exactly how Semantic Mini-
malism differs from the Gricean view, and secondly, it responds to
Carston’s objection. However, once we deﬂate this version of her objection,
we will take up an obvious variation on it directed squarely at us.

Carston’s Psychological Objection

Version 1

Carston’s argument against the view that she calls Gricean Minimalism can
be presented in ﬁve steps, C1–C5.

(C1) If the what-was-said is cognitively redundant, then it is not

‘plausible’ (Carston 1988, p. 39; see also p. 40) that the what-
was-said plays a role in the mental life of communicators.

(C2) If an implicature can play all the roles in reasoning that the

what-was-said can play, then ‘the explicature has no function in
the mental life that cannot be played by the implicature’ (1988,
p. 38), and hence, is cognitively redundant.

(C3) On Grice’s construal of the distinction between what-was-said

and implicatures, the latter can often play all the roles in rea-
soning that the what-was-said can play.

(C4) So: The Gricean what-was-said will often be cognitively redun-

dant.

(C5) So: It is in many cases implausible that the Gricean what-was-

said plays a role in the mental life of communicators.

A framework that gives you a what-was-said that plays no role in the mental
life of communicators should, Carston claims, be rejected. To endorse such
a framework is ‘to ignore the nature of communication and of cognition
in general in the interest of a formal principle which has absolutely no
bearing on human psychology’ (1988, p. 40).

Carston’s discussion of sentence (1) illustrates her argumentative

strategy.

(1) The park is some distance from where I live.

According to Carston, the Gricean what-was-said expressed by an utterance
of (1) is the proposition that there is some distance or other between the speaker’s
home and the park. In most contexts, this claim would obviously not be worth
making; it would be entirely superﬂuous and trivial. What the speaker
wants to communicate, for instance, could be that the distance between the
speaker’s home and the park is longer than what the hearer expects it to be.

Semantic Minimalism and Speech Act Pluralism

178

This more pragmatically enriched proposition would in Grice’s frame-

work have to be classiﬁed as an implicature, derived from the minimalistic
Gricean what-was-said using the maxims of conversation. To elicit the cog-
nitive redundancy of the Gricean notion of what-was-said, Carston asks:

But what function then does the explicature [i.e., the Gricean what-
was-said] have in the mental life of the hearer? It is entailed by the
implicature: if the park is further away from my house than the hearer
had been assuming, it follows that it is some distance or other from
my house. When this entailment relation holds between putative
implicature and explicature the probability of functionally indepen-
dent propositional forms is very low. (Carston 1988, p. 40)

Cognitively redundant Gricean what-was-saids, according to Carston, are
ubiquitous. Other cases she discusses include (2)–(6), where (a) is a sen-
tence, (b) an approximation of the Gricean what-was-said expressed by a
certain utterance of (a), and (c) a proposition that Grice, according to
Carston, would classify as an implicature of that utterance.

(2) (a) I went to the exhibition and ran into John.

(b) I went to the exhibition and I ran into John (somewhere or

other).

(3) (a) She took the gun, walked into the garden, and killed her

mother.

(b) She took the gun and she walked into the garden and she

killed her mother (with no indication of a temporal ordering,
no indication of where the killing took place, or how the
killing was done).

killed her mother with the gun in the garden.

(4) (a) Mr. Smith has been insulted and he is going to resign.

(b) Mr. Smith has been insulted and Mr. Smith is going to resign

(with no indication of the reason why he will resign).

(5) (a) Mr. Jones has three children.

(b) Mr. Jones has at least three children.
(c) Mr. Jones has no more than three children.

Semantics and Psychology

179

1 She continues, ‘What the hearer is going to remember from this utterance is some estimate
of the distance involved, not the fact that there is a distance, and any inferences he draws on
the basis of the utterance will involve the proposition concerning this amount of distance,
rather than the basic proposition concerning the existence of a distance’ (Carston 1988, p. 40).
2 Note: (5c) does not entail (5b).

(6) (a) I’ve had breakfast.

(b) I’ve had breakfast at some point or other.
(c) I’ve had breakfast recently.

In all of these cases, the Gricean implicature (tentatively characterized by
(c)) entails the Gricean what-was-said (i.e., (b)). When this is so, there is,
according to Carston, nothing the what-was-said can do that the implica-
ture cannot do as well; so, there is no function for the Gricean what-was-
said to serve:

whatever role the explicated assumption [her expression for the
Gricean what-was-said] might play in chains of reasoning the impli-
cated assumption could also play. . . . in other words, whatever the
explicature can do, so can the implicature plus more. (Carston 1988,
p. 38)

She concludes that minimalism should be rejected:

It is difﬁcult to see any justiﬁcation for a principle along the lines of
‘Use the maxims just in order to get a minimal truth-bearing vehicle’.
This is to ignore the nature of communication and of cognition in
general in the interest of a formal principle which has absolutely no
bearing on human psychology. (Carston 1988, p. 40)

Why this version of Carston’s objection does not apply to our

version of Semantic Minimalism

As formulated, Carston’s objection to Gricean Minimalism is completely
off target as an objection to our version of Semantic Minimalism. Recall:

䊉

We agree with her that you need a contextually shaped content to
generate implicatures in all of the cases she discusses. That is, we
agree that the (b) part of her examples would be insufﬁcient to gen-
erate implicatures. What’s needed in order to derive the implicature
in these cases is a contextually shaped content, i.e., a contextually
shaped what-is-said.

Semantic Minimalism and Speech Act Pluralism

180

3 Further examples, from her book (Carston 2002), include:

(7) It will take some time to understand relevance theory.
(8) Something has happened.
(9) There’s nothing on telly tonight.

䊉

Our semantic content (the proposition semantically expressed) is
not identical to what the speaker said. It cannot, and is not meant
to, play the roles that what the speaker said can play. It is not meant
to serve functions (cognitive or otherwise) that the speech act
content can serve.

䊉

More generally: We are happy to agree with Carston that an appro-
priate notion of what the speaker said must allow for contextual
inﬂuences that go far beyond what the speaker said.

We are sometimes asked: Aren’t you just disguised Radical Contextualists/
Relevance Theorists? The answer is resoundingly ‘no’ for at least ﬁve
reasons:

(a) We think, and the Relevance Theorists deny, that there is a

minimal semantic content or proposition that is semantically
expressed by (almost) every utterance of a well-formed English
sentence. This proposition is not a ‘skeleton’; it is not fragmen-
tary; it’s a full-blooded proposition with truth conditions and a
truth value. (For elaboration, see Chapter 11.) This is a substan-
tive disagreement about the metaphysics of content.

(b) We think, and the Relevance Theorists deny,

that this minimal

semantic content is an essential part of all communicative inter-
actions. The minimal semantic content has a function in the cog-
nitive life of communicators that no other content can serve.
(This is what we argue later in this chapter.)

tic content has a psychological role that no other ‘level of content’
can ﬁll. (Again, we argue for this later in this chapter.)

(d) We think, and the Relevance Theorists deny, that theories that

do not recognize minimal semantic content are empirically
inadequate and internally inconsistent. (See Chapters 7–9 for
elaboration.)

(e) Finally, we should mention here, though we only discuss it in the

next chapter, that we emphatically reject theories that are ‘Origi-
nal Utterance Centrist.’ All Relevance Theorists are Original
Utterance Centrists. (For elaboration, see SPAP

in Chapter 13.)

Carston’s Psychological Objection

Version 2

In the light of all that, here’s a second version of Carston’s objection:

Semantics and Psychology

181

4 See, for example, the quotes from Recanati and Carston above.

(C1

¢) If the minimal semantic content is cognitively redundant, then

it is not ‘plausible’ that it plays a role in the mental life of
communicators.

(C2

¢) If the speech act content (what-is-said and what is implicated)

can play all the roles in reasoning that the minimal semantic
content can play, then the semantic content has no function in
the mental life of communicators that cannot be played by the
speech act content, and hence, is cognitively redundant.

(C3

¢) The speech act content can often play all the roles in reasoning

that the minimal semantic content can play.

(C4

¢) So: The minimal semantic content will often be cognitively

redundant.

(C5

¢) So: It is in many cases not plausible that the minimal semantic

content plays a role in the mental life of communicators.

As with the ﬁrst version, this objection also concludes: A framework that
gives you a semantic content that plays no role whatsoever in the mental
life of communicators should be rejected. To endorse such a framework is
‘to ignore the nature of communication and of cognition in general in the
interest of a formal principle which has absolutely no bearing on human
psychology’ (Carston 1988, p. 40).

Why the Psychological Objection Version 2 fails

What we’re about to say in the next few paragraphs might seem very simple
(we hope it is) and obvious (we hope it is), but we can’t overemphasize its
importance, so please pay close attention. First, we remind you of some
basic facts about communication. Then we respond directly to this second
version of the Psychological Objection.

1. Basic facts about speakers and audiences who share a context. We know

the following about speakers and audiences who share a context:

䊉

Speakers are sometimes wrong (or have incomplete information)
about their audience. They can be wrong (or have incomplete infor-
mation) about many things concerning their audience, including
each of the following:

Semantic Minimalism and Speech Act Pluralism

182

5 This entire section is based on there being a clear notion of a shared context. We doubt
there is one, but we’ll place our reservations to the side for the sake of argument. If there’s no
such thing as a shared context, it makes things even harder for our opponents.

䊊

what the audience believes and knows;

䊊

what the audience remembers about previous conversations;

䊊

how the audience has interpreted previous conversations;

䊊

how the audience perceives their shared environment; and

䊊

what the audience believes about the speaker.

䊉

Audiences are sometimes wrong (or have incomplete information)
about speakers. They can be wrong (or have incomplete informa-
tion) about each of the following:

䊊

what the speaker believes and knows;

䊊

what the speaker remembers about previous conversations;

䊊

how the speaker has interpreted previous conversations;

䊊

how the speaker perceives their shared environment; and

䊊

what the speaker believes about the audience.

䊉

Audiences and speakers are both often wrong (or have incomplete
information) about the following facts about the context that they
ﬁnd themselves in:

䊊

what their perceptual environment is; and

䊊

what the contents of preceding conversations were.

These mistakes can transpire for a variety of reasons, the most
normal of which is boredom. Boring conversations distract both the
audience and speaker. Distracted people don’t pay close attention
to what’s going on around them.

䊉

Speakers and audiences know that they can be wrong and have
incomplete information about each other in the ways speciﬁed
above. They know that there are, literally, innumerable sources of
confusion and misunderstanding. They know that failed commu-
nicative efforts are ubiquitous. They know that communicative
success is a precarious ideal.

2. Basic facts about speakers and audiences who do not share a context.

Sometimes the audience of an utterance doesn’t share a context with the
speaker. This can happen in any of several ways, the most salient of which
is the reproduction of a speech act. Printed books, for example, often have
this effect. Take you, our reader: We have no idea who you are; we know
next to nothing about your beliefs; we don’t know anything about your per-
ceptual environment; all we know is that it is not shared with us (you’re not
here in 5stC with us). We are aware of no known shared previous conver-
sations with you. Yet, nonetheless, we have an audience for this book and
you’re it.

Another typical device through which a speech act can reach an audience

in another context is through indirect quotation. This is the device through
which a speaker S says in a context C to an audience A what another speaker
S

¢ said in another context C¢ to another audience A¢.

Semantics and Psychology

183

In these cases the sources of confusion mentioned in section (a) above

are multiplied. The added complications should be obvious, so we won’t go
into them in any detail. Sufﬁce it to say, however, that there is not even the
illusion of a shared context in such cases.

3. Basic facts about inter-contextual content sharing. Finally, three facts about

how the same contents can be expressed in different contexts:

(1) People can and often do say the same thing in different contexts.

For example, people in different contexts can say that Napoleon
was short.

