Full download test bank at ebook ebookmass.

com

Mood 1st Edition Paul Portner

CLICK LINK TO DOWLOAD

https://ebookmass.com/product/mood-1st-
edition-paul-portner/

ebookmass.com
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

From Melancholia To Depression: Disordered Mood In

Nineteenth-Century Psychiatry 1st Edition Edition Åsa
Jansson

https://ebookmass.com/product/from-melancholia-to-depression-
disordered-mood-in-nineteenth-century-psychiatry-1st-edition-
edition-asa-jansson/

Distillation 1st Edition Jean-Paul Duroudier

https://ebookmass.com/product/distillation-1st-edition-jean-paul-
duroudier/

Lectures on Imagination 1st Edition Paul Ricœur

https://ebookmass.com/product/lectures-on-imagination-1st-
edition-paul-ricoeur/

Bryozoan Paleobiology 1st Edition Paul D. Taylor

https://ebookmass.com/product/bryozoan-paleobiology-1st-edition-
paul-d-taylor/
Fluid Transport: Pipes 1st Edition Jean-Paul Duroudier

https://ebookmass.com/product/fluid-transport-pipes-1st-edition-
jean-paul-duroudier/

African Entrepreneurship 1st ed. Edition Leo-Paul Dana

https://ebookmass.com/product/african-entrepreneurship-1st-ed-
edition-leo-paul-dana/

OKRs For Dummies 1st Edition Paul R. Niven

https://ebookmass.com/product/okrs-for-dummies-1st-edition-paul-
r-niven/

Investing in Stocks For Dummies 1st Edition Paul

Mladjenovic

https://ebookmass.com/product/investing-in-stocks-for-
dummies-1st-edition-paul-mladjenovic/

POW/MIA Accounting 1st ed. Edition Paul M. Cole

https://ebookmass.com/product/pow-mia-accounting-1st-ed-edition-
paul-m-cole/
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

Mood

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

OXFORD SURVEYS IN SEMANTICS AND PRAGMATICS

general editors: Chris Barker, New York University, and Chris Kennedy,
University of Chicago
advisory editors: Kent Bach, San Francisco State University; Jack
Hoeksema, University of Groningen; Laurence R. Horn, Yale University; William
Ladusaw, University of California Santa Cruz; Richard Larson, Stony Brook
University; Beth Levin, Stanford University; Mark Steedman, University of
Edinburgh; Anna Szabolcsi, New York University; Gregory Ward, Northwestern
University

published
1 Modality
Paul Portner
2 Reference
Barbara Abbott
3 Intonation and Meaning
Daniel Büring
4 Questions
Veneeta Dayal
5 Mood
Paul Portner

in preparation
Aspect
Hana Filip
Lexical Pragmatics
Laurence R. Horn
Conversational Implicature
Yan Huang

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

Mood

PAUL PORTNER

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

3
Great Clarendon Street, Oxford, ox2 6dp,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Paul Portner 2018
The moral rights of the author have been asserted
First Edition published in 2018
Impression: 1
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, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2017939633
ISBN 978–0–19–954752–4 (hbk.)
978–0–19–954753–1 (pbk.)
Printed and bound by
CPI Group (UK) Ltd, Croydon, cr0 4yy
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

Contents
General preface vii
Acknowledgments viii
List of figures and tables ix

 Introduction 
. What do we study when we study mood? 
.. Conceptual preliminaries 
.. The general concept of mood 
. Main findings about the nature of mood 
. Background on modality 
.. Classifications of modality 
.. Modality in possible worlds semantics 
. The flow of information in discourse 
.. The dynamic approach 
.. Speech act theory 
.. Update potential and illocutionary force 
. Looking ahead 
 Verbal mood 
. Subsentential modality 
. Indicative and subjunctive 
.. Ideas about the indicative/subjunctive contrast 
.. Semantic theories of verbal mood in complement clauses 
.. Clauses which are not complements to a selecting predicate 
. Beyond verbal core mood 
.. Other mood-indicating forms 
.. The roles of semantics, syntax, and non-grammatical factors 
 Sentence mood 
. Sentence mood, clause type, and sentential force 
.. Clause types as grammatical categories 
.. Sentential forces as pragmatic categories 
.. The syntax/sentence mood interface 
. Sentence mood in speech act theory 
.. The performative hypothesis 
.. Adjustments to classical speech act theory 
.. The dynamic force hypothesis in speech act theory 
. Sentence mood in the dynamic approach 
.. Declaratives in the dynamic approach 
.. Interrogatives in the dynamic approach 
.. Imperatives in the dynamic approach 
.. Minor types: optatives and exclamatives 

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

vi contents

. Theories of clause type systems and sentence mood 

.. Clause type systems in speech act theory 
.. Clause type systems in the dynamic approach 
.. Other properties of clause type systems 
. Looking ahead 
 Core mood, reality status, and evidentiality 
. Prospects for a unified theory of core mood 
. Reality status and evidentiality 
.. Reality status 
.. Evidentiality 
.. Final remarks 

References 
Index 

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

General preface
Oxford Surveys in Semantics and Pragmatics aims to convey to the reader the
life and spirit of the study of meaning in natural language. Its volumes provide
distillations of the central empirical questions driving research in contemporary
semantics and pragmatics, and distinguish the most important lines of inquiry
into these questions. Each volume offers the reader an overview of the topic at
hand, a critical survey of the major approaches to it, and an assessment of what
consensus (if any) exists. By putting empirical puzzles and theoretical debates into a
comprehensible perspective, each author seeks to provide orientation and direction
to the topic, thereby providing the context for a deeper understanding of both
the complexity of the phenomena and the crucial features of the semantic and
pragmatic theories designed to explain them. The books in the series offer research-
ers in linguistics and related areas—including syntax, cognitive science, computer
science, and philosophy—both a valuable resource for instruction and reference
and a state-of-the-art perspective on contemporary semantic and pragmatic theory
from the experts shaping the field.
Paul Portner’s survey on mood provides a welcome new platform for the work
on a major but understudied topic in the semantics of natural language. The vast
majority of modern semantic studies have focused on the ways in which the
morphosyntactic and semantic properties of the constituent elements of declarat-
ive sentences—and to a somewhat lesser extent, interrogatives and imperatives—
interact with each other and with aspects of the context of utterance to determine
truth-conditional content. However, as an increasingly large number of scholars
have begun to appreciate, the full set of linguistically marked distinctions in
clause type is finer-grained than the traditional three-way distinction between
declarative, interrogative, and imperative reflects, richly subtle in its detail, and
indicative of a systematic relation between clausal and verbal morphosyntax and
constraints on the ways that a sentence can be used to perform a speech act.
Comprehensive, careful, instructive, and insightful, Portner thoroughly explores
the intricate tension between clausal and verbal morphosyntax and illocutionary
content, covering performativity, speech act theory, dynamic update, and eviden-
tiality. This volume bridges traditional studies of clause typing and contemporary
semantic and pragmatic theory, and lays the foundation for future advances in our
study of the relation between sentential morphosyntax and illocutionary force.
Chris Barker
New York University
Christopher Kennedy
University of Chicago

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

Acknowledgments
The idea for this book goes back to a proposal I made in 2003 to Oxford University
Press for a volume on modality and mood, and from that time I have appreciated
the patience and support of both the editors of the Oxford Surveys series, Chris
Barker and Chris Kennedy, and the editors at the press, especially John Davey and
Julia Steer. It turned out that it was far from feasible to include my thoughts on
mood and modality, and the relation between them, in a single volume. I am very
grateful that the press allowed me to pursue this project in two parts. I revised the
portion of the manuscript on modality into Modality (Portner 2009), and then set
out to write a second volume on mood. This latter part of the project turned out
to be much more difficult, because the range of relevant ideas in the literature is so
much more varied and disconnected, but for me the process of engaging with that
literature has also led again and again to the feeling that a new insight is waiting to
be understood. I hope that readers will experience to some extent the benefits, as I
have, of finding new ideas and connections within the literature on mood.
As I wrote the book I continued my various research projects on modality and
mood, and my collaborators, discussants, and reviewers during that time have
helped me immensely in this project. I can mention especially my collaborat-
ors Graham Katz, Elena Herburger, Miok Pak, Aynat Rubinstein, and Raffaella
Zanuttini. I received helpful comments on Chapter 3 from Malte Willer and on the
entire manuscript from an anonymous reviewer. In presentations of ideas which
flow from the perspective on mood adopted in this book, I have received crucial
feedback from Maria Aloni, Gennaro Chierchia, Nate Charlow, Liz Coppock, Kai
von Fintel, Anastasia Giannakidou, Magda Kaufmann, Angelika Kratzer, Alda
Mari, Craige Roberts, and Steve Wechsler. This feedback has in many cases led
me to rethink my understanding of the literature and trends present in current
research. I thank my students for sharing their insights and for help with research
and editing, in particular Lissa Krawczyk, Hillary Harner, and Akitaka Yamada.
My family has remained confident in my ability to complete this project and,
amazingly, convinced of its value through the long years that it hung about the
house as a competitor for my attention. I love the fact that my kids Noah and
Ben believe in the importance of research and writing, and I deeply appreciate the
respect my wife Sylvia holds for any project which has value to me. The feeling that
what one of us cares about, we all do, has given me the confidence to move forward
whenever the project seemed too complex and difficult. I dedicate the book to my
family and especially to Sylvia.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

List of figures and tables

Figures
1.1 Strength via accessibility relation 17
1.2 Modal base and ordering source 19
2.1 Entailment pattern under strong modal analysis 53
3.1 The question as function from worlds to answers 191
3.2 The question as a partition 191
3.3 The context as a partition 195
3.4 Asking a question in the partition context 196
3.5 The inquisitive context 198

Tables
1.1 Semantic classifications for mood and modality 7
1.2 Versions of the dynamic approach 28
3.1 Terminology for sentence moods 124
3.2 Structured discourse context of Portner (2004) 181
3.3 Main contributions of dynamic theories of imperatives 215

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 20/9/2017, SPi

i i

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

1
Introduction

Mood is a fundamental and traditional concept used in the description of human

language and in theoretical inquiries into the nature of language. Because it is
fundamental and traditional, one might think that the theory of mood is an
important topic within linguistics. Yet mood as such, in its most general sense, is
not a topic of research, and there are several reasons for this. One is that the concept
of mood is so fundamental that different scholarly traditions have molded their
understanding of what mood is to fit their conception of the nature of language.
We see it used in different ways in philosophy, formal semantics and pragmatics,
and descriptive/typological linguistics. And even setting aside broad differences in
approach to the study of language, mood is a difficult topic because all linguists
agree that it relates to at least the two different grammatical phenomena: verbal
mood, the category which includes indicatives and subjunctives, and sentence
mood, the category which includes declaratives, interrogatives, and imperatives.
It has generally not seemed likely that a general theory of mood incorporating
both of these categories would be possible, and so, in many instances, linguists take
care to explain that they are talking about one kind of mood and not the other. It
will emerge as we go forward, however, that similar ideas and puzzles recur as we
try to understand the two kinds of mood. They might not be as different as we
often assume.
In writing this book, I have two primary goals. The first is to provide scholars
interested in the semantics and pragmatics of mood with a deeper source of
background than is found in any single previous work. In doing this, I draw on as
much of the literature on mood as I am able, but I think and write from within the
formal semantics tradition. And second, I seek to formulate and advance new ideas
about the semantics and pragmatics of mood. As I survey, organize, and explain
the literature, I will sometimes sketch out novel analyses. In some cases, I do this
as a model for discussion without entirely endorsing the analysis, while in others,
I advocate for a certain hypothesis.
It must not go without saying that there is a great deal of insightful research
on mood which will not be mentioned in this book. The fields of linguistics and
philosophy are too large and the concepts of mood are too useful—they lead to
such a vast literature that even the subset of it that is worth studying today is so
extensive that no single book can incorporate a responsible discussion of it all.
What is included here is the result of my best judgment about what combination
of literature, explanation, and new ideas can best further the goals of the book.

Mood. First edition. Paul Portner.

© Paul Portner 2018. First published 2018 by Oxford University Press.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

2 introduction

There is a great deal of additional important scholarship on mood which I would

recommend. I hope that this book can help readers make better use of it.

Some terms and conventions. Before beginning the substantive discussion,

I would like to specify some of my terminology and conventions.

1. Notational conventions. In mentioning a word, phrase, or sentence:

(a) The use of small caps indicates a technical term whose precise meaning
is being discussed in the immediate context; in many cases, the term is
actually defined there.
(b) The use of boldface merely serves to draw the reader’s attention to a
particular word or words.
(c) The use of italics indicates that the material is being referred to as a
linguistic object. Exception: Within numbered examples and figures,
such material will not be italicized, unless there is other material within
the numbered example which is not being referred to as a linguistic
object.
(d) The use of ‘single quotes’ indicates that the material stands for the
linguistic objects in a number of languages, under the assumption that
all of those objects are similar enough across the relevant languages to be
spoken of as if they are the same. Example: “In most Romance languages,
‘believe’ selects the indicative.” Single quotes are also used for glosses and
paraphrases.
(e) The use of “double quotes” indicates that the quoted material should be
understood as spoken or written by someone. Double quotes are also
used in the traditional way as “scare quotes,” indicating that the quoted
material should not necessarily be taken literally.
(f) [Brackets] indicate some aspect of the context in which an example is to
be understood.
2. I use the term logical form in two ways:
(a) When capitalized, as “Logical Form,” it refers to a level of linguistic repres-
entation at which certain semantic properties, such as scope, are explicitly
represented. Logical Form can be a level of syntax or an independent level
between syntax and semantics.
(b) When not capitalized, as “logical form,” it refers to a representation
which displays or elucidates important abstract properties of meaning.
The logical form usually has some logic-like properties, such as a simple
vocabulary and an unintuitive syntax. Unlike Logical Form, a logical form
is not understood to be a level of linguistic representation.
3. When using a pronoun anaphoric to an antecedent which could refer to
a male or female individual, I roughly alternate genders. Example: “If a
philosopher reads this, she might disagree” (next time “he”). Occasionally,
however, it will be better to use “he or she,” “s/he,” or singular “they.”

In difficult cases, I will apply these conventions in the way I feel is most helpful.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

what do we study when we study mood? 3

1.1 What do we study when we study mood?

1.1.1 Conceptual preliminaries
One of the most fundamental issues to be addressed by semantic theory concerns
the kind of meanings that simple declarative sentences have, and we can judge
any approach by how well it helps us understand such basic properties as: they
can be (and typically are) either true or false; they can be used to share factual
information; they are appropriate for representing the contents of thought and
some other cognitive states; and they stand in logical relations to one another like
entailment and synonymy. Consider the declarative sentence (1a). It has all of these
properties:

(1) a. Ben is holding a bird.

b. This is true because: Ben is holding John Fei.
c. Noah told me: “Ben is holding a bird.” And now I know what Ben is
holding.
d. Noah thinks that Ben is holding a bird.
e. Ben is holding a bird. Entails: Ben is in close proximity to a bird.

Moreover, semanticists think they have a good (though not yet perfect) under-
standing of how each of these properties should be explained, an understanding
which is based on the theoretical construct of a proposition.
In formal semantics, our theories of sentence meaning most commonly work
with the idea that propositions can be defined in terms of possible worlds. A
possible world is a way things could be, complete through space and time, an
alternative history of the universe. Our own universe-history can be referred to
as the “actual world” or “real world.” For the purposes of linguistic semantics,
let us assume that we have a set of possible worlds conceivable by humans. By
this I mean that any difference in how things could be which a human could
recognize, imagine, or describe corresponds to a difference between possible
worlds in the set of all worlds W. For example, if I tell you that Ben has a
cockatiel, you can imagine that it is grey, or that it is white. Therefore, W
should contain at least one possible world in which it is grey and at least one
in which it is white. The set of worlds conceivable by humans seems sure to be
adequate for doing natural language semantics, if any theory based on possible
worlds is.
A proposition in possible worlds semantics is a subset of W. For example, the
meaning of (1a) is, or can be characterized in terms of, the set (2):

(2) {w : Ben is holding a bird in w}

(= the set of worlds in which Ben is holding a bird.)

This definition of proposition helps to explain the properties summarized in (1).

For example, (1b) amounts to the claim that the actual world is one of the worlds
in the set. Though the conception of propositions as sets of possible worlds is not

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

4 introduction

without its problems, we can ignore them for now. Our goal is to use it to develop
a useful way of thinking about mood.
Another feature of language we need to understand if we are to get a handle
on the concept of mood is modality. Modality is “the linguistic phenomenon
whereby grammar allows one to say things about, or on the basis of, situations
which need not be real” (Portner 2009, p.1). The noun Ben in (1a) is not modal; it
is used to refer to a particular, actual person. In contrast, the auxiliary should in (3)
is modal:

(3) Ben should put down the bird.

This sentence says that situations in which Ben puts down the bird are in some
respect better than, or preferable to, situations in which he does not. Since he either
will or will not put down the bird, some of these situations will never be real, and
so the word should counts as modal.

1.1.2 The general concept of mood

With these preliminaries in place, we can attempt a definition of mood:

Mood is an aspect of linguistic form which indicates how a proposition is used

in the expression of modal meaning.

As we start out, it’s acceptable to be vague about what we mean by the “use” of a
proposition, because linguists have employed the term “mood” in many different
ways. Let us consider two examples:

1. In many languages, subordinate clauses (and sometimes root clauses as well)

come in different forms, known as verbal moods, such as indicative mood
and subjunctive mood. These clauses can be used to help represent various
cognitive states and mental events, such as beliefs, desires, and dreams. When
they are used in a subordinate clause to help talk about a desire (as in (4)),
they typically take the subjunctive form; in contrast, when they are used to
describe a dream ((5)), they typically take indicative form.

(4) Pierre veut que Marie soit heureuse. (French)

Pierre wants that Marie is.subj happy
‘Pierre wants Marie to be happy.’
(5) J’ai rêvé qu’il était président.
I dreamed that he was.indic president
‘I dreamed that he was president.’

 Under certain circumstances, they may take a different form, such as infinitive. For our purposes

here, it’s helpful to pretend there are only two possibilities.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

what do we study when we study mood? 5

We call this kind of mood “verbal mood” because it is frequently (though not
always) marked in a grammatical sense on the verbal head of the clause; for
example, the verb soit is a subjunctive form of ‘be’ in French.
2. In all languages, root sentences have various functions, including easy-to-
intuit ones like directing somebody to do something (as in (6a)), requesting
information ((6b)), and providing information ((6c)).

(6) a. Pick up a bird!

b. Is Ben holding a bird?
c. Ben is holding a bird.

Each of these functions is associated in an intuitively clear way (which is

nonetheless hard to define) with particular members of a paradigm of forms
known as sentence moods. The most prominent sentence moods are the
three just illustrated: the imperative, interrogative, and declarative.

Each of these two important concepts of mood realizes the general description of
mood indicating how a sentence’s proposition is to be used: verbal mood tells us
something about how it is to be used, within the compositional computation of
meaning, to describe an individual’s mental life, while sentence mood indicates
how it is to be used, in a multi-party exchange, to achieve specified communicative
functions.
Given an understanding of mood like the one just developed, what would a
linguistic theory of mood look like? We could spell out a too-simple theory of
verbal mood like this: Assume that verbs like want and dream, verbs which take
sentential complements and talk about some aspect of their subject’s mental life,
are modal words which express a relation between two arguments, an individual
and a proposition. Their logical form can be represented as V(x, p), for example
wants(Pierre, {w : Marie is happy w}). Mood choice is determined by the following
principle:

(7) If the relation expressed by the verb concerns a preference about how the
future will be, the clause which denotes the proposition argument of this
relation should be in the subjunctive mood. Otherwise, it should be in the
indicative mood.

Given this principle, if you want to say something which means wants(Pierre,
{w : Marie is happy w}), the verb ‘be’ will be subjunctive, but if you want to say
something which means dreams(I, {w : Pierre was president w}), it will be indicative.
Nothing semantic changes between the two cases, other than the main relation,
wants or dreams.
And we could spell out a too-simple theory of sentence mood like this: Assume
that the function of a root sentence used in dialogue is to adjust the speaker’s and
hearer’s shared assumptions. For example, sometimes the speaker may want to
create a shared assumption that one way the future could be is preferable to another,
perhaps threatening some sort of punishment upon the addressee if the preferred

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

6 introduction

future does not come about. We can describe this as the speaker directing the
addressee to do something. There is a particular grammatical form, the imperative,
for a sentence used with this purpose, as in (6a). This reasoning suggests the
following principle:

(8) If a sentence is used to create an assumption that it is preferable that the

proposition expressed by the sentence comes to be true (as opposed to false),
it should be expressed using imperative mood.

There would be similar principles for interrogative sentences (Will Ben put down
the bird? creates an assumption that the addressee will help the speaker know some-
thing) and indicative sentences (Ben will put down the bird creates an assumption
that a certain fact holds). Nothing would differ among these cases in terms of the
proposition involved, but there would be crucial differences in the goal which the
speaker aims to achieve by using a sentence which denotes that proposition.
These pictures of subjunctive and imperative clauses have been presented to
help convey an understanding of the idea behind the informal definition of mood.
This conception of mood will be very important in the book, as it serves as an
unarticulated intuition behind the actual practices linguists have in describing
phenomena as “mood,” and because it plays a role in many attempts to provide
concrete semantic or pragmatic analyses of mood forms. However, the specific
statements about the subjunctive and imperative above are not to be taken as
serious proposals. Besides oversimplifying the relevant phenomena, they make
many assumptions which could turn out to be wrong: for example, they assume that
all sentences have propositions as their basic meanings and that the various moods
have no effect themselves on the meaning of a given sentence. All such assumptions
must be carefully evaluated as part of any serious investigation of mood. Much
of the work in this book is to examine some of the phenomena which meet the
characterization of mood based on the concept of “modal use” and to consider
various theories of them. In the next subsection, I give a preview of where this way
of thinking about mood will lead us.

1.2 Main findings about the nature of mood

The literature on mood is extensive, contradictory, and at times confusing.
Nevertheless, I hope we won’t end up in a muddle. The discussion will lead not only
to lots of questions, but to a few linguistically significant conclusions as well. Here
at the outset, I would like to highlight the work’s most basic, broadly conceptual
findings:

1. While verbal mood and sentence mood are distinct both in terms of
morphosyntax and in terms of meaning, they are closely related. They
are related because of tight parallels between the modal semantics of
sentence-embedding constructions, which determine verbal moods, and the

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

main findings about the nature of mood 7

communicative functions served by root sentences in conversation, in terms

of which sentence moods are characterized.
2. At a formal level, many theories of verbal mood and sentence mood can be
expressed within a unified framework. This framework states that both kinds
of mood encode the interaction between a clause’s meaning and a contextually
given partially ordered set of worlds (the posw). The posw may be
derived from either the grammatical or the discourse context. Though the
posw-framework is clearly not the last word on the relation between verbal
mood and sentence mood, it can point the way towards a more general and
explanatory theory of the two kinds of mood and the relation between them.
On the assumption that some types of verbal mood and some types of
sentence mood can be analyzed in terms of a single theoretical system (be
it the posw or something else), I label the broader linguistic category which
includes them both core mood.
3. There is a range of forms and constructions across languages whose relation
to core mood is not currently well understood. Among these are evidentials,
the kinds of forms which Elliott (2000) describes as marking “reality status,”
and numerous other specific moods and modes used in the description
of languages typologically different from those which have been the focus
of research on verbal mood. We do not yet have a good enough under-
standing of all the forms with labels like “realis,” “irrealis,” “conjunct,” “in-
ferential,” “hearsay,” “conjecture,” “potential,” “desiderative,” “assertive,” and
“contingent” to say how they are related to core mood.

I think that core mood covers more than what linguists typically think of as mood,
rather than less, and I don’t know of any linguistic forms which at once should
clearly be classified as mood, yet also clearly not as core mood. However, some types
of elements, for example evidentials, which I think could reasonably be thought
of as core mood, are not typically thought about in that way, and so they may
exemplify non-core mood. More significantly, few phenomena which have been
described in terms of the concepts of reality status have been analyzed in a precise
enough way for it to be clear what their relation to core mood is. Table 1.1 outlines

Table . Semantic classifications for mood and modality

Modality
Core mood Non-core mood (The rest of modality)
Verbal Sentence Other core
mood mood mood

indicative, imperative, evidentials? evidentials? epistemic, priority,

subjunctive, declarative, deontic, dynamic
certain interrogative reality status? reality status? modals; modal
infinitives adjectives, adverbs

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

8 introduction

this way of thinking about the relationships among various kinds of mood and
between mood and modality.

1.3 Background on modality

As described above, for the purposes of this book I define modality as the linguistic
phenomenon whereby grammar allows one to say things about, or on the basis of,
situations which need not be real. Since, on this conception, mood fits within this
broader category of modality, it will be essential to have an understanding of the
semantics of modality. In this section I will provide an overview of some key ideas
in the semantic theory of modality, drawing on the more extensive discussion by
Portner (2009).

1.3.1 Classifications of modality

We begin with two ways of classifying modal expressions: by the linguistic level at
which they are expressed and by the meanings they convey.

Classifications of modality by level of linguistic organization. Modality may

be classified according to the level of linguistic organization on which it operates.
Specifically, it’s useful to distinguish subsentential, sentential, and discourse
modality.

1. Subsentential modality operates below the level of proposition expressed by a

complete sentence. It includes modal adjectives, modal nouns, propositional
attitude verbs (verbs which take an argument which expresses a proposition),
verbal mood, and infinitives, among other types.

(9) (a) A possible solution to this problem is to call the recalcitrant reviewer.
(b) The probability of success is low.
(c) I think/hope/regret that she arrived on time.
(d) I hope to be happy.

Of course, many of these elements also affect the meaning of the complete
sentence, but they do so via the meaning of some smaller constituent. For
example, the subject phrases in (9a–b) have noun phrase-type meanings
which have been built up using the modal concepts expressed by possible and
probability, and the predicate in (9c) denotes a property, like other predicates
do, but this property involves consideration of not-necessarily-real situations
which are important in the speaker’s mental life.
2. Sentential modality operates at the level of the complete proposition. In other
words, if a sentence contains a constituent which denotes a proposition, and
then a modal element combines with this to create another propositional
constituent, we have a case of sentential modality. In grammatical terms,
this means that it is typically realized above the level of the main subject–
predicate structure in the clause, that is, above the S, IP, vP, or other roughly

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 9

equivalent syntactic unit. Sentential modality affects the primary meaning of

the sentence; thus, in the case of a root declarative clause, it affects the truth
conditions of the sentence; in the case of a root interrogative, it affects what
information is literally requested, and so forth.
Sentential modality has commanded the vast majority of scholarly atten-
tion on modality within linguistics. In English, there has been a great deal
of work on the modal auxiliaries like can in (10a). We can also classify as
expressing sentential modality such elements as modal verbs (by this I mean
verbs which are not auxiliaries, but which are distinct from regular lexical
verbs as well, as in (10b)), modal adverbs, and some tense and aspect forms.