(2) No two contexts are the same with respect to the parameters that

ﬁx content according to contextualists, e.g., the intentions are not
the same; the background knowledge is not the same; previous
conversations are not the same; what’s normal is not the same;
and so on.

(3) It is possible to say in a context C that people in a range of con-

texts C

–C

said the same thing, e.g., there are true reports of the

form ‘They all said that Napoleon was short’ made in a context C
about different speakers’ utterances in contexts C

–C

Remember from Chapter 8: if someone denies (1)–(3), we don’t want to talk
to her or about her (because she doesn’t think she can say what we say, so
she can’t deny what we say, and (according to her) we can’t say what she
said, and so we can’t say that we disagree with what she said).

The cognitive role of minimal semantic content

What’s the cognitive role of minimal semantic content? The answer should
be (almost) self-evident by now:

(F1) Speakers know that their audience can be (and often are) mis-

taken (or have incomplete information) about the communica-
tion-relevant facts about the context of utterance (i.e., the facts
listed in Bezuidenhout’s (i)–(vi) discussed in Chapter 8). The
proposition semantically expressed is that content the speaker
can expect the audience to grasp (and expect the audience to
expect the speaker to expect them to grasp) even if they have mis-
taken or incomplete communication-relevant information.

(F2) Audiences know that the speaker can be (and often is) mistaken

(or has incomplete information) about the communication-
relevant facts about the context of utterance (i.e., the facts listed
in Bezuidenhout’s (i)–(vi)). The proposition semantically
expressed is that content the audience can expect the speaker to

Semantic Minimalism and Speech Act Pluralism

184

grasp (and expect the speaker to expect the audience to grasp,
etc.) even if she has such mistaken or incomplete information.

(F3) The proposition semantically expressed is that content which

can be grasped and expressed by someone who isn’t even a par-
ticipant in the context of utterance.

(F4) The proposition semantically expressed is that content which

speakers and audiences know can be transmitted through indi-
rect quotation or reproduction (in the form of tapes, video
recordings, etc.) to those who ﬁnd themselves in contexts radi-
cally different from the original context of utterance.

In short: the proposition semantically expressed is our minimal defense
against confusion, misunderstanding, mistakes and it is that which guar-
antees communication across contexts of utterance.

Possible counter-reply. We expect this sort of reply from Carston (and other
Relevance Theorists): ‘Hold it: You’re saying that the minimal semantic
content is a “shared fallback content” and that this content serves to guard
against confusion and misunderstandings. But given what you’ve told us
about minimal propositions, how could they serve that purpose? Consider,
for example, an utterance of “A is red.” Suppose a speaker utters this sen-
tence in order to communicate that A is red when peeled, washed, and
observed under normal daylight conditions (or something like that). That’s
what the speaker is trying to say. That’s what the speaker is asserting. Let’s
assume, as you (almost) granted in Chapter 11, that the proposition that A
is red is true as long as A is red on some surface, under some conditions.
Everything is red on some surface under some conditions. So, that propo-
sition is trivially true. How does it help an audience to know that this
minimal proposition was expressed? It’s not what the speaker asserted. It’s
trivially true. What help could it be to know that this proposition was
expressed?’

Our response is simple: It is a starting point. Suppose, for the sake of

argument, that the proposition that A is red is trivially true (not something
we have endorsed, simply something we speculated about in Chapter 11).
The audience knows that the speaker is talking about A and its redness, not,
for example, about oysters, France, or Relevance Theory. There’s a lot of
stuff to talk about in the universe. The proposition semantically expressed
pares it down considerably. Knowledge that this proposition was semanti-
cally expressed provides the audience with the best possible access to the
speaker’s mind, given the restricted knowledge they have of that speaker. It
is trivial that A is red on some surface or other under some condition or
other. The audience can assume that the speaker knew that this was trivial
and was not interested in conveying such trivialities with his utterance and

Semantics and Psychology

185

can, therefore, infer that there’s work to be done in order to ﬁgure out
exactly what the speaker was trying to communicate. In general, audiences
know what to look for in such situations; they know what kind of informa-
tion would help narrow down more closely what the speaker wanted to
communicate.

To sum up this reply to the Psychological Objection, here is how we are

responding to the following charge from Recanati. He writes of minimal
propositions:

Let the semanticist use it if he or she wants to, provided he or she
agrees that . . . the minimal proposition has no psychological reality.
It does not correspond to any stage in the process of understanding
the utterance, and need not be entertained or represented at any point
in that process. (Recanati 2001, p. 89)

If there’s a difference between having a cognitive function and corre-
sponding to a stage in processing or having psychological reality, we don’t
know what that difference consists in. If F1–F4 aren’t sufﬁcient to ‘corre-
spond to any stage in the process of understanding the utterance, and need
not be entertained or represented at any point in that process,’ then we
don’t know what is.

In some sense, what we’ve said here is a bit of a stab in the dark since

we’re not at all sure what the contextualists have in mind by their require-
ment that the explicature be psychologically real. We think what we have said
is sufﬁcient to render the propositions semantically expressed, according to
Semantic Minimalism, psychologically real, but we’re genuinely confused
by the requirement, since we have no idea how contextualists themselves
can satisfy their own requirement. The rest of this chapter is devoted to an
elaboration of that point.

The Psychological Objection Reversed (or, Why

Recanati’s Account of What-is-Said doesn’t

Satisfy his Own Availability

Principle)

Suppose we focus, as contextualists tend to, on the context of the speaker
and her audience. The factors that ﬁgure into ﬁxing the contextualist’s
what-was-said (i.e., the explicature) include, inter alia, (i)–(iv):

(i) information triggered in the speaker and the audience by prior

discourse contents;

Semantic Minimalism and Speech Act Pluralism

186

(ii) information conversational partners share about each other;
(iii) information the conversational partners have acquired through

observation of their mutual perceptual environment;

(iv) information conversational partners have about each other’s

purposes and abilities (e.g., whether the person is being deceitful
or sincere, whether the person tends to verbosity or is a person
of few words, etc.).

This in no way exhausts the facts that, according to contextualists, are
content determinants, but what we have to say about (i)–(iv) generalizes.
Here’s the problem: Suppose (i)–(iv) are factors that ﬁx the explicature of
an utterance u of some sentence S. Now (i)–(iv) involve the mental states of
several people (i.e., the speaker and her audience). None of the participants
knows all the relevant facts about all the other participants. For example,
Cappelen doesn’t know all the information triggered in Lepore by our many
previous discussions; Lepore doesn’t know what information Cappelen has
about him (Cappelen undoubtedly knows things about Lepore that Lepore
doesn’t even know he knows); Cappelen doesn’t always know what Lepore
pays attention to in their shared perceptual environment; etc.

Our point here is rather obvious: If the explicature is ﬁxed by these sorts

of facts, then no one of the participants has direct access to the explicature.
It is ﬁxed intra-personally, and so, there’s no reason to think the resulting
content is ‘represented’ at any stage of that person’s processing of the rel-
evant utterance. There is no reason to think that the resulting proposition
is psychologically real.

One way (the only way?) to avoid this troubling problem is to let the

content determining facts about the context of utterance be the speaker’s
beliefs about (i)–(iv), i.e., the speaker’s beliefs about what is shared infor-
mation, shared perceptual input, etc.

This is something the speaker has

access to, and if those beliefs shape the context, then our objection doesn’t
apply. But, of course, this is a deeply unattractive view since it, for all prac-
tical purposes, makes what was said inaccessible to any audience. And so
this is not a view contextualists would want to endorse. (If you let it be
the audience’s beliefs about (i)–(iv) that determine the content, then
the speaker doesn’t know what she has said. This view is obviously equally
unattractive.)

Recanati discusses a version of this objection and his reply illustrates just

how hard it is for contextualists to satisfy their own psychological reality
requirement. In particular, it illustrates why Recanati can’t satisfy his own
Availability Principle (his version of the Psychological Requirement).

Semantics and Psychology

187

6 Alternatively: the audience’s beliefs about (i)–(iv); the point we make in what follows still
applies.

Hence my ‘Availability Principle’ (Recanati 1993, p. 248), according to
which ‘what is said’ must be analysed in conformity to the intuitions
shared by those who fully understand the utterance – typically the
speaker and the hearer, in a normal conversational setting. I take the
conversational participants’ intuitions concerning what is said to be
revealed by their views concerning the utterance’s truth-conditions. I
assume that whoever fully understands a declarative utterance knows
which state of affairs would possibly constitute a truth-maker for that
utterance, i.e., knows in what sort of circumstance it would be true.
(Recanati 2004, p. 14)

Recanati’s theory, based on his Availability Principle, is supposed to be

an alternative to theories according to which the explicature (i.e., content,
what-is-said) is not psychologically accessible. Recanati’s idea is that, since
his what-is-said corresponds to the speaker’s and audience’s intuitions about
what is said, it will ﬁgure in the process of understanding (an utterance of)
the sentence. But he then raises the following worry:

Have we not equated what is said with their [i.e., the speaker’s and
audience’s] understanding of what is said?

We have not. We have equated what is said with what a normal

interpreter would understand as being said, in the context at hand.
A normal interpreter knows which sentence was uttered, knows
the meaning of that sentence, knows the relevant contextual facts
(who is being pointed to, etc.). Ordinary users of the language
are normal interpreters, in most situations. They know the relevant
facts and have the relevant abilities. But there are situations. . . . where
the actual users make mistakes and are not normal interpreters.
In such situations their interpretations do not ﬁx what is said. To
determine what is said, we need to look at the interpretation that a
normal interpreter would give. This is objective enough, yet remains
within the conﬁnes of the pragmatic construal. (Recanati 2004,
pp. 19–20)

But wait! What’s normal is not something that speakers have psychological
access to.

What’s normal need not ‘be in the speaker’s mind when the

sentence is understood’; it certainly needn’t ﬁgure in any psychological
processes that the speaker goes through when understanding (an utterance
of) a sentence. This is so for several obvious reasons, (a)–(d) being perhaps
the most obvious ones:

Semantic Minimalism and Speech Act Pluralism

188

7 Not consciously and certainly not unconsciously.

(a) A speaker can be abnormal, but think that she is normal.
(b) A speaker might know that she is not normal, but not know what

normal is.

(c) A speaker might think that she is not normal, but be normal.
(d) More generally: Even for speakers who are normal and know that

they are normal, they might not know what counts as a normal
understanding of some speciﬁc feature of a context that they
happen to ﬁnd themselves in.

A lot of situations have no ‘normal’ set of expectations associated with
them. Suppose Lepore meets someone in the 5th Street café on a hot New
York City summer day. What ‘normality’ are we looking for? Normal for
Lepore when he’s talking to strangers in a café on 5th Street in New York
City on a hot summer day? There’s no such thing!

In other words: If what’s normal, in part, determines what-is-said, and if

what’s normal is not represented at any stage in the processing of the utter-
ance (and it can’t be if any one of (a)–(d) is true), then the resulting what-
is-said cannot be so represented. Then, we suppose (though, as we have
admitted, we’re not sure we entirely understand the contextualists here),
Recanati’s what-is-said is not psychologically real.

In sum: the Semantic Minimalist has a response to the Psychological

Objection. It is the contextualist who surprisingly does not.

Semantics and Psychology

189

190

C H A P T E R 1 3

Speech Act Pluralism

So far we have

(a) outlined what we think are the sources of semantic context

sensitivity;

(b) refuted the alternative accounts – MC and RC;
(c) told you what we can (admittedly, not much) about the nature of

propositions semantically expressed; and

(d) responded to objections to Semantic Minimalism, in particular,

to the view that we express minimal propositions.

Throughout, we have said that what an utterance says, states, claims, etc.
differs from the proposition it semantically expresses. In this chapter we
elaborate on that claim.

We call the view we defend Speech Act Pluralism (SPAP, for short). SPAP is

not really a theory; it’s a collection of observations, one of which is that
there can be no systematic theory of speech act content. Since we’re inclined
to insist that a theory has to be systematic, we refuse to call SPAP a theory.