(10) (a) Noah can swim.

(b) Devo partire domani. (Italian; Squartini 2004)
must.pres-1sg leave tomorrow
‘I must leave tomorrow.’
(c) Probably, Ben was there.
(d) She will be late.
(e) The team is building a bridge.

In some cases, the decision whether to treat a given modal element as

representing subsentential or sentential modality requires a careful syntactic
and semantic analysis. For example, my labeling (10b) as sentential is based
on the assumption that dovere (‘must’) is syntactically distinct from regular
lexical verbs, and the idea that the progressive in (10e) exemplifies sentential
modality assumes that it is a sentence-level operator. If either of these points
turns out not to be correct, then they are not examples of sentential modality.
3. Discourse modality is the direct contribution of modality to meaning in
discourse. I say “direct contribution” because, of course, any type of modality
ultimately contributes to the broader discourse; what distinguishes discourse
modality is that it does not do so via an effect on the regular semantic meaning
expressed by a sentence. Though (10b) may cause the discourse to contain the
information that the speaker favors future situations in which she leaves over
ones in which she does not, this effect is mediated by its ordinary sentence-
level semantics; hence, it is not discourse modality. In contrast, discourse
modality is separate from any modal contribution towards regular sentence
meaning. Some plausible examples of discourse modality are evidentiality,
sentence mood, and the performative meanings of modal auxiliaries and
modal verbs.

(11) (a) Para-sha-n-mi/-si/-chá. (Cusco Quechua, Faller 2006b)

rain-prog-3-bpg/rep/conj
‘It is raining.’ (Direct evidence/Reported/Conjectural.)
(b) Leave right away!
(c) It might rain.

In (11a), we see the three evidential markers in Cusco Quechua, the first
indicating that the speaker has direct evidence (what Faller calls ‘best possible

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

10 introduction

grounds’), the second that he is conveying a report, and the third that he is
making a conjecture. In (11b), we have an imperative. These two exemplify
discourse modality on the assumption that the evidentials and imperative
form do not cause the sentence to denote a modalized proposition—for
example that (11a) does not mean ‘I have direct evidence that it is raining’ (if
it did mean this, it would show sentential modality), but rather conveys this
meaning without affecting the primary proposition ‘it is raining.’ Example
(11c) can be considered discourse modality if we accept the claim (Portner
2008) that it not only means that rain is compatible with our information
(its sentential modality), but also makes the question of whether it will rain a
topic of conversation (additional discourse modality).

The examples of each subtype above are given only in order to help make clear what
should fall under each subcategory. It wouldn’t be surprising if further research
showed some of them to be miscategorized or even not modal at all. For example,
while many think that the progressive is modal, this is somewhat controversial
(Portner 2011a).

Classifications of modality by meaning. Modality involves saying something

about, or on the basis of, situations which need not be real, and it is useful to
classify modal elements according to the way in which its meaning is based on
those situations. Roughly speaking, (12) says that the relevant situations (those in
which the girl gets the prize) are good (perhaps more specifically, they are good in
relation to the criteria set up for the contest):

(12) That girl should get the prize.

In the literature on modality, we find various systems for classifying modal ele-
ments, especially sentential modals, along such parameters; in Portner (2009),
I outline a top-level classification into epistemic, priority, and dynamic modality,
with various subtypes:

1. Epistemic modality has to do with what can be concluded based on

someone’s knowledge.
True epistemic modality is not commonly divided into subtypes, but in
logic and philosophy, there is consideration of such related concepts as
alethic modality (concerning concepts of logical possibility and neces-
sity) and metaphysical modality (concerning metaphysical possibility and
necessity). It is not clear that natural language modal elements ever express
these concepts, except in the context of technical logical/philosophical dis-
course. Historical modality is a kind of metaphysical modality where the
relevant worlds are those with a future that is compatible with everything that
has already occurred, and should perhaps be classified as a kind of dynamic
modality.
2. Priority modality has to do with reasons for assigning priority, or prefer-
ence, to one type of situation over another.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 11

There are several subtypes, such as: deontic modality concerns priority
based on rules or right and wrong, buletic modality concerns priority based
on desire, and teleological modality concerns priority based on goals. We
do not assume that these subcategories are mutually exclusive.
3. Dynamic modality has to do with the possible courses of events in the
world, based on the factual circumstances.
The most prominent subtype is volitional modality. Volitional modals
concern the actions available to a volitional individual, with sub-subtypes
including ability modality (focus on the individual’s abilities), opportun-
ity modality (focus on the circumstances surrounding the individual), and
dispositional modality (focus on the individual’s dispositions). There are
also forms of dynamic modality which are not tied to a volitional individual,
and I will call these intrinsic modality. A somewhat special variety of
dynamic modal are the quantificational modals, which seem to involve
quantification over individuals.

Examples of all of these subtypes from Portner (2009, ch.4) are given in (13)–(15):

(13) Epistemic
(a) A typhoon may hit the island.
(b) Mary must have a good reason for being late.
(14) Priority
(a) Deontic: The rich must give money to the poor.
(b) Buletic: You should try this chocolate.
(c) Teleological: You could add some more salt to the soup.
(15) Dynamic
(a) Volitional:
(i) John can swim. (ability)
(ii) You can see the ocean from here. (opportunity)
(iii) Mary will laugh if you tell her that. (dispositional)
(b) Intrinsic:
(i) The cup is breakable.
(ii) Every empire eventually falls. (historical)
(c) Quantificational:
(i) A spider can be dangerous. (existential)
(ii) A spider will be dangerous. (universal)

Following Kratzer (1981, 1991), priority and dynamic modality are often grouped
together as circumstantial modality (i.e. modality which makes reference
to factual circumstances, rather than only an individual’s knowledge or beliefs),
and following this terminology, dynamic modality can be called “pure” cir-
cumstantial modality (circumstantial modality where priorities do not play
a role).
One feature of traditional classifications which might be less than ideal is the
great difference it implies between two uses of words like likely, certain, and chance.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

12 introduction

According to the standard classification above, likely is ambiguous between an

epistemic meaning in (16a) and a dynamic meaning in (16b).

(16) a. Given the evidence, Bob is likely to be hiding in the basement.

b. The way that storm is developing, it is likely to spawn a tornado.

It is useful to have a cover term for types of epistemic and dynamic modality which
express either subjective or objective chance; we can use the term predictive
modality for this class.
I will refer to the differences in meaning along the dimensions outlined in (13)–
(16) as differences of judgment type. (In the literature and especially in the spoken
jargon of semantics, they are often described as “flavors” of modality.) See Portner
(2009) for discussion of other classification schemes. Note that many of the above
examples involve English modal auxiliaries. Other varieties of modality are found
in other languages, and in other constructions within English, but these have not
made their ways into the generally shared terminology of semanticists.
Yet another parameter along which modal meanings vary is that of strength.
Scholars who study modality make at least a two-way distinction between strong
and weak modals, although not every language may actually have modals of both
strengths (Deal 2011). Both should and may can be deontic, but (17a) is stronger
than (i.e. it entails) (17b):

(17) (a) He should vote for her. (strong)

(b) He may vote for her. (weak)

Similar oppositions exist within each of the other subtypes of modality, as can be
seen in (13)–(15). In those examples, must, should, and will would be classified as
strong modals, while the others would be classified as weak. Strong modals are
sometimes called necessity modals, and weak ones possibility modals, on the
grounds that it is necessary that is strong and it is possible that is weak.
When we look beyond modal auxiliaries, it becomes clear that strength is not
a two-way distinction. In the following, (18a) is stronger than (18b), and so forth
down the line:

(18) (a) It will certainly rain.

(b) It will almost certainly rain.
(c) It will probably rain.
(d) There is a reasonable chance that it will rain.
(e) It is just possible that it will rain.

It seems that modal strength is gradable, and it is natural to think of this gradability
as being similar to the gradability of concrete properties, such as height and weight.
(Gradable modality is currently a topic of much study in semantics; see for example

 Rubinstein et al. () find that speakers have significant difficulties making the distinction

between epistemic and dynamic modality in certain texts.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 13

Portner 2009; Yalcin 2010; Katz et al. 2012; Klecha 2014; Lassiter 2016.) As we see in
these few examples, variation in strength can arise from a combination of lexical
choice (certainly is stronger than probably) and compositional semantics (almost
certainly is weaker than certainly). Among strong modals, elements including
should and ought are sometimes called “weak necessity modals,” because they feel
weaker than other (“strong”) necessity modals like must. (The terminology is some-
what confusing here, because a weak necessity modal is still a strong, or necessity,
modal by our terminology. A weak necessity modal is the weaker subtype of strong
modal.) It’s not yet clear whether weak necessity modals are logically weaker than
their strong counterparts, or whether they differ in meaning in some other way
(e.g. von Fintel and Iatridou 2008; von Fintel and Gillies 2010; Rubinstein 2012).

1.3.2 Modality in possible worlds semantics

In order to fully engage with the ideas and debates in the analysis of mood which
will occupy the bulk of this book, it is important for readers to have a good
understanding of semantic theories of modality. The goal of Portner (2009) was
to introduce the analyses of modality within four semantic frameworks: modal
logic, Kratzer’s ordering semantics, dynamic semantics, and cognitive semantics.
I cannot recapitulate that discussion here, but instead suggest that readers without
sufficient background begin with that book or other resources on the semantics
of modality. In this section, I will review some of the key ideas of the important
tradition of modal semantics based on possible worlds. This tradition began within
modal logic and has evolved into the standard account within formal (linguistic)
semantics. It is of the most relevance here because the bulk of theoretically-
informed work on mood takes place within this framework. My goal with this
review of possible worlds analysis of modality is to emphasize aspects of the
approach which will prove especially important in the rest of the book, to establish
my own terminological and notational preferences, and to jog the memories of
readers who may know about the semantics of modality, but not have thought
about it in a while.
The standard theory of modality within possible worlds semantics focuses on
sentential modal constructions, and aims to explain their judgment type and
strength.

Judgment type. Within possible worlds semantics, judgment type is determined

by the particular set of possible worlds on which the modal’s meaning is based.
For example, with the epistemic example (13a), the judgment being made is that a

 Subsentential and discourse modality are considered important topics to which the theory should

be extended (and much of this book concerns such attempts), but not exemplars of the standard
theory as it stands. An exception to this statement is subsentential modality that gets treated as if it
were sentential modality, for example the use of modal adjectives with sentential complements: It is
necessary/likely/possible that it is raining. These treatments generally ignore features of the subsentential
modal constructions which are not shared by sentential modals.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

14 introduction

typhoon hitting the island is compatible with the speaker’s knowledge. Suppose we
identify the set K of worlds compatible with what the speaker knows; in that case,
we can express the judgment of (13a) by saying that some of the worlds in K contain
situations in which a typhoon hits the island. In more formal terms, modal logic
bases the semantics of a particular modal element on an accessibility relation
between worlds R. In (13a), we might use Rep :

(19) For any worlds w∗ , and w: Rep (w∗ , w) iff everything the speaker knows in w∗
also holds in w.

Taking w∗ to be the actual world, the set of accessible worlds K is {w : Rep (w∗ , w)}.
We can define the truth conditions of (13a) as follows:

(20) A typhoon may hit the island is true in a world w∗ iff there is some world w
such that both Rep (w∗ , w) and a typhoon hits the island in w.

Though the definitions in (19)–(20) show technically how an accessibility relation

works (and so might be useful for some purposes), it is far too simple. For instance,
it assumes that the only person whose knowledge we care about is the speaker’s
and the only knowledge of his that we care about is that which he has at the present
moment. If there are examples where the knowledge of an individual other than
the speaker, or at a time other than the present, is relevant, we’d need a different
accessibility relation.
Let us examine some data to see who can be the “knower” and what can be the
“knowing time” in cases of sentential epistemic modality.

(21) [Two children are discussing whether the creature they caught is a newt or a
salamander.]
(a) This might be a salamander.
(b) It might have been a salamander.
(c) Ryan said that it might be a salamander.

In (21a), the knower is either the child speaking or the two children jointly, and
the knowing time is the speech time. In (21b), the knower is again the speaker
or two children jointly, while the knowing time could be either the speech time
or some time in the past, for example, when the children still had the creature in
their hands. (To see the latter possibility, consider the continuation . . . but it turns
out it wasn’t.) In (21c), Ryan is the relevant knower, and the knowing time is the
(past) time at which he spoke. Clearly, the details of the accessibility relation can
vary from case to case, and yet this variation is very much limited by grammatical
factors. It would be very difficult for one of the children to use (21a) to make a
modal statement based on what Ryan knew.
It is an important goal for modal semantics to come up with an adequate theory
of the kind of variation illustrated in (21). One way to do this is to incorporate a
context situation s into the definition of the accessibility relation. The knower and
knowing time are extracted from the context situation, while the context situation

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 15

itself (as the name suggests) is determined by the linguistic or extralinguistic

context, in ways to be determined:

(22) For any context situation s and worlds w∗ , w: Rep (s)(w∗ , w) iff everything
that the thinking participant(s) of s know in w∗ at the time of s also holds
in w.

There is some redundancy in (22) because the set of accessible worlds depends on
both a context situation s and a world w∗ . But if we assume that each situation is
only part of a single world, and that it only determines accessibility from the world
of which it is a part, we can base the accessibility relation on the context situation
alone, as follows:

(23) For any context situation s and world w: Rep (s, w) iff everything which the
thinking participant (or participants) of s know in s also holds in w.

In root sentences, it is always possible (and usually preferred, if not outright

required) for s to include the speaker as among the thinking participants, and
for the time of s to include the time at which the sentence itself was used. Thus,
the typical interpretation of a root sentence with might concerns the speaker’s
knowledge at the time of utterance. This is what we see in (21a). However, the time
can diverge from the speech time, particularly in the presence of perfect aspect
(have been in (21b)), and it’s even possible for the thinking participants not to
include the speaker, for example in a narrative. In embedded clauses, the situation
s is typically determined by grammatical factors; in (21c), we see the matrix verb
said controlling the interpretation of the modal in the embedded clause by making
sure that Ryan is the thinking participant of s.
Other judgment types can be associated with accessibility definitions similar to
(23), such as the following for a deontic modal used to make a statement about what
the law requires (for example, That guy should put a coin in the parking meter):

(24) For any context situation s and world w: Rlegal (s, w) iff all of the laws in force
in s are fully complied with in w.

We describe a world accessible by the relation associated with an epistemic modal

as an epistemically accessible world. Similarly, we may talk about deontic-
ally accessible worlds, buletically accessible worlds, and so forth. We can
also use less esoteric language like “knowledge worlds” (for epistemically accessible
ones) and “desire worlds” (for buletically accessible ones).

Strength. Differences in strength are analyzed within modal logic and semantic
systems closely based on modal logic as a difference between universal and
existential quantification. The strength of a given modal is also known as its

 Hacquard () argues that the meanings expressed by modal auxiliaries are relative to events in

a way analogous to (). She also assumes that the judgment type can be determined from s, so that a
single general-purpose R can work for all flavors of modality.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

16 introduction

modal force. Within the meaning of the weak modal might in (13a), we find the
phrase “there is some world,” an existential quantifier over accessible worlds. A
strong modal like must, in contrast, is used to make a statement to the effect that all
relevant worlds have a certain property. In other words, it is a universal quantifier
over accessible worlds. The universal meaning of epistemic must in (13b) can be
stated as follows:

(25) Mary must have a good reason for being late, used in a context situation s in
w∗ , is true in w∗ iff, for every w such that Rep (s, w), Mary has a good reason
for being late in w.

This is like (20) except that it involves universal quantification and it has been
improved through the inclusion of the context situation motivated above.
The quantificational analysis of modal strength in standard possible worlds
semantics only can distinguish two strengths, strong (universal quantification) and
weak (existential quantification), and it is clear that this is not enough to explain
the full range of variation in strength. This point is most clear with subsentential
modality, as illustrated by (18) above, but it may also be seen in the relation
between strong and weak necessity modals. Many authors (including von Fintel
and Iatridou 2008; Finlay 2010; Kolodny and MacFarlane 2010; Rubinstein 2012;
Lassiter 2016) discuss the fact that the strong necessity must and have to appear to
be stronger than weak necessity should and ought to (example from von Fintel and
Iatridou):

(26) Everybody ought to wash their hands; employees must.

The simplest approach to explaining intuitive differences in strength beyond the

basic strong–weak contrast focuses on the choice of accessibility relation. The idea
here follows from a logical observation. Suppose we have two accessibility relations
R and R such that the set of worlds accessible by the former is always a subset of
the set of worlds accessible by the latter:

(27) For any context situation s: {w : R(s, w)} ⊆ {w : R (s, w)}.

Then, if we have weak (existential) modal sentences differing in that the first uses
R and the second R , the first will entail the second. The reverse holds for strong
(universal) modal sentences. To see why, consider Figure 1.1, based on Portner
(2009). At a given situation s, R makes accessible the set of worlds indicated by the
arrow labeled R, and R makes accessible a superset of that set. A weak modal is
true, using a given accessibility relation, if some accessible world is one in which
the portion under the scope of the modal (call it S) is true. The portion of the
diagram on the left shows the situation relevant to a weak modal: if S is true at
some world in the smaller set accessible by R, it will obviously be true at some
world accessible by R (at least, the very same world). The right-hand side of the
diagram shows that the opposite holds with a strong modal: if S is true at all worlds
accessible by R , it will obviously also be true at all worlds accessible by R. To
summarize, we have the following:

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 17

(28) For any sentence S and any accessibility relations R and R related as in (27):
(a) ♦R (S) entails ♦R (S)
(b) R (S) entails R (S)

(The symbol  is drawn from modal logic to indicate any modal whose meaning is
expressed using universal quantification over worlds, while ♦ is the corresponding
symbol for a modal whose meaning is expressed using existential quantification.)
We might use the observations in (28) to analyze (26) by saying that the accessibility
relation associated with ought makes accessible a subset of the worlds made
accessible by that associated with must. In other words, ought is like R in (28),
while must is like R .
The main issue with trying to explain differences in modal strength in terms of
subset relations among accessibility relations is that it becomes unclear precisely
what content is associated with each relation. In the case of (26), we are to assume
that ought’s R determines a subset of must’s R , but precisely which worlds do R(s)
and R (s) make accessible? Intuitively, we want to say that R(s) makes accessible all
worlds in which both the very important and the less important rules applicable in
s are followed, while R (s) makes accessible all worlds in which the very important
rules in s are followed, but possibly not the less important rules. It would be helpful
if we could explicitly define the accessible worlds for ought and must in terms
of the rules of varying degrees of importance. Kratzer’s premise-based ordering
semantics for modality discussed next allows us to think of differences in modal
strength in precisely this way.

weak modal strong modal

S is true S is true

w∗ R w∗ R

R R

Figure . Strength via accessibility relation

Kratzer’s ordering semantics. In Portner (2009), I presented two versions of

modal logic based on the ideas about modal force and judgment type outlined
above. These theories, a basic modal logic similar to what is taught in introductory
modal logic texts and a “modal logic for linguists” which modifies it to deal with
some of the most obvious differences between natural languages and logical ones,
are useful because they are widely understood and serve as a precise baseline
against which other theories can be measured. While forms of modal logic like
those are useful, they are not taken as realistic theories of modal semantics in
natural language. They simply do not allow one to express in a theoretically
perspicuous way many important ideas about modal semantics and pragmatics.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

18 introduction

For this reason, linguists have extended and modified the framework of modal logic
to produce more linguistically useful theories of modality.
The standard theory of modality within formal semantics was developed
by Kratzer (1977, 1981, 1991, 2012). This influential approach extends the basic
possible worlds analysis of modality in several ways. Perhaps the most important
development is that the modals do not simply classify a possible world as either
wholly accessible or inaccessible, but rather rank or order them according to
some relevant criteria. Exactly which worlds are accessible in a given case can be
determined in a flexible way based on this ordering. We can label Kratzer’s theory
as a prime example of the ordering semantics approach to modality.
Portner (2009) provides a detailed introduction to Kratzer’s theory, but it will be
useful to have a brief review here. One key concept in her framework is that of a
conversational background. A conversational background is a function from
situations to sets of propositions, and can be given by context, linguistic material,
or a combination of the two. A conversational background can serve either of two
basic functions. It can specify a set of relevant worlds or it can define an ordering of
worlds. In its former function, a conversational background is known as a modal
base, while in the latter function, it is known as an ordering source.

(29) Conversational backgrounds

(a) A conversational background is a function from situations to sets of
propositions. (Its domain is S and its range P(W).)
(b) Functions of conversational backgrounds
(i) Modal base. For any situation s, a conversational background m
(functioning
 as a modal base) defines a set of relevant worlds
m(s).
(ii) Ordering source. For any situation s, a conversational background
o (functioning as an ordering source) defines an ordering ≤s , as
follows:
For any worlds w and v, w ≤s v iff, for all p ∈ o(s), if v ∈ p, then
w ∈ p.

The modal base defines as relevant (at a given situation s) those worlds in which all
propositions in the modal base are true. The ordering source ranks worlds so that
one world is ordered ≤ another (with respect to a given situation s) if and only if
all of the propositions in the ordering source which are true in the latter are also
true in the former. So for example:

(30) (a) Modal base: {‘It is hot’, ‘It is sunny’}.

(b) Ordering source: {‘We have cold drinks’, ‘We have a parasol’}.

The relevant worlds defined by the modal base are ones in which it’s both hot and
sunny. The ordering defined by the ordering source ranks those worlds in which

 In Kratzer’s work, a conversational background is a function from worlds to sets of propositions.

I will treat the domain as situations rather than worlds, in order to incorporate the idea of a context
situation.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

background on modality 19

We have cold drinks and a parasol. † Relevant:

w
It’s both hot Ordering w≤v:
and sunny. v

We have We have a parasol;

cold drinks; no cold drinks.
no parasol.

We have neither cold drinks nor parasol.

Figure . Modal base and ordering source

we have both cold drinks and a parasol the highest, worlds in which we have either
cold drinks or a parasol less high, and worlds in which we have neither cold drinks
nor a parasol the lowest. (Worlds in which we have cold drinks but no parasol, and
those in which we have a parasol but no cold drinks, are not ordered with respect
to each other.) These relations are illustrated in Figure 1.2.
Within ordering semantics, the meanings of modal operators are defined in
terms of structures like the one illustrated in Figure 1.2. In simple cases (like the
one in the figure), the modal base and ordering source combine to define a set
of accessible worlds which can be thought of in pretty much the same way as
the accessibility relation in modal logic. The accessible worlds are the best-ranked
relevant worlds; this set is indicated by † in Figure 1.2 (i.e. the grey-shaded worlds
in the upper sector). The meanings of strong and weak modal operators can then
be defined in terms of this set in the usual way. However, the ordering structures
can be more complicated; for example, there may be multiple, incompatible sets
of “best-ranked” worlds (worlds compared to which there are no higher-ranked
worlds), or infinite series of ever-better worlds, with no stopping point at a highest-
ranked world. In those cases, more sophisticated definitions will be called for; see
Kratzer’s papers or Portner (2009) for details.
Kratzer’s framework has been used to provide a better analysis of modal strength
than the one which comes from modal logic. Recall the problem of must and ought
illustrated in (26). In order to explain the fact that must is stronger, we want to
say that the set of accessible worlds for must is always a superset of the set of
accessible worlds for ought. Within Kratzer’s system, the set of accessible worlds for
a modal is determined by the conversational backgrounds which function as modal
base and ordering source. Suppose that it is part of the lexical meaning of deontic
must that it uses a particular conversational background which includes only the
most important priorities and rules in the context; in contrast, deontic should uses
both this conversational background and another one which includes some less-
important priorities and rules. From this intuitively reasonable difference between
must and should, we can derive that the sentence with must is stronger than the
corresponding sentence with should.
Let me show this by way of example. Consider the following conversational
backgrounds:

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

20 introduction

(31) a. For any situation s, cbcrucial (s) = the set of the most important priorities
and rules in s (for example: you wash hands, you cook dinner).
b. For any situation s, cbgood (s) = the set of relevant but lesser priorities in
s (for example: you clean up the house before dinner).

Now consider (32):

(32) a. You must cook dinner.

b. You must clean up the house beforehand.
c. You should clean up the house beforehand.

The ordering source for (32a–b) is cbcrucial , while that for (32c) is a combination
of cbgood and cbcrucial . Given this, (32a) and (32c) will be true, but (32b) will
be false because there are best-ranked worlds accessible according to cbcrucial in
which you do not clean up the house. This example illustrates how the notion of
conversational background allows us to state in an explicit and intuitive way how
various contextual factors combine to determine which worlds are accessible in
modal semantics.
Ordering semantics is useful for dealing with a number of other problems in the
semantics of modality, including further variation in modal strength (as in (18)),
comparative modality, and the interaction between modals and conditionals, but it
would take us too far from our main concerns to pursue them here. Our discussion
so far has been enough to highlight two points at which the ordering semantics
approach is helpful in the theory of mood. First, Figure 1.2 is an example of the kind
of partially ordered set of worlds (or posw) which will show up in analyses of both
verbal mood and sentence mood, and which may serve as the basis for developing
a general theory of core mood. And second, the discussion of must and ought
gives a sense of how we can formalize intuitively meaningful factors (like “the most
important priorities and rules”) to give an account of the lexical semantics of modal
expressions. In order to understand verbal mood, we will develop techniques for
explaining a variety of features of lexical meaning within a formal theory of modal
semantics.
Before moving on, it is important to note that ordering semantics as sketched
above is not the last word on the semantics of modality. At the current time, we
see both a great deal of work aiming to modify and extend the basic framework
and some important arguments that the entire framework should be replaced.
When it comes to the semantics of mood, however, ordering semantics remains
the touchstone (though not necessarily the precise form of ordering semantics
proposed by Kratzer). Given the tight link between verbal mood and modality
in human language, it is certain that the correct theory of modality will play an
important role in the correct theory of mood, and so it should be seen as a test of

 There are several different possibilities for how to combine them. We might combine the ordering

sources as λs[cbcrucial (s) ∪ cbgood (s)] or merge them in another way. Or we might assign ought both
cbcrucial and cbgood ordering sources, adjusting the definition of ≤s to allow for multiple ordering
sources. See von Fintel and Iatridou (), Rubinstein (), and Katz et al. () for discussion.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

the flow of information in discourse 21

old and new theories of modal semantics whether they can produce useful insights
when used for thinking about mood.

1.4 The flow of information in discourse

Linguistic research on sentence mood builds on a very diverse range of ideas
about discourse meaning from pragmatics, philosophy, and logic, and there is
no foundational framework with the same near-standard status that the possible
worlds semantics for modality has in research on verbal mood. In order to
investigate sentence mood, we need some ideas about the nature of “information
flow” in discourse. That is, we need some understanding of how information is
introduced, solicited, and deployed for planning, reasoning, and social interaction
in conversations or other genres like narratives and records. (Exactly what falls
under this term “flow” must be left vague at this point. Theories differ, and it is up
to a specific theory of discourse meaning to state what it encompasses and what it
does not, and to justify its choices.) In this section, I will give a brief overview of two
approaches to information flow which are important in contemporary research on
sentence mood: the dynamic approach and classical speech act theory.