Our presentation and defense of SPAP has four parts:

(1) First, we register some methodological observations: since SPAP

is a collection of observations about the nature of speech acts, we
start by outlining, in general terms, how we think speech acts
should be investigated.

(2) We present what we call the Central Observation. This provides the

basis for our nontheory theory of speech act content.

(3) We then present observations about speech act content. These

eight observations constitute SPAP.

(4) We distinguish SPAP from four related views (some of which we

ourselves once held) and explain why we don’t endorse these
views.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

Methodology: Nothing is Hidden

We are naive about speech act content; we take our nontheoretic beliefs and
intuitions about what speakers say, assert, claim, etc. at face value (unless
given overwhelming reason to do otherwise). We’re naive for a reason:
Speech act content isn’t deeply hidden somewhere, where only a theorist
can excavate it. That would undermine its purpose. What speakers say,
claim, assert, etc. is what we hear when we listen to them. It is what we
respond to, question, laugh at, take seriously, ignore, agree or disagree with.
It is not concealed; it is not something requiring an investigation to
uncover.

Intuitions and nontheoretic assumptions about speech act content can,

of course, be overridden should important theoretical considerations lead
us to reject speciﬁc intuitions or general considerations. We are, however,
unaware of any such intuitions or considerations relevant to the cases we
discuss here.

How, then, do we identify the clear cases we take at face value? Well, we

just sit here at the café on 5th Street and think about what people say, and
what people say about what people say. Then we check with others sitting
at adjoining tables to ensure our reactions aren’t idiosyncratic. And ﬁnally
we check whether there are any overwhelming reasons for rejecting these
observations.

The Central Observation

Consider this verbatim transcript of an utterance, the so-called ‘Smoking
Gun’ utterance (‘. . .’ indicates pauses):

When you get in these people, when you get these people in, say:
‘Look, the problem is that this will open the whole, the whole Bay of
Pigs thing, and the president just feels that,’ ah, without going into
the details . . . don’t, don’t lie to them to the extent to say there is no
involvement, but just say this is sort of a comedy of errors, bizarre,
without getting into it, ‘the president believes that it is going to open
the whole Bay of Pigs thing up again, and ah because these people are
plugging for, for keeps and that they should call the FBI in and say
that we wish for the country, don’t go any further into this case.’
Period. That’s the way to put it, do it straight.

Speech Act Pluralism

191

All of this is not to deny that confusion, obscurity, misunderstanding, and related pheno-

mena are not ubiquitous. They are. But those cases illustrate our point even more clearly.

Let’s reﬂect on what is said by this utterance. We want a naive description
of what it says, the sort of description you would give if you weren’t encoun-
tering it with a philosophical axe to grind. Notice ﬁrst the following:

(a) This quote is typical in that almost none of it includes a fully

grammatical sentence. Indeed, very few well-formed English sen-
tences ever get uttered.

(b) As a result, to ascertain what’s said, you ﬁrst have to reconstruct

utterances to a point where they express thoughts. There are
many ways to achieve this end, as illustrated by this quote. No one
way is uniquely correct.

it), it obviously helps to know some basic facts about it, such as
who the speaker and audience are and where the utterance took place. It
helps, for example, to know that the speaker was President
Richard Milhous Nixon, the thirty-seventh President of the
United States; that the audience was R. H. Haldeman (Nixon’s
Chief of Staff); that the locution ‘these people’ refers to one or all
of CIA Director Richard Helms and his deputy, General Vernon
Walters (a longtime associate of the President’s), and FBI Acting
Director Pat Gray; that the conversation takes place in the Oval
Ofﬁce, June 23, 1972, from 10:04 to 11:39 a.m. (It’s from a tran-
script of the so-called Smoking Gun Tape.)

Observation: Such factors inﬂuence how we describe what Nixon said,
asserted, claimed, ordered, etc. Our hypothesis, formulated as SPAP

below,

is that there’s no single way to put all of this together in order to come up
with a unique description of what Nixon said. There are many different
ways to do it, no one of which is more correct than all others.

So, what did Nixon say? Well, the current standard reports on this tape

go something like this (found in many history books, innumerable con-
temporaneous newspaper articles and investigative reports, etc.):

Nixon told Haldeman to tell the CIA to tell the FBI not to pursue
their investigation into the Watergate burglary.

Nixon is clearly heard telling his Chief of Staff, Bob Haldeman, to
implement John Dean’s idea that the CIA be used to pressure the FBI
to limit the Watergate investigation.

Semantic Minimalism and Speech Act Pluralism

192

It’s not clear whether ‘those people’ refers to (CIA Director) Richard Helms, (Deputy CIA

Director) Vernon Walters, (FBI Director) Pat Gray, or to all of them. Reports actually vary, and
if you read the transcript carefully, no unique answer emerges and there’s no reason to think
there would be one even if you were able to go back in time and look into Nixon’s head.

Nixon wanted the CIA Director, Richard Helms, to thwart the FBI’s
probe of the Watergate burglary by saying it was a CIA operation.

Nixon told Haldeman to tell Helms that Nixon wanted him to stop
the Watergate investigation.

Nixon told Haldeman to break the law.

These reports all attribute different sayings to the utterance; and they con-
stitute but just a modest start. Nixon’s utterance clearly said lots of other
things, e.g.:

He told Haldeman to tell someone at the CIA to tell the FBI that there
was a connection between the Bay of Pigs invasion and the Watergate
burglary.

He said that Haldeman should give the FBI few details about the con-
nection between the Bay of Pigs and the Watergate burglary.

And so on and so on.

What’s crucial to notice here (and in general) is that our intuitions about

what speakers say with their utterances are inﬂuenced by, at least, the fol-
lowing sorts of considerations:

(1) Facts about the speaker’s intentions and beliefs. These reports make

assumptions about what Nixon believes, for example, that he
thinks ‘those people’ hold certain positions and that they have
certain kinds of power; he has certain beliefs about the CIA and
the FBI, the legal system, etc.

(2) Facts about the conversational context of this particular utterance. The

reports of what Nixon said are inﬂuenced by information we have
about whom Nixon and Haldeman have been talking about, the
topic of their conversation, etc.

(3) Other facts about the world. What’s illegal (i.e., that it would be a

crime for the President of the United States to ask the CIA to ask
the FBI to stop an investigation), that getting the CIA to talk to
the FBI in certain ways constitutes undue inﬂuence, etc.

(4) Logical relations. The most obvious examples are conjuncts of con-

junctions or trivial logical implications. For example, if Nixon
said that he wanted the CIA Director, Richard Helms, to thwart
the FBI’s probe of the Watergate burglary by saying it was a CIA
operation, then it follows he also said that he wanted the CIA
Director, Richard Helms, to thwart the FBI’s probe of the Water-
gate burglary – where the latter follows logically–semantically
from the former.

Speech Act Pluralism

193

(5) In light of (1)–(4), we can easily substitute coextensive predicates

and referring expressions. Take, for example, Haldeman. Since he
was Nixon’s Chief of Staff, one true report would be: ‘Nixon told
his Chief of Staff to break the law.’

(6) There’s no reason to think (1)–(5) exhaust all the factors that inﬂuence

our intuitions about what speakers say.

The general point illustrated by (1)–(6) is that our intuitions about what
speakers say depend on a wide range of considerations not all of which are
encoded solely in the meanings of the words uttered. It is only when these
considerations are combined with the meanings of the words used that it
makes sense for us to ask what an individual said with his utterance.

Thought experiment

To see the effect of a wide range of background beliefs on our intuitions
about what Nixon said, imagine encountering his utterance with these
peculiar beliefs:

That Nixon is a French construction worker, an illegal immigrant
working as a carpenter on a housing project in Washington DC (but
knowing enough about Nixon, for example, being able to recognize
him etc., to refer to him); that Haldeman is a clerk in a video store;
and that Gray and Helms run a laundromat together; that Nixon is
obsessed with US foreign politics and that he’s trying to inﬂuence
what the guy in the video store says to the guys who run the
laundromat; that in his own confused way, he’s trying to defend the
President.

There’s nothing in the words of Nixon’s utterance (their meaning) that pre-
vents these beliefs from being true. Now imagine hearing Nixon’s Smoking
Gun utterance with these beliefs as your background.

When you get in Gray and Helms [the guys the audience believes run
the laundromat], when you get these people in, say: ‘Look, the
problem is that this will open the whole, the whole Bay of Pigs thing,
and the president just feels that,’ ah, without going into the details
. . . don’t, don’t lie to them to the extent to say there is no involve-
ment, but just say this is sort of a comedy of errors, bizarre, without
getting into it, ‘the president believes that it is going to open the
whole Bay of Pigs thing up again, and ah because these people are
plugging for, for keeps and that they should call the FBI in and say
that we wish for the country, don’t go any further into this case.’
Period. That’s the way to put it, do it straight.

Semantic Minimalism and Speech Act Pluralism

194

If you plug these beliefs into the equation, aren’t you going to miss a lot of
what was said? You’re clearly not going to have the intuition that

Nixon told Haldeman to tell the CIA to tell the FBI not to pursue
their investigation into the Watergate burglary.

Nixon is clearly heard telling his Chief of Staff, Bob Haldeman, to
implement John Dean’s idea that the CIA be used to pressure the FBI
to limit the Watergate investigation.

Wouldn’t you miss a lot of what was said?

Additional illustrations

The above example is typical. Take any speech act, and you can use it to
illustrate the same point. We love these cases, but we worry that if we do
too much of it we’ll lose you, but the temptation is too great for us not to
provide just a few additional illustrations.

‘Liddy an Asshole?’ This next case also is from the Smoking Gun transcript
of the conversation between Nixon and Haldeman. Nixon is trying to ﬁnd
out who arranged the payment to those who broke into the hotel room in
the Watergate Hotel. The conversation goes as follows:

PRESIDENT NIXON

: He [Hunt] didn’t know how it was going to be

handled though, with Dahlberg and the Texans and so forth? Well
who is the asshole that did? Is it Liddy? Is that the fellow? He must
be a little nuts.

HALDEMAN

: He is.

PRESIDENT NIXON

: I mean, he isn’t well screwed on is he? Isn’t that

the problem?

Here are some questions for you to think about. Does Nixon say:

䊉

Liddy is clinically insane?

䊉

Liddy is an asshole?

䊉

Liddy has a low IQ?

䊉

Liddy handled the situation in an irresponsible way?

䊉

Liddy mishandled the situation because he’s stupid?

䊉

He (Nixon) has no conﬁdence in Liddy?

䊉

Does he blame Liddy?

䊉

Does he ask Haldeman whether Liddy is an asshole?

Speech Act Pluralism

195

‘O.J. Dressing.’ Here is the kind of crucial information we were constantly
being assailed with during O.J.’s infamous trial. O.J. uttered:

At 11:05 p.m. I put on a white shirt, a blue Yohji Yamamoto suit, dark
socks, and my brown Bruno Magli shoes.

It is easy to imagine these as reports of what O.J. said:

He said that he dressed up in some really fancy clothes late in the
evening.

He said that he changed his clothes right after 11 p.m. (said in a
context where it is shared knowledge that he was wearing a different
set of clothes before 11 p.m.).

He said that he stopped exposing himself to the neighbors right after
11 p.m. (said in a context where it is common knowledge that he was
standing naked in front of the window before 11 p.m.).

He said that he gave the sign at 11:05 (said in a context where it is
common knowledge that putting on the brown Bruno Magli shoes is
a sign of some signiﬁcance or other).

‘The moronic clown.’ Suppose A is a philosopher we tend to describe as a
moronic clown even though we both know A is neither really a clown nor
really moronic. It’s just how we tend affectionately to describe A. Suppose
Cappelen hears B utter ‘A just wrote a book.’ Cappelen could naturally
report B’s utterance to Lepore with:

B said that the moronic clown just wrote a book.

What’s distinctive about this example is that the subject term in the indi-
rect report not only does not refer to A, it doesn’t even describe A, and the
speaker and the audience know this is so.

Two objections

Here are a couple of objections we have encountered to the Central Obser-
vation.

Did Nixon really say what we attributed to him? Did he, strictly

speaking, say it? Did he literally say it? Let’s see: We know this particular utter-

Semantic Minimalism and Speech Act Pluralism

196

For further discussion and many more examples, see Cappelen and Lepore (1997; 1998).

See also Richard (1998) and Reimer (1998). Both raise objections to our view. We respond

in Cappelen and Lepore (1998).

ance was the cornerstone of the impeachment case against Nixon. It
was, in effect, one of the central causes of Nixon’s resignation. Imagine
the absurdity of a defense of Nixon that he didn’t, strictly speaking, ask
the CIA to block the FBI investigation. Or that he didn’t really ask it. Or
that he didn’t literally order it. Any such defense would have been just plain
silly.

As we mentioned earlier, we are terribly naive in that we take the prac-

tice of talking about, and thinking about, what people say at face value. We
think the best way to ﬁnd out about what speakers say is to listen to them,
and then think about what they have said, and ask other people what they
have said. Recanati objects to our modest procedure, and suggests an
alternative:

. . . I strongly disagree with Cappelen and Lepore’s surprising
statement:

We ourselves don’t see how to elicit intuitions about what-is-said
by an utterance of a sentence without appealing to intuitions
about the accuracy of indirect reports of the form ‘He said that
. . .’ or ‘What he said is that . . .’ or even ‘What was said is that . . .’
(Cappelen and Lepore 1997, p. 280)

I ﬁnd this statement surprising, because there obviously is another
way of eliciting truth-conditional intuitions. One has simply to
provide subjects with scenarios describing situations, or, even better,
with – possibly animated – pictures of situations, and to ask them to
evaluate the target utterance as true or false with respect to the situa-
tions in question . . . . this procedure presupposes that normal inter-
preters have intuitions concerning the truth-conditional content of
utterances. On my view, those intuitions correspond to a certain ‘level’
in the comprehension process – a level that a proper theory of lan-
guage understanding must capture. That is the level of ‘what is said’
(as opposed to, e.g., what is implied). (Recanati 2004, pp. 92–3)

Let’s see. Recanati’s idea is that there’s a level of content, a level he calls
‘what is said,’ that this level corresponds to a level ‘in the comprehension’
process, and that’s revealed, not by intuitions about what a speaker said,
not by asking people what they think was said, but rather by looking at ani-
mated pictures. (We’re not sure exactly what ‘animated’ comes to here, but
we’ll let that pass.)

Try to apply Recanati’s proposal to Nixon’s Smoking Gun utterance. The

suggestion, if we understand it correctly, is that we should look at a little
picture, like this:

Speech Act Pluralism

197

Or, even better, we should watch a movie, maybe Oliver Stone’s Nixon, with
Anthony Hopkins playing the role of Nixon, in which there is a scene very
close to this one.

We have three brief comments on/objections to Recanati’s alternative

methodology:

1. Our main objection, and it has nothing in particular to do with the

speciﬁc features of the Smoking Gun utterance, is this: We’re not sure what
we’re supposed to compare to the picture or animation because we are not
sure what Recanati is referring to by ‘utterance.’ If the utterance is just the
sounds made, then we don’t know how to evaluate it with respect to a
picture. If by ‘utterance’ he doesn’t have in mind the sounds uttered, then
he must be referring to a proposition expressed. If this is what he has in
mind, then we need to choose which expressed proposition, i.e., we have to
decide what is said by the utterance before we can do the comparison.
Hence, the comparison can’t be what gets us what is said. No matter what
you do you must rely on assumptions about what was said, asserted, etc. in
order to evaluate ‘the utterance’ with respect to the animated picture.

2. We’re not even sure how to understand Recanati’s animated pictures.

We take it to be plain obvious that pictures don’t have intrinsic represen-

Semantic Minimalism and Speech Act Pluralism

198

Haldeman

Nixon

Don't investigate

Watergate!

tative properties. There’s nothing in a particular drawing that requires that
drawing to be a representation of Nixon. So, the comparison presupposes
extensive interpretation of the picture by the audience. In effect, the audi-
ence has to decide the content of the picture before a comparison. Again,
this shows that the idea that a naive, nontheoretic intuition can be elicited
in this way is deeply confused.

3. Here’s a point that has more speciﬁcally to do with the Smoking Gun

utterance: The utterance is a piece of advice. Nixon is telling Haldeman
what he should do; he’s not describing what Haldeman did or will do. But
how do you ‘animate’ an order, or a suggestion, or a recommendation?
More generally, how do you animate modality?

There are other interesting issues we could pursue here, but limitations

of space and time force us to push forward.

Observations about Speech Act Content

The Central Observation and the sort of data we have been discussing lead
us to the ﬁrst thesis of Radical Speech Act Pluralism, which we call SPAP

SPAP

: Pluralism

No one thing is said (or asserted, or claimed, or . . . ) by any utterance:
rather, indeﬁnitely many propositions are said, asserted, claimed, stated,
etc.

Corollary. When you report what is said, many things are said by your
report, i.e., indirect reports are, of course, no exception to SPAP

. In other

words, Pluralism about speech act content applies just as much to sentences
that report something about what speakers’ utterances say as it does to the
reported speakers’ utterances. As a result: Not only is there no one correct
answer to what was said by an utterance, there’s no one correct answer to
what was said by a report of what was said by an utterance either. There’s
no meta-language in which the speech act content is ﬁxed and determinate.
Pluralism applies all the way through.

Comparison to related views we don’t endorse. It is important to distinguish
Pluralism from two other views, neither of which we mean to endorse by
virtue of endorsing Pluralism (nor do we mean to be rejecting either of these
views as well; we don’t have to take any stand whatsoever on these views
here).

Speech Act Pluralism

199

1. Pluralism

π semantic context sensitivity of ‘said that.’ Pluralism is not the

view that the English locution ‘said that’ is semantically context sensitive.
That would involve the claim that ‘said that’ passes our three tests for
context sensitivity (cf., Chapter 7). That’s not the claim we are making here
(though we’re not rejecting it either

). We’re simply making the claim that

any one utterance asserts many distinct propositions.

2. Pluralism

π relativization to context. Nor are we rejecting nor endorsing

the view that at least one of the propositions said (or asserted, or claimed,
or . . . ) by an utterance of ‘A said that p’ includes a reference to the context
of utterance of that indirect report, i.e., it is consistent with Pluralism that
there is some sort of relativization to the context of utterance of the indi-
rect report at the level of what is said by a report of what is said. In that
sense, what a speaker said might depend on the context of utterance of the
report of what that speaker said.

Here’s what we have in mind: Return to our old friend A, whom, recall,

we regrettably tend to call a moronic clown; we both know A isn’t really a
clown and we’re exaggerating wildly when we describe him as a moron.
Nonetheless, if we hear B utter ‘A just wrote a book,’ Cappelen might report
that utterance to Lepore by saying ‘B said that the moronic clown just wrote
a book.’ It is compatible with SPAP that one of the things Cappelen said by
reporting what B said about A is that B said that the guy we (in such and such
contexts) tend to describe as a moronic clown wrote a book.

SPAP

: What’s said and what’s semantically expressed

One of the many propositions asserted by an utterance is the semantic
content of that utterance (the proposition semantically expressed). We
argued for this in Chapters 7–10.

SPAP

: The nontheory theory of speech act content

We have been unable to ﬁgure out how to devise an algorithm that takes
the proposition semantically expressed and delivers all the propositions
said, asserted, etc. There might not be any systematic theory from which
one can derive all of which is said by an utterance. In this respect, what was
said by u might be similar to what u can be used for, or what’s interesting
about u, or what u is similar to, or what is strange about u. In all of these

Semantic Minimalism and Speech Act Pluralism

200

Remember, we’re not really doing semantics in this book, so we’ll not make any substan-

tial claims about the semantics for speciﬁc expressions. We intend to remain neutral here.

cases, there is no one unique answer. There are many things u is similar to;
there are many things that are interesting about u; and so on.

We don’t have a proof that there is no such theory, so we’re open-minded

about the prospect that someone might devise one, but we strongly suspect
that no such theory is ever forthcoming.

SPAP

: Contextual Two-Dimensionalism (against Original

Utterance Centrism)

Our theory is opposed to theories that are speaker (or audience) centrist,
and more generally, to Original Utterance Centrism. These are theories
according to which the speech act content is ﬁxed by facts about the
speaker, his audience, and their common context. On our view, that’s deﬁ-
nitely not the case.

We think facts not known or available to the speaker (or his audience)

can make a difference. Assumptions made by people who have nothing at
all to do with the original context of utterance can ﬁx what the speaker said.
To take a simple illustration, suppose you uttered (1) several weeks ago:

(1) The table is covered with books.

Suppose that whatever table is under discussion currently sits comfortably
in your father’s ofﬁce (though it did not sit there when you uttered (1)).
Haven’t you said with your utterance of (1) that the table in your father’s
ofﬁce is covered with books? Note that no other account we are aware of
can factor being in your father’s ofﬁce into what was said by your utterance
of (1), since you yourself were ignorant or misinformed about what would
or wouldn’t be in your father’s ofﬁce at the time of your utterance – maybe
the table got moved there right after you spoke.

Suppose Frank utters (1) in a context where only one table is present.

Later, after another table is brought in, a question arises about what Frank
said with his earlier utterance of (1). Suppose that on the sole table present
when Frank spoke now sits a vase, and on the added table sits nothing.
Anyone who reports Frank’s utterance of (1) in this context as having said
that the table is covered with books might grossly misrepresent what he
said, but anyone who reports him as having said that the table with a vase
on it is covered with books has gotten him just right. If you disagree, how
would you in the context described usefully and correctly answer the ques-
tion ‘What did Frank say?’ Should you conclude you cannot correctly report
his utterance?

For additional examples, consider the moronic clown utterance again.

The speaker doesn’t believe that A is a moronic clown; indeed, he isn’t. Even

Speech Act Pluralism

201

if you go back to the Smoking Gun utterance, it’s clear that Nixon asked
Haldeman to put undue pressure on the CIA even if Nixon didn’t think he was.

SPAP

: Speakers don’t have privileged access to the content

of their speech acts

It is a corollary of SPAP

that speakers don’t have privileged access to all

the propositions they assert or say in uttering a sentence. In the above
example, the speaker didn’t know that she said that the table with a vase
on it is covered with books. Such examples need not involve coextensive
referring expressions. If someone points at our friend the moronic clown
(who is neither a clown nor moronic) and utters ‘That man is shady,’ she
has said that that moronic clown is shady. If Justine bought the picture, and if
Justine is French, and Jack utters the sentence ‘Justine bought the picture,’
there are contexts in which one would say something true by saying that Jack
said that a French woman bought the picture, even though Jack doesn’t know
that Justine is French.

The semantic content of an uttered sentence provides an additional, but

more complex, illustration of the same point. In Chapter 11 we argued that
all a semanticist has to say about the semantic content of an utterance of
‘A is red’ is that it expresses the proposition that A is red, and is true just in
case A is red. We semanticists are not in the business of telling you what it
is to be red. It is possible that at the end of the day, color-metaphysicians
will conclude that being red is to be red under conditions C, and this might come
as a bit of a surprise, both to the semanticist and to ordinary speakers. It
might, for example, turn out that it is much easier to satisfy this condition
than one would pretheoretically expect (and harder not to satisfy it). A
speaker who is not a color-metaphysician might utter the sentence ‘A is red’
primarily intending to communicate that A is red on the outside when
washed. In so doing she need not be aware of the fact that she also expressed
a content that’s metaphysically equivalent to the claim that A is red under
conditions C (of course, she does know that she expressed the proposition
that A is red; what she doesn’t know is that this is metaphysically equivalent
to the proposition that A is red under conditions C).