1.4.1 The dynamic approach

The central virtue of the dynamic theories is that, through all their variations,
they give an explanation of the nature of information in discourse which is both
precise and amenable to a simple account of how that information changes. I will
introduce the ideas of dynamic semantics from a slightly unorthodox direction,
beginning with the work of Robert Stalnaker (1970, 1974, 1978, 1984), and then
discuss how Stalnaker’s ideas develop into dynamic semantics in the form of the
File Change Semantics of Irene Heim (1982, 1983, 1988, 1992). I will then give
overviews of two other important, and somewhat different, theories of dynamic
semantics: Kamp’s Discourse Representation Theory (Kamp 1981; Kamp and Reyle
1993) and the “Amsterdam school” of Dynamic Logic (Groenendijk and Stokhof
1990, 1991; Groenendijk et al. 1996). In this section, we will not go into more recent
work which explicitly applies the dynamic approach to understanding verbal mood
or sentence mood; we will delve into such material in Chapters 2 and 3.

Stalnaker on pragmatic presupposition and assertion. Stalnaker (1978) pro-

poses that “the central concept needed to characterize speech contexts” is speaker
presupposition:

A proposition is presupposed if the speaker is disposed to act as if he assumes or believes

that the proposition is true, and as if he assumes or believes that his audience assumes or
believes that it is true as well. (Stalnaker 1978, p.321)

Building on this concept of presupposition and the associated model of the

discourse context, we will be able to characterize the dynamic perspective on
information flow in discourse.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

22 introduction

Stalnaker takes propositions to be sets of possible worlds (as we have done

above), and so it is possible to conjoin all of the speaker’s presuppositions into a
single proposition which represents all the ways the world could be which match
all of those presuppositions. This proposition is the context set of the speaker.
More generally, any participant in a linguistic exchange has a context set (relative
to that exchange):

(33) The
 context set of A in speaking to B =
{p : A presupposes p in speaking to audience B}

In a given conversation, each participant has her own context set. These context
sets might be different, since one or more participants could be mistaken about
what her audience assumes or believes. If the participants in the conversation
have the same presuppositions in speaking to one another, we have a non-
defective context, and theories of dynamic semantics typically make the ideal-
izing assumption that our understanding of language and communication can
make progress by considering only non-defective contexts. While one might fear
that this idealization takes us too far from the nature of real conversation, in
that most contexts are probably defective, it is perhaps reasonable to assume that
most contexts are “close enough to non-defective” (as Stalnaker says) that their
deficiencies do not matter.
In a non-defective context, we can speak of the common ground of the
context and the context set of the context:

(34) a. The common ground of a non-defective context c:

cgc = the set of presuppositions of all of the participants in c in speaking
to one another. 
b. The context set of a non-defective context c: csc = cgc

The context set of a non-defective context is identical to the context set of any
participant in the conversation (in speaking to any other participant).
In thinking about Stalnaker’s notion of context, it is worth noticing that the
discussion has moved quickly from a definition of presupposition based on a
speaker’s disposition to act as if he assumes or believes some things (in speaking
to an audience) to an analysis of context which relies on the concept of “the
conversation.” Specifically, the definitions of defective and non-defective contexts
only make sense relative to a delineation of a set of participants all of whom count as
speakers towards and audiences of one another. Thus, Stalnaker’s approach seems
to make the assumption that the nexus of speakers and auditors which constitutes a
conversation is clear and unproblematic. Although Stalnaker’s approach to defining
the common ground of a context is sufficient for the purposes for which he and
other dynamic semanticists use it, a more precise definition of context will require

 He or she might even have different context sets for each other addressee in the conversation, for

example if he or she is talking to two people, trying to deceive one with the support of another.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

the flow of information in discourse 23

a better-developed notion of conversation and with it a clearer understanding of

what it means to be a speaker’s “audience” (cf. Bell 1984).
Given the ability to model the conversational context with the common ground
and context set, Stalnaker discusses the concepts of assertion and linguistic pre-
supposition. He writes that

. . . the essential effect of an assertion is to change presuppositions of the participants in the

conversation by adding the content of what is asserted to what is presupposed. This effect is
avoided only if the assertion is rejected. (Stalnaker 1974, p.323)

These ideas about assertion can be expressed in terms of a rule affecting the
common ground or context set:

(35) a. The essential effect of an assertion of S with respect to a non-defective

context c with common ground cgc is to change cgc to cgc ∪ {[[ S ]] c }.
• As a formula: cgc + S = cgc ∪ {[[ S ]] c }
This effect comes about unless the assertion is rejected.
b. The essential effect of an assertion of S with respect to a non-defective
context c with context set csc is to change csc to csc ∩ [[ S ]] c .
• As a formula: csc + S = csc ∩ [[ S ]] c
This effect comes about unless the assertion is rejected.

The formulaic versions above rely on interpreting the ‘+’ symbol appropriately, as
shorthand for the extended description given by Stalnaker.

Heim’s File Change Semantics. Irene Heim (1982, 1983, 1988) takes us from
Stalnaker’s concepts of assertion and presupposition to the core ideas of dynamic
semantics. In Heim’s work, we can see a number of important developments
which lead from the philosophical and somewhat abstract perspective of Stalnaker
to a framework more suited to research into linguistically interesting questions
about assertion and presupposition. The most obvious differences result from
Heim’s concern for a much wider range of presupposition phenomena. First of
all, whereas Stalnaker expressed skepticism that it is ever crucial to talk about the
presuppositions of linguistic forms (as opposed to the pragmatic presuppositions
of the speakers who use those forms), Heim develops a notion of the felicity
conditions of a presupposition trigger within dynamic semantics. For example,
Heim (1992) defines the felicity condition of too in such a way that the second
sentence in (36a) (uttered in that context and with the coindexing indicated) is
subject to the condition (36b):

(36) a. Mary1 is happy. SusanF is happy too1 .

b. Felicity condition of too, given context c indicated by (36a): csc + SusanF
is happy too1 is defined iff Mary is happy in every world in csc .

Intuitively, the second sentence of (36a) presupposes that the referent associated
with index 1, Mary, is both distinct from Susan and happy. Heim does not formulate

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

24 introduction

this presupposition directly in terms of the speaker’s disposition to act as if he

assumes or believes something, but rather expresses it as a condition on when
csc + S is defined. This way of formalizing the felicity condition triggered by too
ultimately leads back to Stalnaker’s pragmatic speaker’s presupposition, since on
the assumption that csc is non-defective, if the condition in (36b) fails, this is
because the speaker and addressee do not presuppose (in the sense of pragmatic
presupposition) that Mary is happy. However, crucially it is not identical to
Stalnaker’s notion of presupposition, because it is seen as a linguistic property
of the sentence which constrains speakers’ utterances based on their (speaker)
presuppositions.
In the context of (36a), the felicity condition of too will be satisfied, if the
assertion of the first sentence is not challenged, because the first sentence entails
that Mary is happy and the assertion adds this information to the context set. Heim
uses similar ideas about presupposition to attack the famous Projection Problem
for presupposition (see for example Karttunen 1973; Karttunen and Peters 1975;
Gazdar 1979; Soames 1982; Heim 1988, 1992).
Other crucial differences between Heim’s work and Stalnaker’s flow from Heim’s
concern for presupposition phenomena related to anaphora and definiteness.
Example (37b) is felicitous if preceded by (37a), intuitively because a soldier in (37a)
provides a referent for he in (37b):

(37) a. I saw a soldier in the park.

b. He waved to me.

To be more precise, within Heim’s theory the felicity of a pronoun depends on

both the preceding context and indexing of noun phrases in Logical Form. On
the reading indicated by the indexing (38a), the second sentence is felicitous.
On the reading indicated by the indexing in (38b), in contrast, it is not felicitous
unless the referent for he is determined by previous material or extralinguistic
context.

(38) a. I saw [a soldier]1 in the park. He1 waved to me.

b. I saw [a soldier]1 in the park. He2 waved to me.

Definite and indefinite noun phrases show complementary patterns: A definite is

felicitous if it is coindexed with an antecedent and it is entailed that the referent
has the property denoted by the definite’s common noun phrase. (It can also be
felicitous due to extralinguistic context.) An indefinite is felicitous unless it is
coindexed with an antecedent.

(39) a. I saw [a soldier]1 in the park. The soldier1 waved to me.

b. I saw [a soldier]1 in the park. ∗ A soldier1 waved to me.

 The description here leaves out many complexities which are well-studied in the literature on

anaphora and definiteness, but this oversimplified version is sufficient to allow us to understand Heim’s
contributions to dynamic semantics.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

the flow of information in discourse 25

In (37)–(39) we see, in each case, that the initial sentence I saw a soldier in the park
introduces a soldier into the context, and that soldier can serve as a referent for
a pronoun or definite; it cannot, however, serve as the referent for a subsequent
indefinite, and indeed each indefinite must introduce its own, new individual into
the context. The main effort of Heim’s work on definiteness is to take this intuitive
description and develop it into an explanatory formal theory.
Under the inspiration of Stalnaker’s ideas about presupposition and assertion,
one can see (37)–(39) as showing that the common ground and context set contain
information not just about what is presupposed to be the case, but also about which
individuals the participants presuppose they are able to talk about. For example,
assertion of the first sentence in (38a) results in a context set which represents a
speaker presupposition that the speaker saw a soldier in the park and a speaker
presupposition that subsequent assertions can affect the speaker presuppositions
about this same presupposed soldier. Thus, after the second sentence in (38a) is
asserted, the context set encodes a speaker presupposition that the speaker saw a
soldier in the park who waved at her. The beauty of analyzing anaphora in terms of
the context set is that this overall presupposition of (38a) can be recorded without
there needing to be any actual soldier in the park—in the end, the context set
encodes a speaker presupposition to the effect that the speaker is disposed to act
as if she assumes or believes that she saw a soldier in the park who waved at
her. The soldier being talked about is a mere discourse referent, in Karttunen’s
(1976) terminology. We can understand discourse referents to be the referential
component of speaker presupposition.
The technical side of Heim’s work focuses on giving a precise, compositional
version of the above ideas. She replaces Stalnaker’s notion of the context set of the
context csc , a set of possible worlds, with the satisfaction set of the context satc ,
a set of pairs w, g consisting of a world w and assignment of values to indices g.
Note that g is a partial function from indices to values (not every index need be
in its domain), but all of the assignments in a satisfaction set must have the same
domain. The satisfaction set allows her to represent contextual information both
about what is the case (via w) and about discourse referents (via g). Specifically, if
index i is assigned a value by the assignments in the satisfaction set, i corresponds
to a discourse referent in the context. For example, satc in (40) represents a context
in which Max is a soldier, Al is a baker, we’re not sure whether Max walked, the
first discourse referent is one of the two men Max or Al, and there are no other
discourse referents.

(40) satc = { w1 , g1 , w1 , g2 w2 , g1 , w2 , g2 }, where

i. w1 , g1 :
w1 : Max is a soldier, Al is a baker, Max walked.
g1 (1) = Max. For all i = 1, g1 (i) is not defined.
ii. w1 , g2 :
w1 : (as above)
g2 (1) = Al. For all i = 1, g2 (i) is not defined.

 Instead of talking about the context, Heim talks about the satisfaction set of the “file,” a metaphor-

ical picture of context as a file cabinet filled with shared information.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

26 introduction

iii. w2 , g1 :
w2 : Max is a soldier, Al is a baker, Max didn’t walk.
g1 : (as above)
iv. w2 , g2 :
w2 : (as above)
g2 : (as above)

Heim has to adjust Stalnaker’s analyses of pragmatic presupposition and assertion

to make sure that pronouns and definite and indefinite noun phrases affect dis-
course referents in the way observed in (37)–(39). Let’s work with a simpler example
than those, (41a):

(41) a. [A soldier]1 walked. He1 waved.

b. i. soldier(x1 ) ∧ walked(x1 )
ii. waved(x1 )

Building on ideas of Lewis (1975), Heim proposes that pronouns and definite and
indefinite noun phrases are realized as variables at logical form, and that, while
these variables become bound in some cases, in simple ones like this they do
not. The logical form of the sequence (41a) is a pair of open sentences (41bi–ii).
Although an open sentence does not have proper truth conditions, we can talk
about the conditions under which an assignment of individuals to the free variables
leads to its being true in a world. For example, a pair consisting of a possible
world w and assignment of individuals to indices g satisfies (41bi) if and only if the
individual g(1) assigned to index 1 is a soldier who walked in w. The satisfaction
set of a sentence S, Sat(S), is the set of world-assignment pairs which satisfy S.
Satisfaction sets allow a treatment of assertion parallel to Stalnaker’s:

(42) The essential effect of an assertion of S with respect to a non-defective

context c with satisfaction set satc is to change satc to satc ∩ Sat(S).
• As a formula: satc + S = satc ∩ Sat(S)
This effect comes about unless the assertion is rejected.

The assertion of (41bi) in a context c results in a satisfaction set in which every

w, g pair has g(1) a soldier who walked in w. For example, if it is asserted in the
context with satc as in (40), the result is { w1 , g1 }.
The satisfaction set also allows an attractive theory of the presuppositions of
definites and indefinites. In the first sentence of (41a), the index 1 is associated with
an indefinite. This means that index 1 is “new”; in other words, it is presupposed
not to be in the domain of the satisfaction set. When we get to the second sentence,
however, 1 is associated with a definite; at this point, 1 is presupposed to be in the
domain of the satisfaction set:

(43) Presuppositions of indefinites and definites:

a. If variable xi is associated with an indefinite NPi in S, then satc + S is only
defined if i is not in the domain of satc .