SPAP

: Speakers need not believe everything they sincerely say

A corollary of SPAP

and SPAP

is that speakers need not believe everything

they say even when their saying it is sincere. At least there’s no reason for
saying the person who uttered ‘Justine bought the picture,’ and thereby said
that a French woman bought the picture, believes that a French woman

Semantic Minimalism and Speech Act Pluralism

202

bought the picture. There’s no reason to think that the person who uttered
‘That man is shady,’ and thereby said that the moronic clown is shady,
believes that the moronic clown is shady. There’s no reason to think that
someone who utters ‘The table is covered in books,’ and thereby said that
the table in your father’s ofﬁce is covered in books, believes that the table
in your father’s ofﬁce is covered in books. We don’t have to think that when
Nixon uttered the sentences heard on the Smoking Gun tape, and thereby
asked Haldeman to break the law, he believed he was asking Haldeman to
break the law (he might have, we don’t know, but he might not have).

SPAP

: Expressed belief need not be equivalent to semantic content

A further corollary is that the principle Hawthorne calls the True Belief
Principle is false (Hawthorne 2003, p. 99):

The True Belief Principle (TBP). If a speaker sincerely accepts an utter-
ance u and u has semantic value p, then the belief manifested by his
sincerely accepting u is true iff p is true.

TBP is mistaken for at least two reasons.

First, as formulated, it presupposes that there’s a single belief manifested

by a speaker’s sincere acceptance of an utterance u. That’s not so. Just as an
utterance expresses a wide range of distinct contents, the speaker’s sincere
acceptance of u might manifest a range of distinct beliefs.

Second, it is not the case that all of the propositions a speaker might

manifest a belief in when sincerely accepting an utterance u are true just in
case the semantic content of u is true. SPAP provides an indeﬁnite range of
counterexamples here, but to make it vivid we’ll stick with two.

Imagine a context in which several children are arranged by their nation-

ality. Suppose this is some kind of competition in which nationalities are
of utmost signiﬁcance. All the participants are wrapped in their national
ﬂags. They are there as representatives of their nation, not as individuals.
Suppose, in this context, Cappelen points at a child wrapped in a French
ﬂag and utters, ‘He’s suspicious.’ In so doing he has exhibited a belief in the
proposition that a French child is suspicious and the proposition that the French
participant is suspicious. That proposition is not equivalent to the semantic
content of the uttered sentence.

Or, consider an utterance of ‘The moronic clown is shady.’ The speaker

in uttering this sentence might manifest a belief in the proposition that Pro-
fessor Smyth is shady, but that proposition is not true just in case the seman-
tic content of u is true.

More generally: We have no idea why anyone should accept TBP. If it is

some kind of a priori philosophical principle that’s supposed to be self-

Speech Act Pluralism

203

evident, then all we can say is that it’s not self-evident for us. Indeed, we
think it’s pretty obviously false if considered as an empirical observation
about the relationship between utterances, beliefs, and semantic content.

SPAP

: Against Grice: For those engaged in ‘semantic vs. pragmatics’ talk:

sayings and implicatures are both on the pragmatic side of the divide

Finally, we want to reemphasize a point made in earlier chapters. The very
same contextual features that determine the implicatures of an utterance
inﬂuence what speakers say and assert by that utterance. There is no fun-
damental theoretical divide between sayings and implicatures. They are
both on the side of speech act content. Whatever mechanisms might gen-
erate implicatures are also all used to generate what speakers say.

Pluralism: Summary

SPAP

–SPAP

and their corollaries are some of the most salient implications

of the Central Observation with which we began this chapter. They are by
no means the only relevant such observations, and we think there’s rich
material here for understanding subtle distinctions between various kinds
of speech acts and about the relationship between reports of what people
say and the context of those reports. We’ll not engage in a detailed investi-
gation here, since the overall goals of this book are simply to defend
Semantic Minimalism and to show how, in broad outlines, it ﬁts hand in
glove with Speech Act Pluralism.

Objection 1: How is communication nonmiraculous?

At several crucial points in this book we have argued that contextualism
makes it impossible to understand the varieties of ways in which speakers
share content across diverse contexts. Test 1 in Chapter 7 appeals to facts
about how speakers report on what others say by using disquotational indi-
rect reports. Chapter 8 accuses contextualists of not being in a position to
account for how speakers could communicate across diverse contexts. The
second and third inconsistency charges in Chapter 9 rely on assumptions
about what Radical Contextualists have to say about what speakers say in
various thought experiments. In Chapter 10 we claim that two central
virtues of Semantic Minimalism are related to its ability to avoid what we
called ‘contextual isolationism.’

Semantic Minimalism and Speech Act Pluralism

204

In the light of this last chapter, the following might be a natural ques-

tion to us: Isn’t communication across contexts miraculous on our view
too? On our account, a bunch of propositions get expressed, some of which
the speaker doesn’t even know about. Communication concerns that which
is said, that which the speaker understands. Doesn’t this view undermine
the possibility of communication just as much as Radical Contextualism?

Our response is fourfold:

1. How does our theory account for the ease with which we disquota-

tionally report each other across contexts; i.e., how do we explain that any
utterance of, for example,

(1) Peter’s duck is brown

can be reported disquotationally, i.e., by

(2) The speaker said that Peter’s duck is brown,

even when the report takes place in a context radically different from the
context of the reported utterance of (1)? On our view, this is possible
because every utterance of ‘Peter’s duck is brown’ semantically expresses the
proposition that Peter’s duck is brown. This minimal semantic content is one
of the many propositions said by any utterance of (1). Hence, an utterance
of (2) will succeed as an indirect report, even in those cases where the
context of utterance for (2) is radically different from the context of the
reported utterance of (1). Of course, in most cases, this minimal content is
not the only, or even the most important, proposition the speaker intended
to communicate to her audience. She typically would intend to communi-
cate something more speciﬁc about the relationship between Peter and the
duck (ownership, possession, proximity, or what have you), something
more speciﬁc about the brownness (located on the feathers, the beak, or
what have you), and no doubt much more. But despite all this, the speaker
can’t help but say that Peter’s duck is brown by uttering (1). That’s why we can
indirectly report each other disquotationally across contexts.

2. How is it that we can understand what was said by an utterance of

(1) when that utterance took place in a context radically different from ours
(and we know little or nothing about that context)? The answer should be
obvious by now: We can always understand part of what the speaker said,
namely, that Peter’s duck is brown. That part is easy to understand and it
doesn’t require knowing much about the original context of utterance. You
might even have mistaken beliefs about that context and still understand
that the speaker said that Peter’s duck is brown.

Speech Act Pluralism

205

3. This third aspect of our account of inter-contextual communication

is a bit tricky to understand (in part, because it is based on such an unusual
view, i.e., SPAP

). Remember two features of our view:

(i) Any utterance u succeeds in saying indeﬁnitely many propositions

(see SPAP

(ii) Our view is not Original Utterance Centrist (see SPAP

), i.e., we

think what was said by an utterance u of (1) in a context C can be
determined by features of the context in which we describe what
was said by u. So when we say what was said by u, we can draw on
features of the context we are in now, say, 5stC. These features of
5stC can be constitutive of what was said by u.

Given (i) and (ii) it isn’t surprising that we can have easy access to some of
what was said by u. In some cases, we might even have access to parts of
what the speaker said that the speaker herself doesn’t have access to (see
SPAP

and SPAP

5–7

). You might think that is a strange kind of communi-

cation, but not so if to communicate is to understand each other and if
understanding someone involves understanding what she said and if we are
right about SPAP

, i.e., if we are right in our rejection of Original Utterance

Centrism.

4. An adequate account of inter-contextual communication should not

only explain how it is possible for speakers in diverse contexts to under-
stand each other. It should also explain why it sometimes seems so mind-
bogglingly hard to understand what other people have said. Someone with
a strong sense of failing to understand what someone else said would not,
typically, be particularly relieved by being told that she can fall back on an
inter-contextual disquotational indirect report. A theory that makes com-
municative interaction seem ridiculously easy is no better than one that
makes communication impossible.

We don’t think successful communication is easy. On our view, any

utterance succeeds in expressing an indeﬁnite number of propositions. One
of these, the proposition semantically expressed, is easy to grasp. Others are
extremely hard to access and there is no reason to think that any one person
can ever grasp all that was said by an utterance, not even the speaker. This
is how Semantic Minimalism combined with Speech Act Pluralism can
account for both the sense in which communication is easy and the sense
in which it is impossibly difﬁcult.

Semantic Minimalism and Speech Act Pluralism

206

In ‘Shared Content’ (2004 MS), we pursue the connection between these ideas and various

nonlinguistic practices.

Objection 2: How to talk about and test for semantic content

In this book we have talked repeatedly about the semantic content of utter-
ances (the proposition semantically expressed). We have presented tests for
how to ﬁnd the semantic content of utterances. We have done that by utter-
ing sentences, by making claims, and assertions. But in so doing we assert
an indeﬁnite number of propositions, only one of which is the proposition
semantically expressed. Doesn’t that undermine our tests? Doesn’t it make
the very project of this book impossible?

Again, the answer should be fairly obvious. It takes stage setting to get

readers to ‘see’ or ‘notice’ the semantic content of an utterance. In most
regular-life contexts, the semantic content is not what is focused upon.
We’re interested in all the extra stuff; that’s typically where the juice is.
However, in certain peculiar contexts, such as articles on semantics, the goal
is to draw the audience’s attention to the semantic content of an utterance,
and the audience recognizes that this is the goal. With a little stage setting
from us, and some receptiveness from the audience, we almost always
succeed.

When you read our tests in Chapter 7, for example, you obviously have

to focus on what’s communicated; we are after all communicating with you,
the reader, so the tests are tests that go via the communicated content. But
that’s not a problem if we ensure that the context is such that it is the
semantic content that’s being focused on. Think of the tests like this: They
are ways to get the audience to notice the semantic features of (utterances
of) sentences. They create contexts in which our attention is drawn to fea-
tures of the semantic content.

Can we be sure that we will succeed? Can we be sure we have succeeded

in constructing the right kind of context for this? No. But we are fairly
conﬁdent.

Conclusion: The Diagram of Communication

We noticed that other authors who write books on the topics of this book
tend to employ many interesting and complicated diagrams. They have
elaborate pictures of how various levels of content are interrelated, what
kinds of content can be expressed by an utterance, and so on and so on.
We’ve always been impressed by (and to be honest a little bit jealous of)
these diagrams, and so it seems appropriate to end our book with what we
call the Diagram of Communication.

Speech Act Pluralism

207

Etc. . . .

Indeﬁnitely many contexts of interpretation

for utterances of interpretation of U

Indeﬁnitely many contexts of interpretation

for U

Context of Utterance

Utterance of S

Fixes (no theory, not systematic?)

Speech Act Content: Indeﬁnitely many

propositions

Proposition Semantically Expressed

(minimal context sensitivity)

Diagram of Communication

208

209

References

Austin, J. L. (1961), How to Do Things with Words (Oxford: Clarendon Press).
—— (1962), Sense and Sensibilia, ed. Geoffrey J. Warnock (Oxford: Clarendon Press).
Bach, K. (1999), ‘The Myth of Conventional Implicature,’ Linguistics and Philosophy,

22/4: 327–66.

—— (1987), Thought and Reference (Oxford: Oxford University Press).
—— (1994a), ‘Conversational Implicature,’ Mind and Language, 9/2: 124–62.
—— (1994b), ‘Semantic Slack: What is Said and More,’ in S. Tsohatzidis (ed.), Foun-

dations of Speech Act Theory (London: Routledge), pp. 267–91.

—— (1997), ‘Do Belief Reports Report Beliefs?’ Paciﬁc Philosophical Quarterly, 78:

215–41.

—— (1999), ‘The Semantics Pragmatics Distinction: What it Is and Why it Matters,’

in K. Turner (ed.), The Semantics–Pragmatics Interface from Different Points of View
(Oxford: Elsevier), pp. 65–84.

—— (2000), ‘Quantiﬁcation, Qualiﬁcation, and Context: A Reply to Stanley and

Szabó,’ Mind and Language, 15: 262–83.

—— (2001), ‘You Don’t Say?’, Synthese, 128: 15–44.
—— (2002a), ‘Seemingly Semantic Intuitions,’ in J. Campbell, M. O’Rourke, and D.

Shier (eds.), Meaning and Truth: Investigations in Philosophical Semantics (New York:
Seven Bridges Press), pp. 21–33.

—— (2002b), ‘Semantic, Pragmatics,’ in J. Campbell, M. O’Rourke, and D. Shier

(eds.), Meaning and Truth: Investigations in Philosophical Semantics (New York: Seven
Bridges Press), pp. 284–92.

Bezuidenhout, A. (1996), ‘Pragmatics and Singular Reference,’ Mind and Language,

11: 133–59.

—— (1997), ‘Pragmatics, Semantic Underdetermination and the Referential/Attribu-

tive Distinction,’ Mind, 106/423: 375–409.

—— (2002), ‘Truth-Conditional Pragmatics,’ Philosophical Perspectives, 16: 105–34.
Blair, Don (2004), ‘Domain Restrictions, Relational Nouns, and Weak Crossover,’

MS.

Burge, T. (1986), ‘On Davidson’s “Saying That,” ’ in E. Lepore (ed.), Truth and Inter-

pretation (Oxford: Blackwell), pp. 190–208.

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

Cappelen, H., and E. Lepore (1997), ‘On an Alleged Connection between Indirect

Quotation and Semantic Theory,’ Mind and Language, 12: 278–96.

—— —— (1998), ‘Reply to Richard and Reimer,’ Mind and Language, 13: 117–621.
—— —— (2002a), ‘Indexicality, Binding, Anaphora and A Priori Truth,’ Analysis (Oct.),

271–81.

—— —— (2002b), ‘Insensitive Quantiﬁers,’ in J. Campbell, M. O’Rourke, and D. Shier

(eds.), Meaning and Truth: Investigating Philosophical Semantics (New York: Seven
Bridges Press), pp. 197–213.

—— (2004), ‘Shared Content,’ MS.
Carston, R. (1988), ‘Implicature, Explicature, and Truth-Theoretic Semantics,’ in R.

Kempson (ed.), Mental Representations: The Interface between Language and Reality
(Cambridge: Cambridge University Press), pp. 155–81.

—— (1993), ‘Conjunction, Explanation, and Relevance,’ Lingua, 90: 27–48.
—— (2002), Thoughts and Utterances: The Pragmatics of Explicit Communication (Oxford:

Blackwell).

—— (2004), ‘Relevance Theory and the Saying/Implicating Distinction,’ in L. Horn

and G. Ward (eds.), Handbook of Pragmatics (Oxford: Blackwell), pp. 633–56.

Casteneda, Hector-Neri (1980), ‘The Theory of Questions, Epistemic Powers, and

the Indexical Theory of Knowledge,’ Midwest Studies in Philosophy, 5: 193–238.

Clapp, Lenny (2002), ‘What Unarticulated Constituents Could Not Be,’ in J.

Campbell, M. O’Rourke, and D. Shier (eds.), Meaning and Truth: Investigating
Philosophical Semantics (New York: Seven Bridges Press), pp. 231–56.

Cohen, Stewart (1988), ‘How to be a Fallibilist,’ Philosophical Perspectives, 2: 91–123.
—— (1991), ‘Skepticism, Relevance, and Relativity,’ in B. McLaughlin (ed.), Dretske

and His Critics (Cambridge, Mass.: Blackwell), pp. 17–37.

—— (1999), ‘Contextualism, Skepticism, and the Structure of Reasons,’ Philosophical

Perspectives, 13: 57–89.

Condoravdi, C., and J. M. Gawron (1995), ‘The Context Dependency of Implicit

Arguments,’ in M. Kanazawa, C. Pinon, and H. de Swart (eds.), Quantiﬁers, Deduc-
tion and Context (Stanford, Calif.: CSLI), pp. 1–32.

Crimmins, M. (1992), Talk about Belief (Cambridge, Mass.: MIT Press).
—— and J. Perry (1989), ‘The Prince and the Phone Booth: Reporting Puzzling

Beliefs,’ Journal of Philosophy, 86: 685–711.

Davidson, D. (1967), ‘The Logical Form of Action Sentences,’ in Davidson, The Logic

of Decision and Action (Pittsburgh: Pittsburgh University Press); repr. in Davidson,
Essays on Actions and Events, 2nd edn. (Oxford: Oxford University Press, 2001), pp.
105–21.

—— (1968), ‘On Saying That,’ repr. in Davidson, Inquiries into Truth and Interpretation

(Oxford: Oxford University Press), pp. 93–108.

—— (1976), ‘Reply to Foster,’ repr. in Davidson, Inquiries into Truth and Interpretation

(Oxford: Oxford University Press), pp. 171–80.

—— (1984), Inquiries into Truth and Interpretation (Oxford: Oxford University Press).
DeRose, Keith (1992), ‘Contextualism and Knowledge Attributions,’ Philosophy and

Phenomenological Research, 52/4: 913–29.

—— (1995), ‘Solving the Skeptical Problem,’ Philosophical Review, 104: 1–52.
—— (1999), ‘Contextualism: An Explanation and Defense,’ in John Greco and Ernie

Sosa (eds.), The Blackwell Guide to Epistemology (Oxford: Blackwell), pp. 187–205.

References

210

—— (2004), ‘Single Scoreboard Semantics,’ Philosophical Studies, 119: 1–21.
Dreier, James (1990), ‘Internalism and Speaker Relativism,’ Ethics, 101: 6–26.
Evans, G. (1985), ‘Does Tense Logic Rest on a Mistake?’ in Evans, Collected Papers

(Oxford: Clarendon Press), pp. 343–63.

Farkas, Donka (1997), ‘Evaluation Indices and Scope,’ in A. Szabolcsi (ed.), Ways of

Scope Taking (Dordrecht: Kluwer), pp. 183–215.

Fodor, J., and E. Lepore (1992), Holism: A Shopper’s Guide (Oxford: Blackwell).
—— —— (1999), ‘All at Sea in Semantic Space: Churchland on Meaning Similarity,’

Journal of Philosophy, 46/8: 381–403.

Frege, G. (1977), ‘The Thought,’ trans. as ‘Thoughts,’ in Frege, Logical Investigations,

ed. and trans. P. T. Geach (Oxford: Oxford University Press), pp. 1–30.

Gaulker, Christopher (1997), ‘Domain of Discourse,’ Mind, 106: 1–32.
—— (2002), Words without Meaning (Cambridge, Mass.: MIT Press).
Gibbs, Raymond, and Jessica Moise (1997), ‘Pragmatics in Understanding what is

Said,’ Cognition, 62: 51–74.

Graff, Delia (2000), ‘Shifting Sands: An Interest Relative Theory of Vagueness,’ Philo-

sophical Topics, 28/1: 45–81.

Grice, P. (1989), Studies in the Ways of Words (Cambridge, Mass.: Harvard University

Press).

Hawthorne, J. (2003), Knowledge and Lotteries (Oxford: Oxford University Press).
Heal, Jane (1997), ‘Indexical Predicates and Their Uses,’ Mind, 106: 619–40.
Heller, Mark (1999), ‘Contextualism and Anti-Luck Epistemology,’ Philosophical Per-

spectives, 13: 115–29.

Higginbotham, J. (1985), ‘On Semantics,’ Linguistic Inquiry, 16/4: 547–93.
Kamp, Hans (1975), ‘Two Theories of Adjectives,’ in E. Keenan (ed.), Formal Seman-

tics of Natural Language (Cambridge: Cambridge University Press), pp. 23–55.

—— (1981), ‘The Paradox of the Heap,’ in U. Monnich (ed.), Aspects of Philosophical

Logic (Dordrecht: Reidel), pp. 225–77.

Kaplan, David (1989a), ‘Demonstratives,’ in J. Almog, J. Perry, and H. Wettstein

(eds.), Themes from Kaplan (Oxford: Oxford University Press), pp. 481–563.

—— (1989b), ‘Afterthoughts’, in J. Almog, J. Perry, and H. Wettstein (eds.), Themes

from Kaplan (Oxford: Oxford University Press), pp. 565–614.

King, Jeffrey C., and Jason Stanley (forthcoming), ‘Semantics, Pragmatics, and the

Role of Semantic Content,’ in Zoltán Gendler Szabó (ed.), Semantics vs. Pragmatics
(Oxford: Clarendon Press).

Kripke, S. (1977), ‘Speaker Reference and Semantic Reference,’ Midwest Studies in Phi-

losophy, 2: 6–27.

—— (1980), Naming and Necessity (Cambridge, Mass.: Harvard University Press).
Larson, R., and P. Ludlow (1993), ‘Interpreted Logical Forms,’ Synthese, 95: 305–56.
—— and G. Segal (1995), Knowledge and Meaning (Cambridge, Mass.: MIT Press).
Lepore, Ernie (2004), ‘An Abuse of Context in Semantics: The Care of Incomplete

Deﬁnite Descriptions,’ in M. Reimer and A. Bezuidenhout (eds.), Descriptions and
Beyond (Oxford: Oxford University Press), pp. 42–68.

Lewis, D. (1970), ‘General Semantics,’ Synthese, 22: 18–67.
—— (1973), Counterfactuals (Oxford: Blackwell).
—— (1980), ‘Index, Context and Content,’ in S. Kanger and S. Ohman (eds.), Philoso-

phy and Grammar (Dordrecht: Reidel), pp. 79–100.

References

211

Lewis, D. (1996), ‘Elusive Knowledge,’ Australasian Journal of Philosophy, 74: Papers in

Metaphysics and Epistemology, 549–67.

—— (1998), ‘Index, Context and Content,’ in Lewis, Papers in Philosophical Logic

(Cambridge: Cambridge University Press), pp. 21–44.

Ludlow, Peter (1989), ‘Implicit Comparison Classes,’ Linguistics and Philosophy, 12:

519–33.

—— (1995), ‘The Logic of Determiners,’ Journal of Philosophical Logic, 24: 47–69.
McDowell, J. (1987), ‘In Defence of Modesty,’ in B. M. Taylor (ed.), Michael Dummett

(Dordrecht: Martinus Nijhoff), pp. 59–80.

Montague, R. (1974), ‘Pragmatics,’ in Montague, Formal Philosophy (New Haven: Yale

University Press), pp. 95–118.

Moravcsik, J. M. (1990), Thought and Language (London: Routledge).
—— (1998), Meaning, Creativity, and the Partial Inscrutability of the Human Mind

(Stanford, Calif.: CSLI).

Neale, Stephen (1990), Descriptions (Cambridge, Mass.: MIT Press).
—— (1992), ‘Paul Grice and the Philosophy of Language,’ Linguistics and Philosophy,

15: 509–59.

—— (2000), ‘On Being Explicit: Comments on Stanley and Szabó,’ Mind and Language,

15: 284–94.

Nelson, Michael (2001), MS.
Nunberg, G. (1992), ‘Two Kinds of Indexicality,’ in C. Barker and D. Dowty (eds.),

Proceedings of the Second Conference on Semantics and Linguistic Theory (Columbus:
Ohio State University), pp. 283–301.