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

the flow of information in discourse 27

b. If variable xi is associated with a definite NPi in S, then satc + S is only

defined if i is in the domain of satc .

What’s important here for us is that (43) encodes lexically triggered presuppositions
which pertain to discourse referents, not propositional information.
Although it was not explicit in her earliest work, Heim comes to see the formula
in (42) not just as a description of what happens when someone asserts a sentence,
but rather as giving the very meaning of the sentence.

the meaning of a sentence is its context change potential (CCP). . . . A CCP is a function
from contexts to contexts. (Heim 1992, p.185)

In this quote, Heim goes beyond Stalnaker’s perspective on assertion, according to

which a sentence has semantic content (a proposition) which gets used to change
the context when it is asserted. Instead she adopts a more radical view according
to which the context change potential is itself the semantic content. This view is
the central hypothesis of dynamic semantics, and so we can see Heim’s work as
realizing a dynamic semantics analysis of speaker presupposition, assertion, and
felicity conditions.

Versions of the dynamic approach. For the purposes of discussing theories of

mood in Chapters 2 and 3, it will be useful to have terms for a range of broadly
“dynamic” perspectives on meaning. We will use the term dynamic pragmatics
for the Stalnakerian view that sentences have a static (i.e. non-dynamic) semantic
content such as a proposition, which is then used to define a context change
potential (for example, assertion). We will use dynamic semantics for the view
which says that the content of a sentence is precisely its context change potential.
The difference between dynamic pragmatics and dynamic semantics concerns the
relative priority between static content and context change potential. For dynamic
pragmatics, static content is basic and is dynamicized through operations native to
pragmatics; for dynamic semantics, context change is basic, and if static content is
needed for some analytical purpose, it has to be recovered from the context change
potential. We should note that hybrid views are possible: perhaps some sentences
have a static content while others do not, or perhaps sentences have two aspects of
meaning, one of which conforms to the dynamic pragmatics view and the other
of which requires dynamic semantics. As I will use it, the broad term dynamic

 To give the complete meaning of the sentence in this framework would be to define c+S generally,

not just the effect of S on the satisfaction set of c, satc + S.

 One might wonder whether the difference is more than terminological, since the Sat(S) plays a

role in () parallel to the content of S in Stalnaker’s (b), but be that as it may, Heim is clearly endorsing
a move to understanding sentence meaning in essentially context-change terms.
 Lauer () uses “dynamic pragmatics” for a rather different approach to discourse meaning.

Schlenker () employs the phrase to describe Stalnaker’s theory of presupposition and assertion,
and Stalnaker (to appear) uses the term in the way I do here. Stalnaker () further explores and
advocates for the dynamic pragmatics approach.
 The discussion of the performativity of modals by Portner () suggests this last possibility.

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

28 introduction

Table . Versions of the dynamic approach

The dynamic approach

Dynamic pragmatics Dynamic semantics Combined approaches

Static semantic content, and Semantic content Semantic content

CCP via pragmatics = CCP is varied

approach encompasses dynamic pragmatics, dynamic semantics, and all of these

other possibilities. These versions of the dynamic approach are summarized in
Table 1.2.

Discourse Representation Theory. Beginning at the same time as Heim’s work

on File Change Semantics but continuing for a longer time, Hans Kamp and col-
leagues developed an important representative of the dynamic approach, Discourse
Representation Theory (DRT; Kamp 1981; Kamp and Reyle 1993, among others).
DRT differs in an important way from File Change Semantics and the Stalnakerian
ideas on which the latter is based: According to DRT, the dynamic aspects of
meaning are computed with the essential mediation of a new level of representation
known as the discourse representation structure. A discourse representation
structure, or DRS, is a level of Logical Form which can incorporate the contribu-
tions of a sequence of sentences in discourse. For example, the operation of DRS-
construction applied to the sequence (41a) from above results in, first, (44b), and
then (44c):

(44) a. Initial DRS:

x
b. + A soldier walked. ⇒ soldier(x)
walked(x)

xy
soldier(x)
c. + He waved ⇒ walked(x)
waved(y)
x=y

The box structures in (44) are the conventional way of representing DRSs; officially,
they are just set-theoretic pairs consisting of a set of variables (discourse referents,
on top) and formulas in a logical language (conditions, at bottom). A DRS is

i i

i i
 i i
OUP CORRECTED PROOF – FINAL, 10/10/2017, SPi

i i

the flow of information in discourse 29

interpreted through techniques familiar from logic, and the simple examples in
(44) can be interpreted extensionally in a first-order model. Specifically, the DRS
in (44b) is true in a model M iff there is a soldier who walked in M, and the one in
(44c) is true in M if there is a soldier who walked and waved in M.
From the above sketch, it should be apparent that the semantic mechanisms
of DRT are not dynamic since each DRS is interpreted with a standard static
semantics. What is special about DRT is that the dynamism in the theory is realized
through the DRS-construction process. For example, the dynamic meaning of He
waved in (44c) is captured by the introduction into the DRS of discourse referent
y and conditions waved(y) and x = y. In this way it differs crucially from Heim’s
framework in which the context change potential of a sentence is a function from
contexts (represented in a simplified way as satisfaction sets) to contexts.
I am not aware of any research on the semantics and pragmatics of mood in
which the representationalism of DRT would differentiate it in a crucial way from
other dynamic approaches. In the very basic example used here, note the similarity
between the logical forms used by File Change Semantics, (41), and the DRSs
constructed by DRT, (44). The differences are merely (i) that the DRSs contain
a layer enumerating the discourse referents (x, y) and (ii) that they introduce a
separate discourse referent y for the pronoun which is equated with the previously
introduced one, x. The first difference is due to the fact that, in order to assign
traditional static semantic values to DRSs, all discourse referents must be implicitly
bound at some point, and the discourse referent layer identifies the scope at which
this occurs. (If box (44b) is nested inside a larger box, the value of x cannot be
accessed from the containing box.) The second difference corresponds to the fact
that Heim assumes that anaphoric relations are already resolved in syntax (and
represented by coindexing, as in (41a)), while Kamp assumes that they are resolved
in a “pragmatic” phase of DRS-construction which follows the introduction of
syntactically overt material.

Dynamic logic. Beginning with an important series of papers by Groenendijk

and Stokhof, dynamic logic developed the dynamic approach to meaning through a
series of formal innovations (e.g. Groenendijk and Stokhof 1990, 1991; Groenendijk
et al. 1996). The earliest work in dynamic logic explored the best ways to formally
analyze anaphora in a dynamic framework; the approach opposes the representa-
tional analysis of DRT and in this way is similar to File Change Semantics. However,
in contrast to the way in which Heim’s commitment to dynamic semantics became
clear only gradually, the founders of dynamic logic were interested in dynamic
semantics from the start. Based on these initial results focused on anaphora,
dynamic logic went on to explore many other important topics, including modality
and sentence mood. In this section, we will briefly review some important ideas
about anaphora and epistemic modality developed within dynamic logic. For a
more detailed introduction, see Portner (2009). We will return to dynamic logic
in Chapter 3 as we study its contribution to theories of sentence mood.
Dynamic logic is, as the name indicates, a logic, meaning that it is a syntax and
semantics for a syntactically simple artificial language. It is important to note that
the fundamental ideas of dynamic logic have been employed in several different

i i

i i
Another random document with
no related content on Scribd:
delay augured well, and Harry’s silence was a sign that the alarm
was subsiding.
Here, however, was a letter addressed to him in Harry’s bold hand.
His poor little wife sitting next the tea-things, eyed her husband as he
opened it, with breathless alarm; she saw him grow pale as he
glanced at it; he lowered it to the tablecloth, and bit his lip, his eye
still fixed on it.
As he did not turn over the leaf, she saw it could not be a long
one, and must all be comprised within one page.
“Ry, darling,” she asked, also very pale, in a timid voice, “it’s
nothing very bad. Oh, darling, what is it?”
He got up and walked to the window silently.
“What do you say, darling?” he asked, suddenly, after a little
pause.
She repeated her question.
“No, darling, nothing, but—but possibly we may have to leave this.
You can read it, darling.”
He laid the letter gently on the tablecloth beside her, and she
picked it up, and read—
 “My dear Charlie,
“The old soldier means business. I think you must go up to
London, but be sure to meet me to-morrow at Hatherton, say
the Commercial Hotel, at four o’clock, P.M.
“Your affectionate brother,
“Harry Fairfield.”

“Who does he mean by the old soldier?” asked Alice, very much
frightened, after a silence.
“One of those d——d people who are plaguing me,” said Charles,
who had returned to the window, and answered, still looking out.
“And what is his real name, darling?”
“I’m ashamed to say that Harry knows ten times as well as I all
about my affairs. I pay interest through his hands, and he watches
those people’s movements; he’s a rough diamond, but he has been
very kind, and you see his note—where is it? Oh, thanks. I must be
off in half an hour, to meet the coach at the ‘Pied Horse.’”
“Let me go up, darling, and help you to pack, I know where all your
things are,” said poor little Alice, who looked as if she was going to
faint.
“Thank you, darling, you are such a good little creature, and never
think of yourself—never, never—half enough.”
His hands were on her shoulders, and he was looking in her face,
with sad strange eyes, as he said this, slowly, like a man spelling out
an inscription.
“I wish—I wish a thousand things. God knows how heavy my heart
is. If you cared for yourself, Alice, like other women, or that I weren’t
a fool—but—but you, poor little thing, it was such a venture, such a
sea, such a crazy boat to sail in.”
“I would not give up my Ry, my darling, my husband, my
handsome, clever, noble Ry—I’d lose a thousand lives if I had them,
one by one, for you, Charlie; and oh, if you left me, I should die.”
“Poor little thing,” he said, drawing her to him with a trembling
strain, and in his eyes, unseen by her, tears were standing.
“If you leave this, won’t you take me, Charlie? won’t you let me go
wherever you go? and oh, if they take my man—I’m to go with you,
Charlie, promise that, and oh, my darling, you’re not sorry you
married your poor little Alley.”
“Come, darling, come up; you shall hear from me in a day or two,
or see me. This will blow over, as so many other troubles have
done,” he said, kissing her fondly.
And now began the short fuss and confusion of a packing on brief
notice, while Tom harnessed the horse, and put him to the dogcart.
And the moment having arrived, down came Charles Fairfield, and
Tom swung his portmanteau into its place, and poor little Alice was
there with, as Old Dulcibella said, “her poor little face all cried,” to
have a last look, and a last word, her tiny feet on the big unequal
paving stones, and her eyes following Charlie’s face, as he stepped
up and arranged his rug and coat on the seat, and then jumped
down for the last hug; and the wild, close, hurried whisperings, last
words of love and cheer from laden hearts, and pale smiles, and the
last, really the last look, and the dog-cart and Tom, and the
portmanteau and Charlie, and the sun’s blessed light, disappear
together through the old gateway under the wide stone arch, with
tufted ivy and careless sparrows, and little Alice stands alone on the
pavement for a moment, and runs out to have one last wild look at
the disappearing “trap,” under the old trees, as it rattled swiftly down
to the narrow road of Carwell Valley.
It vanished—it was gone—the tinkling of the wheels was heard no
more. The parting, for the present, was quite over, and poor little
Alice turned at last, and threw her arms about the neck of kind old
Dulcibella, who had held her when a baby in her arms in the little
room at Wyvern Vicarage, and saw her now a young wife, “wooed
and married, and a’,” in the beauty and the sorrows of life; and the
light air of autumn rustled in the foliage above her, and a withered
leaf or two fell from the sunlit summits to the shadow at her feet; and
the old woman’s kind eyes filled with tears, and she whispered
homely comfort, and told her she would have him back again in a
day or two, and not to take on so; and with her gentle hand, as she
embraced her, patted her on the shoulder, as she used in other
years—that seemed like yesterday—to comfort her in nursery
troubles. But our sorrows outgrow their simple consolations, and turn
us in their gigantic maturity to the sympathy and wisdom that is
sublime and eternal.
Days passed away, and a precious note from Charlie came. It told
her where to write to him in London, and very little more.
The hasty scrawl added, indeed emphatically, that she was to tell
his address to no one. So she shut it up in the drawer of the old-
fashioned dressing-table, the key of which she always kept with her.
Other days passed. The hour was dull at Carwell Grange for Alice.
But things moved on in their dull routine without event or alarm.
Old Mildred Tarnley was sour and hard as of old, and up to a
certain time neither darker nor brighter than customary. Upon a day,
however, there came a shadow and a fear upon her.
Two or three times on that day and the next, was Mrs. Tarnley
gliding, when old Dulcibella with her mistress was in the garden,
about Alice’s bedroom, noiselessly as a shadow. The little girl
downstairs did not know where she was. It was known but to herself
—and what she was about. Coming down those dark stairs, and
going up, she went on tiptoe, and looked black and stern as if she
was “laying out” a corpse upstairs.
Accidentally old Dulcibella, coming into the room on a message
from the garden, surprised lean, straight Mrs. Tarnley, feloniously
trying to turn a key, from a bunch in her hand, in the lock of the
dressing-table drawer.
“Oh, la! Mrs. Tarnley,” cried old Dulcibella, very much startled.
The two women stood perfectly still, staring at one another. Each
looked scared. Stiff Mildred Tarnley, without, I think, being the least
aware of it, dropped a stiff short courtesy, and for some seconds
more the silence continued.
“What be you a-doing here, Mrs. Tarnley?” at length demanded
Dulcibella Crane.
“No occasion to tell you,” replied Mildred, intrepidly. “Another one,
that owed her as little as I’m like ever to do, would tell your young
mistress. But I don’t want to break her heart—what for should I?
There’s dark stories enough about the Grange without no one
hangin’ theirself in their garters. What I want is where to direct a
letter to Master Charles—that’s all.”
“I can’t say, I’m sure,” said old Dulcibella.
 “She got a letter from him o’ Thursday last; ’twill be in it no doubt,
and that I take it, ma’am, is in this drawer, for she used not to lock it;
and I expect you, if ye love your young mistress, to help me to get at
it,” said Mrs. Tarnley, firmly.
“Lor, Mrs. Tarnley, ma’am! me to pick a lock, ma’am! I’d die first.
Ye can’t mean it?”
“I knowd ye was a fool. I shouldn’t ’a said nothing to ye about it,”
said Mildred, with sharp disdain.
“Lawk! I never was so frightened in my life!” responded Dulcibella.
“Ye’ll be more so, mayhap. I wash my hands o’ ye,” said Mrs.
Tarnley, with a furious look, and a sharp little stamp on the floor. “I
thought o’ nothing but your mistress’s good, and if ye tell her I was
here, I’ll explain all, for I won’t lie under no surmises, and I think ’twill
be the death of her.”
“Oh, this place, this hawful place! I never was so frightened in my
days,” said Dulcibella, looking very white.
“She’s in the garden now, I do suppose,” said Mildred, “and if ye
mean to tell her what I was about, ’taint a pin’s head to me, but I’ll go
out and tell her myself, and even if she lives through it, she’ll never
hold up her head more, and that’s all you’ll hear from Mildred
Tarnley.”
“Oh, dear! dear! dear! my heart, how it goes!”
“Come, come, woman, you’re nothin’ so squeamish, I dare say.”
“Well,” said Dulcibella; “it may be all as you say, ma’am, and I’ll
say ye this justice, I ha’n’t missed to the value of a pennypiece since
we come here, but if ye promise me, only ye won’t come up here no
more while we’re out, Mrs. Tarnley, I won’t say nothing about it.”
“That settles it, keep your word, Mrs. Crane, and I’ll keep mine; I’ll
burn my fingers no more in other people’s messes;” and she shook
the key with a considerable gingle of the whole bunch from the
keyhole, and popped it grimly into her pocket.
“Your sarvant, Mrs. Crane.”
“Yours, Mrs. Tarnley, ma’am,” replied Dulcibella.
And the interview which had commenced so brusquely, ended with
ceremony, as Mildred Tarnley withdrew.
That old woman was in a sort of fever that afternoon and the next
day, and her temper, Lilly Dogger thought, grew more and more
savage as night approached. She had in her pocket a friendly
fulsome little letter, which had reached her through the post,
announcing an arrival for the night that was now approaching. The
coach that changed horses at the “Pied Horse,” was due there at
half-past eleven, P.M., but might not be there till twelve, and then
there was a long drive to Carwell Grange.
“I’m wore out wi’ them, I’m tired to death; I’m wore off my feet wi’
them; I’m worked like a hoss. ’Twould be well for Mildred Tarnley, I’m
thinkin’, she was under the mould wi’ a stone at her head, and shut
o’ them all.”
 CHAPTER XXV.
LILLY DOGGER IS SENT TO BED.

That night the broad-shouldered child, Lilly Dogger, was up later

than usual. An arrear of pots and saucepans to scour, along with
customary knives and forks to clean, detained her.
“Bustle, you huzzy, will ye?” cried the harsh voice of old Mildred,
who was adjusting the kettle on the kitchen fire, while in the scullery
the brown-eyed little girl worked away at the knife-board. A mutton-
fat, fixed in a tin sconce on the wall, so as to command both the
kitchen and the scullery, economically lighted each, the old woman
and her drudge, at her work.
“Yes’m, please,” she said interrogatively, for the noise of her task
prevented her hearing distinctly.
“Be alive, I say. It’s gone eleven, you slut; ye should a bin in your
bed an hour,” screeched Mildred, and then relapsed into her
customary grumble.
“Yes, Mrs. Tarnley, please’m,” answered the little girl, resuming
with improved energy.
Drowsy enough was the girl. If there had been a minute’s respite
from her task, I think she would have nodded.
“Be them things rubbed up or no, or do you mean to ’a done to-
night, huzzy?” cried Mrs. Tarnley, this time so near as to startle her,
for she had unawares put her wrinkled head into the scullery. “Stop
that for to-night, I say. Leave ’em lay, ye’ll finish in the morning.”
“Shall I take down the fire, Mrs. Tarnley, ma’am, please?” asked
Lilly Dogger, after a little pause.
“No, ye sha’n’t. What’s that ye see on the fire; have ye eyes in
your head? Don’t you see the kettle there? How do I know but your
master’ll be home to-night, and want a cup o’ tea, or—law knows
what?”
Mrs. Tarnley looked put about, as she phrased it, and in one of
those special tempers which accompanied that state. So Lilly
Dogger, eyeing her with wide open eyes, made her a frightened little
courtesy.
“Why don’t ye get up betimes in the morning, huzzy, and then ye
needn’t be mopin’ about half the night? All the colour’s washed out o’
your big, ugly, platter face, wi’ your laziness—as white as a turnip.
When I was a girl, if I left my work over so, I’d ’a the broomstick
across my back, I promise ye, and bread and water next day too
good for my victuals; but now ye thinks ye can do as ye like, and all’s
changed! An’ every upstart brat is as good as her betters. But don’t
ye think ye’ll come it over me, lass, don’t ye. Look up there at the
clock, will ye, or do ye want me to pull ye up by the ear—ten minutes
past eleven—wi’ your dawdling, ye limb!”
The old woman whisked about, and putting her hand on a
cupboard door, she turned round again before opening it, and said—
“Come on, will ye, and take your bread if you want it, and don’t ye
stand gaping there, ye slut, as if I had nothing to do but attend upon
you, with your impittence. I shouldn’t give ye that.”
She thumped a great lump of bread down on the kitchen table by
which the girl was now standing.
“Not a bit, if I did right, and ye’ll not be sittin’ up to eat that, mind
ye; ye’ll take it wi’ ye to yer bed, young lady, and tumble in without
delay, d’ye mind? For if I find ye out o’ bed when I go in to see all’s
right, I’ll just gi’e ye that bowl o’ cold water over yer head. In wi’ ye,
an’ get ye twixt the blankets before two minutes—get along.”
The girl knew that Mrs. Tarnley could strike as well as “jaw,” and
seldom threatened in vain, so with eyes still fixed upon her, she took
up her fragment of loaf, with a hasty courtesy, of which the old
woman took no notice, and vanished frightened through a door that
opened off the kitchen.
The old woman holding the candle over her head, soon peeped in
as she had threatened.
Lilly Dogger lay close affecting to be asleep, though that feat in the
time was impossible, and was afraid that the thump, thump of her
heart, for she greatly feared Mrs. Tarnley, might be audible to that
severe listener.
Out she went, however, without anything more, to the great relief
of the girl.
 Lilly Dogger lay awake, for fear is vigilant, and Mrs. Tarnley’s
temper she knew was capricious as well as violent.
Through the door she heard the incessant croak of the old
woman’s voice, as she grumbled and scolded in soliloquy, poking
here and there about the kitchen. The girl lay awake, listening
vaguely in the dark, and watching the one bright spot on the
whitewashed wall at the foot of her bed, which Mrs. Tarnley’s candle
in the kitchen transmitted through the keyhole. It flitted and glided,
now hither, now thither, now up, now down, like a white butterfly in a
garden, silently indicating the movements of the old woman, and
illustrating the clatter of her clumsy old shoes.
In a little while the door opened again, and the old woman entered,
having left her candle on the dresser outside.
Mrs. Tarnley listened for a while, and you may be sure Lilly Dogger
lay still. Then the old woman, in a hard whisper, asked, “Are you
awake?” and listened.
“Are ye awake, lass?” she repeated, and receiving no answer, she
came close to the bed, by way of tucking in the coverlet, in reality to
listen.
So she stood in silence by the bed for a minute, and then very
quickly withdrew and closed the door.
Then Lilly Dogger heard her make some arrangements in the
kitchen, and move, as she rightly concluded, a table which she
placed against her door.
Then the white butterfly, having made a sudden sweep round the
side wall, hovered no longer on Lilly Dogger’s darkened walls, and
old Mildred Tarnley and her candle glided out of the kitchen.
The girl had grown curious, and she got up and peeped, and found
that a clumsy little kitchen table had been placed against her door,
which opened outward.
Through the keyhole she also saw that Mildred had not taken
down the fire. On the contrary, she had trimmed and poked it, and a
kettle was simmering on the bar.
She did not believe that Mrs. Tarnley expected the arrival of her
master, for she had said early in the day that she thought he would
come next evening. Lilly Dogger was persuaded that Mrs. Tarnley
was on the look out for some one else, and guarding that fact with a
very jealous secrecy.
She went again to her bed; wondering she listened for the sounds
of her return, and looked for the little patch of light on the
whitewashed wall; but that fluttering evidence of Mrs. Tarnley’s
candle did not reappear before the tired little girl fell asleep.
She was wakened in a little time by Mrs. Tarnley’s somewhat noisy
return. She was grumbling bitterly to herself, poking the fire, and
pitching the fire-irons and other hardware about with angry
recklessness.
The girl turned over, and notwithstanding all Mildred’s noisy
soliloquy was soon asleep again.
Again she awoke—I suppose recalled to consciousness by some
noise in the kitchen. The little white light was in full play on the wall
at the foot of her bed, and Mrs. Tarnley was talking fluently in an
undertone. Then came a silence, during which the old Dutch clock
struck one.
Lilly Dogger’s eyes were wide open now, and her ears erect. She
heard no one answer the old woman, who resumed her talk in a
minute; and now she seemed careful to make no avoidable noise—
speaking low, and when she moved about the kitchen treading softly,
and moving anything she had to stir gently. Altogether she was now
taking as much care not to disturb as she had shown carelessness
upon the subject before.
Lilly Dogger again slipped out of bed, and peeped through the
keyhole. But she could not see Mrs. Tarnley nor her companion, if
she had one.
Old Mildred was talking on, not in her grumbling interrupted
soliloquy, but in the equable style of one spinning a long narrative.
This hum was relieved now and then by the gentle click of a teacup,
or the jingle of a spoon.
If Mrs. Tarnley was drinking her tea alone at this hour of night and
talking so to herself, she was doing that she had never done before,
thought the curious little girl; and she must be a-going mad. From
this latter apprehension, however, she was relieved by hearing some
one cough. It was not Mrs. Tarnley, who suspended her story,
however. But there was an unmistakable difference of tone in this
cough, and old Mildred said more distinctly something about a cure
for a cough which she recommended.
Then came an answer in an odd drawling voice. The words she
could not hear, but there could no longer be any doubt as to the
presence of a stranger in the kitchen.
Lilly Dogger was rather frightened, she did not quite know why,
and listened without power to form a conjecture. It was plain that the
person who enjoyed old Mildred’s hospitality was not her master, nor
her mistress, nor old Dulcibella Crane.
As she listened, and wondered, and speculated, sleep overtook
her once more, and she quite forgot the dialogue, and the kitchen,
and Mildred Tarnley’s tea, and went off upon her own adventures in
the wild land of dreams.
 CHAPTER XXVI.
THE LADY HAS HER TEA.