—— (1993), ‘Indexicality and Deixis,’ Linguistics and Philosophy, 16: 1–43.
Pagin, Peter (forthcoming), ‘Compositionality and Context,’ in G. Preyer (ed.), Con-

textualism (Oxford: Oxford University Press); page references are to the MS.

Parsons, T. (1972), ‘Some Problems Concerning the Logic of Grammatical Modi-

ﬁers,’ in G. Harman and D. Davidson (eds.), Semantics of Natural Language
(Dordrecht: Reidel), pp. 127–41.

—— (1990), Events and the Semantics of English (Cambridge, Mass.: MIT Press).
Partee, Barbara (1973), ‘Some Structural Analogies between Tenses and Pronouns

in English,’ Journal of Philosophy, 70/18: 601–9.

—— (1984), ‘Nominal and Temporal Anaphora,’ Linguistics and Philosophy, 7: 243–86.
—— (1989), ‘Binding Implicit Variables in Quantiﬁed Contexts,’ in C. Wiltshire, B.

Music, and R. Graczyk (eds.), Papers from CLS 25 (Chicago: Chicago Linguistics
Society), pp. 342–65.

Perry, John (1986), ‘Thought without Representation,’ Proceedings of the Aristotelian

Society, suppl. vol. 60: 263–83; repr. in Perry, The Problem of the Essential Indexical
and Other Essays (Oxford: Oxford University Press, 1993), pp. 205–25.

—— (1993), The Problem of the Essential Indexical and Other Essays (New York: Oxford

University Press).

—— (1998), ‘Indexicals, Contexts, and Unarticulated Constituents,’ in D. Westerståhl

et al. (eds.), Computing Natural Language (Stanford, Calif.: CSLI), pp. 1–11.

Platts, M. (1979), Ways of Meaning (London: Routledge & Kegan Paul).
Raffman, Diana (1994), ‘Vagueness without Paradox,’ Philosophical Review, 103/1:

41–74.

—— (1996), ‘Vagueness and Context-Sensitivity,’ Philosophical Studies, 81: 175–92.

References

212

Ramsey, F. (1978), Foundations of Mathematics (London: Routledge & Kegan Paul).
Recanati, F. (1989), ‘The Pragmatics of What is Said,’ Mind and Language, 4: 295–

329.

—— (1993), Direct Reference: From Language to Thought (Oxford: Blackwell).
—— (1996), ‘Domains of Discourse,’ Linguistics and Philosophy, 19: 445–75.
—— (2001), ‘What is Said,’ Synthese, 128: 75–91.
—— (2002), ‘Unarticulated Constituents,’ Linguistics and Philosophy, 25: 299–345.
—— (2003), ‘The Limits of Expressibility,’ in B. Smith (ed.), John Searle (Cambridge:

Cambridge University Press), pp. 189–213.

—— (2004), Literal Meaning (Cambridge: Cambridge University Press).
Reimer, M. (1992), ‘Incomplete Descriptions,’ Erkenntnis, 37: 347–63.
—— (1998), ‘What is Meant by “What is Said”? A Reply to Cappelen and Lepore,’

Mind and Language, 13/4: 598–604.

Richard, Mark (1990), Propositional Attitudes: An Essay on Thoughts and How We Ascribe

Them (New York: Cambridge University Press).

—— (1998), ‘Semantic Theory and Indirect Speech,’ Mind and Language, 13/4: 605–16.
—— (2004), ‘Contextualism and Relativism,’ Philosophical Studies, 119/1–2: 215–42.
Rumﬁtt, I. (1993), ‘Content and Context: The Paratactic Theory Revisited and

Revised,’ Mind, 102: 429–53.

Russell, B. (1905), ‘On Denoting,’ Mind, 14: 479–93.
—— (1919), Introduction to Mathematical Philosophy (London: Allen & Unwin).
—— (1940), An Inquiry into Meaning and Truth (London: Allen & Unwin; New York:

W. W. Norton).

—— (1959), ‘Mr. Strawson on Referring,’ in Russell, My Philosophical Development

(London: Allen & Unwin), pp. 238–45.

Rysiew, Patrick (2001), ‘The Context-Sensitivity of Knowledge Attributions,’ Noûs,

35/4: 477–514.

Schiffer, Stephen (1998), ‘Descriptions, Indexicals and Belief Reports: Some Dilem-

mas,’ Mind, 104 (1995), 107–31; repr. in Gary Ostertag (ed.), Deﬁnite Descriptions:
A Reader (Cambridge, Mass.: MIT Press, 1998), pp. 369–95.

—— (1996), ‘Contextualist Solutions to Scepticism,’ Proceedings of the Aristotelian

Society, 96: 317–33.

Schlenker, Philippe (2003), ‘A Plea for Monsters,’ Linguistics and Philosophy, 26:

29–120.

Searle, John (1978), ‘Literal Meaning,’ Erkenntnis, 13: 207–24.
—— (1980), ‘The Background of Meaning,’ in J. Searle, F. Kiefer, and M. Bierwisch

(eds.), Speech Act Theory and Pragmatics (Dordrecht: Reidel), pp. 221–32.

—— (1983), Intentionality (Cambridge: Cambridge University Press).
Sellars, W. (1954), ‘Presupposing,’ Philosophical Review, 63: 197–215.
Soames, Scott (1986), ‘Incomplete Deﬁnite Descriptions,’ Notre Dame Journal of

Formal Logic, 27: 349–75.

—— (1989), ‘Direct Reference, Propositional Attitudes, and Semantic Content,’ Philo-

sophical Topics, 15: 47–87.

—— (1999), Understanding Truth (Oxford: Oxford University Press).
Sperber, D., and D. Wilson (1986), Relevance (Oxford: Blackwell).
Stanley, J. (2000), ‘Context and Logical Form,’ Linguistics and Philosophy, 23: 391–424.
—— (2002a), ‘Making it Articulated,’ Mind and Language, 17: 149–68.

References

213

Stanley, J. (2002b), ‘Nominal Restriction,’ in G. Preyer and G. Peter (eds.), Logical

Form and Language (Oxford: Oxford University Press), pp. 365–88.

—— (2003a), ‘Context, Interest-Relativity, and the Sorites,’ Analysis, 63/4: 269–80.
—— (2003b), ‘Context, Interest-Relativity, and Knowledge,’ MS.
—— and Z. Szabó (2000a), ‘On Quantiﬁer Domain Restriction,’ Mind and Language,

15/2: 219–61.

—— —— (2000b), ‘Reply to Bach and Neale,’ Mind and Language, 15: 261–5.
—— and T. Williamson (1995), ‘Quantiﬁers and Context-Dependence,’ Analysis, 55:

291–5.

Strawson, P. (1950), ‘On Referring,’ Mind, 59: 320–44.
Szabó, Z. (2001), ‘Adjectives in Context,’ in I. Kenesei and R. M. Harnish (eds.),

Perspectives on Semantics, Pragmatics and Discourse (Amsterdam: John Benjamins),
pp. 119–46.

Taylor, Ken (2001), ‘Sex, Breakfast, and Descriptions,’ Synthese, 128: 45–61.
Travis, Charles (1985), ‘On What is Strictly Speaking True,’ Canadian Journal of Phi-

losophy, 15/2: 187–229.

—— (1989), The Uses of Sense (Oxford: Oxford University Press).
—— (1994), ‘On Being Truth-Valued,’ in S. L. Tsohatzidis (ed.), Foundations of Speech

Act Theory (London: Routledge), pp. 167–86.

—— (1996), ‘Meaning’s Role in Truth,’ Mind, 100: 451–66.
—— (1997a), ‘Pragmatics,’ in B. Hale and C. Wright (eds.), A Companion to the Philoso-

phy of Language (Oxford: Blackwell), pp. 97–107.

—— (1997b), ‘Reply to Simmons,’ Mind, 106: 119–20.
Unger, Peter (1975), Ignorance: A Case for Skepticism (New York: Oxford University

Press).

—— (1984), Philosophical Relativity (Minneapolis: University of Minnesota Press).
—— (1995), ‘Contextual Analysis in Ethics,’ Philosophy and Phenomenological Research,

55: 1–26.

Vallée, Richard (2003), ‘Context-Sensitivity Beyond Indexicality,’ Dialogue, 42:

79–106.

Vendler, Z. (1967), Linguistics and Philosophy (Ithaca, NY: Cornell University Press).
Westerståhl, Dag (1985), ‘Determiners and Context Sets,’ in J. van Benthem and

A. ter Meulen (eds.), Quantiﬁers in Natural Language (Dordrecht: Foris), pp. 45–71.

—— (1989), ‘Quantiﬁers in Formal and Natural Languages,’ in D. Gabbay and

F. Guenthner (eds.), Handbook of Philosophical Logic, iv (Dordrecht: Kluwer), pp.
1–131.

Wilson, D., and D. Sperber (1981), ‘On Grice’s Theory of Conversation,’ in P. Werth

(ed.), Conversation and Discourse (London: Croom Helm), pp. 155–78.

References

214

Carnap, Rudolf, 121
Carston, R., 5, 7, 32, 38, 43, 62, 147,

149, 154, 175, 177, 178, 179,
180, 181, 182, 185

Casteneda, Hector-Neri, 28
character, 78, 118, 148
CIA, 192–203
Clapp, Lenny, 22, 24
cohabitating expressions, 12, 77
Cohen, S., 8, 28, 29, 109, 120, 121
collective descriptions, 99–104
collective descriptions test, 99–104, 151
color terms, 44, 68, 92–4, 111, 118–19,

161

communicated content, 22, 113, 122,

149, 207

communication, 88, 91, 107, 112,

123–7, 131, 144, 153, 156, 170,
175, 178, 180, 182, 184, 185,
204–7

comparative adjectives, 9, 21–2, 32, 53,

79, 94, 95, 103, 170, 171, 172

comparative similarity, 25, 26
completion, 34, 35, 62, 65
compositionality, 12, 145
concept construction, 12
conditionals, 25, 26
Condoravdi, C., 1
context shifting, see context shifting

argument

Index

215

a priori truths, 76, 78–9, 151
adicity, 75
adjectives, 1, 79, 95, 102, 103, 144, 172
adverbs, 1, 102, 144
ambiguity, 35, 42, 145
anaphoric relations, 73, 76, 77, 79, 151
Austin, J. L., 5, 6, 30, 32
availability principle, 185–8

see also psychological requirement

Bach, K., 7, 9, 25, 34, 36, 37, 51, 60, 61,

62, 63, 147

basic set (of context sensitive

expressions), 2, 6–8, 12, 14, 17,
39, 40, 42, 59, 63, 89, 90, 93, 94,
100, 106, 108, 113

belief reports, 22–4, 36, 37
Bezuidenhout, A., 31, 35, 43, 44, 68,

92, 123–5, 131, 135–40, 184

bin Laden, Osama, 108, 172
binding arguments, 69–76, 83

see also indexicals, hidden

binding, 70–1, 73
Blair, Dan, 76
Bonaparte, Napoleon, 127, 184
Burge, T., 55
Bush, George W., 47–8, 89

Cappelen, H., 57, 71, 76, 90, 118, 127,

187, 196, 197, 200, 203

Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism

Herman Cappelen, Ernie Lepore

context shifting argument (CSA), 4, 5,

10, 14, 17–33, 35, 37, 38, 39–52,
53–8, 60, 64, 68, 69, 70, 73, 107,
109, 111, 115, 116

instability, 39

local contextualist solutions, 42,

49–52

context shifting intuitions, see context

shifting argument

contextual isolationism, 153, 204
contextual salience, 149
contextual salience absorption, 88,

113–16

contextual two-dimensionalism, 201
contextuals, 1
counterfactual conditionals, 24–6, 30,