“You suffers dreadful, ma’am,” said Mildred Tarnley. “Do you have
them toothaches still?”
“’Twas not toothache—a worse thing,” said the stranger, demurely,
who, with closed eyes, and her hand propping her head, seemed to
have composed herself for a doze in the great chair.
“Wuss than toothache! That’s bad. Earache, mayhap?” inquired
Mrs. Tarnley, with pathetic concern, though I don’t think it would have
troubled her much if her guest had tumbled over the precipice of
Carwell Valley and broken her neck among the stones in the brook.
“Pain in my face—it is called tic,” said the lady, with closed eyes, in
a languid drawl.
“Tic? lawk! Well, I never heard o’ the like, unless it be the field-bug
as sticks in the cattle—that’s a bad ailment, I do suppose,”
conjectured Mrs. Tarnley.
“You may have it yourself some day,” said this lady, who spoke
quietly and deliberately, but with fluency, although her accent was
foreign. “When we are growing a little old our bones and nerves they
will not be young still. You have your rheumatism, I have my tic—the
pain in my cheek and mouth—a great deal worse, as you will find,
whenever you taste of it, as it may happen. Your tea is good—after a
journey tea is so refreshing. I cannot live without my cup of tea,
though it is not good for my tic. So, ha, ha, he-ha! There is the tea
already in my cheek—oh! Well, you will be so good to give me my
bag.”
Mildred looked about, and found a small baize bag with an
umbrella and a bandbox.
“There’s a green bag I have here, ma’am.”
“A baize bag?”
“Yes, ’m.”
“Give it to me. Ha, yes, my bibe—my bibe—and my box.”
 So this lady rummaged and extricated a pipe very like a
meerschaum, and a small square box.
“Tibbacca!” exclaimed Mrs. Tarnley. The stranger interpreted the
exclamation, without interrupting her preparations.
“Dobacco? no, better thing—some opium. You are afraid Mrs.
Harry Fairfield, she would smell id. No—I did not wish to disturb her
sleeb. I am quite private here, and do not wish to discover myself.
Ya, ya, ya, hoo!”
It was another twinge.
“Sad thing, ma’am,” said Mildred. “Better now, perhaps?”
“Put a stool under my feed. Zere, zere, sat will do. Now you light
that match and hold to the end of ze bibe, and I will zen be bedder.”
Accordingly Mildred Tarnley, strongly tempted to mutter a criticism,
but possibly secretly in awe of the tall and “big-made” woman who
issued these orders, proceeded to obey them.
“No great odds of a smell arter all,” said Mrs. Tarnley, approvingly,
after a little pause.
“And how long since Harry married?” inquired the smoker after
another silence.
“I can’t know that nohow; but ’tis since Master Charles gave ’em
the lend o’ the house.”
“Deeb people these Vairvields are,” laughed the big woman,
drowsily.
“When will he come here?”
“To-morrow or next day, I wouldn’t wonder; but he never stays
long, and he comes and goes as secret-like as a man about a
murder a’most.”
“Ha, I dare say. Old Vairvield would cut him over the big shoulders
with his horsewhip, I think. And when will your master come?”
“Master comes very seldom. Oh! ve-ry. Just when he thinks to find
Master Henry here, maybe once in a season.”
“And where does he live—at home or where?” asked the tall
visitor.
“Well, I can’t say, I’m sure, if it baint at Wyvern. At Wyvern, I do
suppose, mostly. But I daresay he travels a bit now and again. I don’t
know, I’m sure.”
 “Because I wrote to him to Wyvern to meet me here. Is he at
Wyvern?”
“Well, faith, I can’t tell. I know no more than you, ma’am, where
Master Charles is,” said Mildred, with energy, relieved in the midst of
her rosary of lies to find herself free to utter one undoubted truth.
“You have been a long time in the family, Mrs. Tarnley?” drawled
the visitor, listlessly.
“Since I was the height o’ that—before I can remember. I was born
in Carwell gate-house here. My mother was here in old Squire’s
time, meanin’ the father o’ the present Harry Fairfield o’ Wyvern that
is, and grandfather o’ the two young gentlemen, Master Charles and
Master Harry. Why, bless you, my grandfather, that is my mother’s
father, was in charge o’ the house and farm, and the woods, and the
tenants, and all; there wasn’t a tree felled, nor a cow sold, nor an
acre o’ ground took up, but jest as he said. They called him honest
Tom Pennecuick; he was thought a great deal of, my grandfather
was, and Carwell never turned in as good a penny to the Fairfields
as in his time; not since, and not before—never, and never will, that’s
sure.”
“And which do you like best, Squire Charles or Squire Harry?”
inquired the languid lady.
“I likes Charles,” said Mrs. Tarnley, with decision.
“And why so?”
“Well, Harry’s a screw; ye see he’d as lief gie a joint o’ his thumb
as a sixpence. He’ll take his turn out of every one good-humoured
enough, and pay for trouble wi’ a joke and a laugh; a very pleasant
gentleman for such as has nothing to do but exchange work for his
banter and live without wages; all very fine. I never seed a shillin’ of
hisn since he had one to spend.”
“Mr. Charles can be close-fisted too, when he likes it?” suggested
the old lady.
“No, no, no, he’s not that sort if he had it. Open-handed enough,
and more the gentleman every way than Master Harry—more the
gentleman,” answered Mildred.
“Yes, Harry Fairfield is a shrewd, hard man, I believe; he ought to
have helped his brother a bit; he has saved a nice bit o’ money, I
dare say,” said the visitor.
 “If he hasn’t a good handful in his kist corner, ’t’aint that he wastes
what he gets.”
“I do suppose he’ll pay his brother a fair rent for the house?” said
the visitor.
“Master Harry’ll pay for no more than he can help,” observed
Mildred.
“It’s a comfortable house,” pursued the stranger; “’twas so when I
was here.”
“Warm and roomy,” acquiesced Mrs. Tarnley—“chimbley, roof, and
wall—staunch and stout; ’twill stand a hundred year to come, wi’ a
new shingle and a daub o’ mortar now and again. There’s a few
jackdaws up in the chimbleys that ought to be drew out o’ that wi’
their sticks and dirt,” she reflected, respectfully.
“And do you mean to tell me he pays no rent for the Grange, and
keeps his wife here?” demanded the lady, peremptorily.
“I know nothing about their dealings,” answered Mrs. Tarnley, as
tartly.
“And ’t’aint clear to me I should care much neither; they’ll settle
that, like other matters, without stoppin’ to ask Mildred what she
thinks o’t; and I dare say Master Harry will be glad enough to take it
for nothing, if Master Charles will be fool enough to let him.”
“Well, he sha’n’t do that, I’ll take care,” said the lady, maintaining
her immovable pose, which, with a certain peculiarity in the tone of
her voice, gave to her an indescribable and unpleasant languor.
“I never have two pounds to lay on top o’ one another. Jarity
begins at home. I’ll not starve for Master Harry,” and she laughed
softly and unpleasantly.
“His wife, you say, is a starved gurate’s daughter!”
“Parson Maybell—poor he was, down at Wyvern Vicarage—meat
only twice or thrice a week, as I have heard say, and treated old
Squire Harry bad, I hear, about his rent; and old Squire Fairfield was
kind—to her anyhow, and took her up to the hall, and so when she
grew up she took her opportunity and married Master Harry.”
“She was clever to catch such a shrewd chap—clever. Light again;
I shall have three four other puff before I go to my bed—very clever.
How did she take so well, and hold so fast, that wise fellow, Harry
Fairfield?”
 “Hoo! fancy, I do suppose, and liken’. She’s a pretty lass. All them
Fairfields married for beauty mostly. Some o’ them got land and
money, and the like, but a pretty face allays along with the fortune.”
The blind stranger, for blind she was, smiled downward, faintly and
slily, while she was again preparing the pipe.
“When will Harry come again?” she asked.
“I never knows, he’s so wary; do you want to talk to him, ma’am?”
said Mildred.
“Yes, I do,” said she; “hold the match now, Mrs. Tarnley, please.”
So she did, and—puff, puff, puff—about a dozen times, went the
smoke, and the smoker was satisfied.
“Well, I never knows the minute, but it mightn’t be for a fortnight,”
said Mrs. Tarnley.
“And when Mr. Charles Fairfield come?” asked the visitor.
“If he’s got your letter he’ll be here quick enough. If it’s missed him
he mayn’t set foot in it for three months’ time. That’s how it is wi’
him,” answered Mildred.
“What news of old Harry at Wyvern?” asked the stranger.
“No news in partic’lar,” answered Mildred, “only he’s well and
hearty—but that’s no news; the Fairfields is a long-lived stock, as
every one knows; he’ll not lie in oak and wool for many a day yet, I’m
thinkin’.”
Perhaps she had rightly guessed the object of the lady’s solicitude,
for a silence followed.
“There’s a saying in my country—‘God’s children die young,’” said
the tall lady.
“And here about they do say, the Devil takes care of his own,” said
Mildred Tarnley. “But see how my score o’ years be runnin’ up; I take
it sinners’ lives be lengthened out a bit by the Judge of all, to gi’e us
time to stay our thoughts a little, and repent our misdeeds, while yet
we may.”
“You have made a little fire in my room, Mrs. Tarnley?” inquired the
stranger, who had probably no liking for theology.
“Yes ’m; everything snug.”
“Would you mind running up and looking? I detest a chill,” said this
selfish person.
 At that hour no doubt Mrs. Tarnley resented this tax on her
rheumatics; but though she was not a woman to curb her
resentments, she made shift on this occasion; that did not prevent
her, however, from giving the stranger a furious look, while she
muttered inaudibly a few words.
“I’ll go with pleasure, ma’am; but I’m sure it’s all right,” she said
aloud, very civilly, and paused, thinking perhaps that the lady would
let her off the long walk upstairs to the front of the house.
“Very good; I’ll wait here,” said the guest, unfeelingly.
“As you please ’m,” said Mildred, and, with a parting look round
the kitchen, she took the candle, and left the lady to the light of the
fire.
The lady was almost reclining in her chair, as if she were dozing;
but in a few moments up she stood, and placing her hand by her ear,
listened; then, with her hands advanced, she crept slowly, and as
noiselessly as a cat, across the floor. She jostled a little against the
table at Lilly Dogger’s door; then she stopped perfectly still, withdrew
the table without a sound; the door swung a little open, and the
gaunt figure in grey stood at it, listening. A very faint flicker from the
fire lighted this dim woman, who seemed for the moment to have no
more life in her than the tall, gray stone of the Druid’s hoe on
Cressley Common.
Lilly Dogger was fast asleep; but broken were her slumbers
destined to be that night. She felt a hand on her neck, and, looking
up, could not for a while see anything, so dark was the room.
She jumped up in a sitting posture, with a short cry of fear, thinking
that she was in the hands of a robber.
“Be quiet, fool,” said the tall woman, slipping her hand over the
girl’s mouth. “I’m a lady, a friend of Mrs. Mildred Tarnley, and I’m
come to stay in the house. Who is the lady that sleeps upstairs in the
room that used to be Mr. Harry’s? You must answer true, or I’ll pull
your ear very hard.”
“It is the mistress, please ’m,” answered the frightened girl.
“Married lady?”
“Yes ’m.”
“Who is her husband?”
 With this question the big fingers of her visitor closed upon Lilly
Dogger’s ear with a monitory pinch.
“The master, ma’am.”
“And what’s the master’s name, you dirdy liddle brevarigator?”
And with these words her ear was wrung sharply.
She would have cried, very likely, if she had been less frightened,
but she only winced, with her shoulders up to her ears, and
answered in tremulous haste—
“Mr. Fairfield, sure.”
“There’s three Mr. Vairvields: there’s old Mr. Vairvield, there’s Mr.
Charles Vairvield, and there’s Mr. Harry Vairvield—you shall speak
plain.”
And at each name in her catalogue she twisted the child’s ear with
a sharp separate wring.
“Oh, law, ma’am. Please ’m, I mean Mr. Charles Fairfield. I didn’t
mean to tell you no story, indeed, my lady.”
“Ho, ho—yes—Charles, Charles—very goot. Now, you tell me how
you know Mr. Harry from Mr. Charles?”
“Oh, law, ma’am! oh, law! oh, ma’am, dear! sure, you won’t pull it
no more, good lady, please—my ear’s most broke,” gasped the girl,
who felt the torture beginning again.
“You tell truth. How do you know Mr. Charles from Mr. Harry?”
“Mr. Charles has bigger eyes, ma’am, and Mr. Harry has lighter
hair, and a red face, please ’m, and Mr. Charles’s face is brown, and
he talks very quiet-like, and Mr. Harry talks very loud, and he’s
always travellin’ about a-horseback, and Mr. Charles is the eldest
son, and the little child they’re lookin’ for is to be the Squire o’
Wyvern.”
The interrogator here gave her a hard pinch by the ear, perhaps
without thinking of it, for she said nothing for a minute nearly, and the
girl remained with her head buried between her shoulders, and her
eyes wide open, staring straight up where she conjectured her
examiner’s face might be.
“Is the man that talks loud—Mr. Harry—here often?” asked the
voice at her bedside.
“But seldom, ma’am—too busy at fairs and races, I hear them
say.”
 “And Mr. Charles—is he often here?”
“Yes ’m; master be always here, exceptin’ this time only; he’s gone
about a week.”
“About a week, Mr. Charles?”
“Oh, la, ma’am—yes, indeed, ma’am, dear, it’s just a week to-day
since master went.”
Here was a silence.
“That will do. If I find you’ve been telling me lies I’ll take ye by the
back of the neck and squeeze your face against the kitchen bars till
it’s burnt through and through—do you see; and I give you this one
chance, if you have been telling lies to say so, and I’ll forgive you.”
“Nothing but truth, indeed and indeed, ma’am.”
“Old Tarnley will beat you if she hears you have told me anything.
So keep your own secret, and I’ll not tell of you.”
She saw the brawny outline of the woman faintly like a black
shadow as she made her way through the door into the kitchen, and
she heard the door close, and the table shoved cautiously back into
its place, and then, with a beating heart, she lay still and awfully wide
awake in the dark.
 CHAPTER XXVII.
THROUGH THE HOUSE.

This stalwart lady stumbled and groped her way back to her chair,
and sat down again in the kitchen. The chair in which she sat was an
old-fashioned arm-chair of plain wood, uncoloured and clumsy.
When Mildred Tarnley returned, the changed appearance of her
guest struck her.
“Be ye sick, ma’am?” she asked, standing, candle in hand, by the
chair.
The visitor was sitting bolt upright, with a large hand clutched on
each arm of the chair, with a face deadly pale and distorted by a
frown or a spasm that frightened old Mildred, who fancied, as she
made no sign, not the slightest stir, that she was in a fit, or possibly
dead.
“For God’s sake, ma’am,” conjured old Mildred, fiercely, “will ye
speak?”
The lady in the chair started, shrugged, and gasped. It was like
shaking off a fit.
“Ho! oh, Mildred Tarnley, I was thinking—I was thinking—did you
speak?”
Mildred looked at her, not knowing what to make of it. Too much
laudanum—was it? or that nervous pain in her head.
“I only asked you how you were, ma’am—you looked so bad. I
thought you was just going to work in a fit.”
“What an old fool! I never was better in my life—fit! I never had a fit
—not I.”
“You used to have ’em sometimes, long ago, ma’am, and they
came in the snap of a finger, like,” said Mildred, sturdily.
“Clear your head of those fits, for they have left me long ago. I’m
well, I tell you—never was better. You’re old—you’re old, woman,
and that which has made you so pious is also making you blind.”
“Well, you look a deal better now—you do,” said Mildred, who did
not want to have a corpse or an epileptic suddenly on her hands,