Crimmins, M., 11, 22–4, 62

Dahlberg, K., 195
Davidson, D., 54, 55, 75
Dean, John, 192, 195
deﬁnite descriptions, 19
demonstratives, 2, 147–50
DeRose, Keith, 8, 27, 28, 107, 109, 119
direct quotation, 51, 121
disambiguation, 34, 131, 177
disquotational truth conditions, 65,

164

domain of discourse, 37

see also quantiﬁer domain

restrictions; quantiﬁers

Dreier, James, 29, 30

enrichment, 12
epistemic attitude attributions, see

knowledge attributions

epistemic contextualism, 26, 28
Evans, G., 118

Farkas, Donka, 71
FBI, 192–203
Fodor, J., 126
fragmentary proposition, 7
free enrichment, 12
Frege, G., 123, 152–3

Gaulker, Christopher, 20, 38
Gawron, Jean-Mark, 1
GEN, 40–52
genitives, 1, 35, 144

see also possessives

genuinely context sensitive expressions,

2, 143

see also basic set

geometrical terms, 30, 94, 96, 103
Gettier, Edmund L., 27
gradable adjectives, 21
Graff, Delia, 118, 171
Graham, Martha, 161
Gray, Pat, 192, 194
Grice, P., 10, 124, 150, 178–9, 204

minimalism, 176–80
what-was-said, 54, 176–80, 186

Haldeman, R. H., 192–203
Hawthorne, J., 76, 171, 203
Heal, Jane, 51
Heller, Mark, 28
Helms, Richard, 192–4
hidden domain variable, 70–7, 103

see also indexicals, hidden

hidden indexical strategy, 12

see also indexicals, hidden

Higginbotham, J., 21
Hopkins, Anthony, 198
Hunt, E. Howard, 195

implicature, 57, 176–82, 204
impoverished context shifting

argument (ICSA), 107–8, 110,
116, 117, 122

incomplete logical form, 7
incompleteness, see semantic

incompleteness

incompleteness argument, 5, 10, 11, 17,

33–8, 52, 59–68, 69, 70, 73, 156,
166

instability, 59–65

psychological reply, 65
subsyntactic constituents reply,

66–8

incorporation, 77

Index

216

indexicals, 1, 8, 9, 28, 54, 69–83, 115,

118, 121, 131, 145, 147–50, 172

hidden, 12, 50, 69–83

syntactic evidence, 52, 69, 70–2

indirect discourse, see indirect reports
indirect reports, 50–2, 55, 88–99, 118,

138, 152, 196, 197, 199, 200,
204, 205, 206

indirect speech, see indirect reports
inter-contextual content sharing, 184
inter-contextual disquotation (ICD)

test, 104–9, 113–14, 151

inter-contextual disquotational

indirect report test, 90, 94, 104,
151

interpretive truth conditions, 146, 164
introspection, 49, 97

justiﬁcation, 28, 129, 180

Kamp, Hans, 102
Kaplan, David, 1, 2, 13, 54–5, 78–9, 82,

117–18, 120–1, 144, 149

Kennedy, John F., 96
King, Jeffrey C., 154, 175
knowledge ascriptions, see knowledge

attributions

knowledge attributions, 8, 26–9, 32, 94,

95, 102, 107, 110–11, 171

Kripke, S., 23

Larson, R., 55, 75
Lepore, E., 57, 71, 76, 80, 89, 90, 118,

126, 127, 187, 189, 196, 197,
200

Lewis, D., 8, 19, 25, 26, 28, 30, 82, 107,

109, 118

Liddy, Gordon, 195
logical form, 9
Ludlow, Peter, 55, 79, 90

McDowell, J., 55, 56
Mill, John Stuart, 26
minimal content, 152
minimal propositions, 177

see also semantic minimalism

metaphysical nihilism, 161–3
minimal semantic content, 180–6, 205

cognitive role, 184

missing arguments, 36
mistaken assumption (MA), 13, 40,

53–8

modal constructions, 94
modals, 96
moderate contextualism (MC), 2, 5,

7–14, 37, 39–52, 54, 57, 59–68,
83, 88, 96, 101, 102, 104, 105,
107, 123, 151, 155, 156, 189

confused semanticism, 151
hidden indexical strategy, 8, 9
misguided semanticism, 7
semantic opportunism, 7, 151
surprise indexical strategy, 8, 9
unarticulated constituent strategy, 8,

monsters, 88, 116–22
Montague, R., 82
moral contextualism, 29
moral judgments, 29, 32, 94
moral terms, 95
Moravcsik, J. M., 43
Mozart, 37

narrow context, 12, 13, 147–8
Neale, Stephen, 19
Nelson, Michael, 71
neo-Wittgensteinians, 5
Nixon, Richard M., 192–203
nominal restriction, 72
nominals, 94, 95, 102, 103
nonliterality, 42
Nunberg, G., 1

occasion-sensitivity, 135
original utterance centrism, 154, 181,

201, 206

Oswald, Lee H., 96

Paderewski, 36, 37
Pagin, Peter, 30
Parsons, T., 21
Partee, Barbara H., 1

Index

217

Perry, John, 9, 11, 22–4, 30, 34, 62, 159
Pietroski, Paul, 77
Platts, M., 55
polysemy, 42, 47
possessives, 35, 53, 94, 95, 103
pragmatic content, 148
pragmatic interpretation, 148, 177
pragmatically enriched proposition, see

implicature

primary scope, 118, 121
PRO, 77
proposition asserted, 150
proposition semantically expressed, 2,

4, 6, 9, 11, 19, 20, 21, 33, 59, 92,
98, 101, 108, 109, 115, 124,
143–50, 154, 155–75, 177,
181–9, 200, 205, 206, 207

see also semantic content

propositional attitude ascriptions,

22–4, 32, 36, 94, 95, 102

propositional fragment, 6, 11, 61
propositional schema, 7
psychological attitude ascriptions, see

propositional attitude
ascriptions

psychological requirement, 187

see also availability principle

psychologism, 152
pure semantic content, 148

quantiﬁed noun phrases, 73
quantiﬁer domain restriction, 35, 41,

49, 70–83, 102, 103

screening off, 81–3
see also quantiﬁers

quantiﬁers, 8, 12, 18–21, 30, 31, 35, 41,

48, 49, 50, 53, 55, 70–83, 94, 95,
102, 103, 104, 158, 159

Quine, Willard Van Orman, 25, 26

Rachmaninov, 36
radical, see propositional fragment
radical contextualism (RC), 2, 5–14, 31,

39–68, 83, 85, 87–140, 151,
155, 156, 181, 189, 199, 204,
205

internal inconsistency, 115, 128, 133

Raffman, Diana, 102
Ramsey, F., 26
real context shifting argument (RCSA),

104–22, 151

Recanati, F., 5, 7, 18, 32–3, 43, 62, 124,

131, 147–9, 154, 175, 177, 181,
186, 186–9, 197, 198

reference assignment, 131, 177
Reimer, M., 196
relativization to context, 200
relevance theory, 5, 32, 124, 175, 181,

185

reported context, 93
reporting context, 93
Richard, Mark, 21, 22, 192, 193, 196
Russell, B., 150
Russert, Tim, 47

‘said that’ context sensitivity, 200
samesaying, 55
saturation, 12, 177
Schiffer, Stephen, 19
Schlenker, Philippe, 120
Searle, John R., 6, 32, 43, 64, 109, 131
Segal, Gabriel, 75
Sellarsians, 5
semantic content, 2, 3, 4, 13, 40, 53, 54,

57–8, 63, 72, 89, 99, 113, 115,
116, 121, 122, 143–54, 164–5,
175, 177, 181, 182, 200, 202,
203, 204, 207

see also minimal semantic content;

proposition semantically
expressed

semantic incompleteness, 33–8, 62–8

see also incompleteness argument

semantic interpretation, 35, 148, 177
semantic minimalism, 2–14, 57, 59, 87,

92, 97, 98, 114, 115, 136, 141,
143–75, 176, 177, 178, 180, 186,
189, 204, 206

argument by elimination, 150, 151
argument from explanatory force,

150, 151

minimal propositions, 61, 132, 143,

156, 185, 186, 189

communicative role, 156, 176–89

Index

218

metaphysical nihilism, 163, 170,

171, 175

metaphysical objection, 155–75
psychological objection, 175,

176–89

semantic potential, 123, 124, 125
semantic scaffolding, 7
semantic skeleton, 7
semantic solipsism, see contextual

isolationism

semantic template, 7
semantic truth conditions, 134, 164
semantic underdetermination, see

semantic incompleteness

sentence meaning, 131
shared context, 125, 175, 182–5
Simpson, O. J., 196
Smoking Gun Tape, 192–203
Soames, Scott, 102, 119, 120
source-situations, 124
speaker’s intentions, 5, 13, 134, 147–9,

193

speaker’s meaning, 131, 148, 177
speech act content, 4, 5, 13, 40, 53,

57–8, 131, 145, 147, 150, 152,
153, 154, 165, 175, 176, 181,
182, 189, 190–207

privileged access, 202

speech act pluralism (SPAP), 2, 4, 5, 14,

97, 141, 143, 150, 154, 175, 176,
189, 190–207

central observation, 190–9
methodology, 191

Sperber, D., 5, 7, 32, 34, 35, 43, 62, 175
Stanley, J., 18, 19, 20, 22, 70–3, 76–9,

95, 102, 103, 154, 171, 175

Stone, Oliver, 198
storytelling context, 107–22
subpropositional content, 177

surprise indexical strategy, 12
surprising context sensitivity, 112
syntactic ellipses, 42, 47
Szabó, Z., 19, 20, 70–3, 76–9, 102, 103

target context, 107–22
Taylor, Ken, 7, 33, 34, 66, 67, 68
temporal reports, 11, 53, 94, 96
tense indicators, 145
Thurman, Uma, 158, 160, 164, 165
Travis, Charles, 6, 31, 32, 43, 47, 54, 56,

62, 63–4, 67, 109, 131, 132–6

true belief principle (TBP), 203
truth-conditional content, 197
truth-conditional intuitions, 197

unarticulated constituent strategy, 12,

unarticulated constituents, 22, 23, 159
underdeterminacy, 38

see also semantic underdetermination

Unger, Peter, 29
use-mention fallacies, 88, 116

vague predicates, 102
vagueness, 25, 26, 42
Vallée, Richard, 1
verb phrase (VP) ellipsis test, 101–3,

151

veriﬁcationism, 164

Walters, Vernon, 192
Watergate, 192–203
weather reports, 11, 30, 33, 53, 94, 96
wide context, 12, 13, 147, 148
Westerståhl, Dag, 103
Williamson, Timothy, 18
Wilson, D., 5, 7, 32, 34, 35, 43, 62, 175
Wittgenstein, Ludwig, 5, 6, 32

Index

219

Wyszukiwarka

Podobne podstrony:
Insensitive Semantics~ A Defense of Semantic Minimalism and Speech Act Pluralism
Little Words Their History, Phonology, Syntax, Semantics, Pragmatics, and Acquisition
In Defense of the Representational Theory of Qualia (Replies to Neander, Rey, and Tye)
Epidemic Profiles and Defense of Scale Free Networks
Estimation of Dietary Pb and Cd Intake from Pb and Cd in blood and urine
Development of Carbon Nanotubes and Polymer Composites Therefrom
Analysis of soil fertility and its anomalies using an objective model
Modeling of Polymer Processing and Properties
DICTIONARY OF AUSTRALIAN WORDS AND TERMS
A Chymicall treatise of the Ancient and highly illuminated Philosopher
Song of Myself Individuality and Free Verse
Extensive Analysis of Government Spending and?lancing the
Comparison of Human Language and Animal Communication
The?onomic Emergence of China, Japan and Vietnam
Hoban The Lion of Boaz Jachin and Jachin Boaz
Comparison of the Russians and Bosnians
Cruelty of Animal Testing Analysis of Animal Testing and A

więcej podobnych podstron